Hartmuth Arenhovel

arenhoev@kph.uni-mainz.de
Hartmut Backe
backe@kph.uni-mainz.de
Dieter Drechsel
drechsel@kph.uni-mainz.de
Jorg Friedrich
friedrch@kph.uni-mainz.de
Karl-Heinz Kaiser
kaiser@kph.uni-mainz.de
Thomas Walcher
walcher@kph.uni-mainz.de

University of Mainz
Institute for Nuclear Physics
Johann-Joachim-Becher-Weg 45
55128 Mainz, Germany

The articles in this book originally appeared on the internet (www.eurphysj.org) as open access
publication of the journal
The European Physical Journal A Hadrons and Nuclei
Volume 28, Supplement 1
ISSN 1434-601X
c SIF and Springer-Verlag Berlin Heidelberg 2006

ISBN-10 3-540-36753-5 Springer Berlin Heidelberg New York

ISBN-13 978-3-540-36753-6 Springer Berlin Heidelberg New York
Library of Congress Control Number: 2006929544

This work is subject to copyright. All rights reserved, whether the whole or part of the material is concerned,
specically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction
on microlm or in any other way, and storage in data banks. Duplication of this publication or parts thereof
is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current
version, and permission for use must always be obtained from SIF and Springer. Violations are liable for
prosecution under the German Copyright Law.

Springer is a part of Springer Science+Business Media

springer.com


c SIF and Springer-Verlag Berlin Heidelberg 2006
Printed in Italy

The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply,
even in the absence of a specic statement, that such names are exempt from the relevant protective laws
and regulations and therefore free for general use.

Typesetting and Cover design: SIF Production Oce, Bologna, Italy

Printing and Binding: Tipograa Compositori, Bologna, Italy
Printed on acid-free paper SPIN: 11804239 5 4 3 2 1 0
Sponsors

Deutsche Forschungsgemeinschaft, Bonn

Institut fur Kernphysik, Universitat Mainz

The European Physical Journal, www.eurphysj.org

ACCEL Instruments GmbH, Bergisch Gladbach

BRUKER, Wissembourg, France

DANFYSIK A/S, Jyllinge, Denmark

SIGMAPHI, Vannes, France

SFAR STEEL, Creusot, France
THALES Electron Devices, Velizy, France
 V

The European Physical Journal A

Volume 28 Supplement 1 2006

 Foreword 71 M. Vanderhaeghen
Two-photon physics

81 M. Ostrick
 Many Body Structure of Strongly Electromagnetic form factors of the nucleon
Interacting Systems Experiments at MAMI

1 R.G. Milner 91 H. Schmieden

The beauty of the electromagnetic probe
7 L.S. Cardman
Physics at the Thomas Jeerson National Accelerator
Facility
Photo- and electro-excitation of the -resonance
at MAMI
101 S. Kowalski
Parity violation in electron scattering

19 W.U. Boeglin
Few-nucleon systems at MAMI and beyond 107 F.E. Maas
Parity-violating electron scattering at the MAMI
facility in Mainz
29 D. Rohe The strangeness contribution to the form factors
A1 and A3 Collaboration of the nucleon
Experiments with polarized 3 He at MAMI

117 N. dHose
39 M. Schwamb Virtual Compton Scattering at MAMI
Few-nucleon systems (theory)

49 H.-W. Hammer 129 H. Merkel

Nucleon form factors in dispersion theory Experimental tests of Chiral Perturbation Theory

59 S. Scherer 139 W. Hillert

Chiral perturbation theory The Bonn Electron Stretcher Accelerator ELSA:
Success and challenge Past and future
VI

149 A. Jankowiak 197 M. El-Ghazaly et al.

The Mainz Microtron MAMI Past and future X-ray phase contrast imaging at MAMI

161 A. Thomas 209 B.A. Mecking

The Gerasimov-Drell-Hearn sum rule at MAMI Twenty years of physics at MAMI What did it
mean?
173 R. Beck
Experiments with photons at MAMI

185 W. Lauth et al.

 Author index
Coherent X-rays at MAMI
Eur. Phys. J. A 28, s01, VII VIII (2006)
DOI: 10.1140/epja/i2006-09-022-5 EPJ A direct
electronic only

Foreword

This volume contains the proceedings of the Symposium on Twenty Years of Physics at the Mainz Microtron
(MAMI), which was held at the Johannes Gutenberg-Universitat Mainz, October 19-22, 2005. The Symposium marks
the retirement of several members of the Institut fur Kernphysik whose work has been devoted primaryly too scientic
research at MAMI over many years. It was the primary aim of the Symposium to review past and current activities in
the eld of hadronic structure investigations with the electroweak interaction. However, the Symposium also gave an
outlook on the physics with the MAMI upgrade, a double-sided mictrotron that is expected to provide a high-quality
beam of up to 1.5 GeV later this year.
The Institut fur Kernphysik was founded in the early 1960s by the late Hans Ehrenberg who served as its director
for more than two decades. He provided the Institute with a 350 MeV pulsed linear electron accelerator, which became
available in 1966 for studies of charge and magnetization distributions in nuclei and nucleons as well as photonuclear
investigations in collaboration with the Max-Planck-Institut fur Chemie.
Hans Ehrenberg knew about the importance of having excellent facilities for performing outstanding physics from
his earlier studies at Bonn and Stanford, with the later Noble Prize winners Paul and Hofstadter, respectively. There-
fore, he dedicated great eort in I) building up a perfect infrastructure of mechanics, electronics, vacuum and computer
workshops, and II) attracting a young accelerator physicist, Helmut Herminghaus, to the Institute.
In the late 1960s it became common wisdom that the next accelerator generation had to provide a high duty-
factor in order to perform coincidence experiments for detailed studies of hadronic physics. Helmut Herminghaus had
conceived a blueprint for such a device in 1975, a three-stage racetrack microtron (RTM). Shortly after a physics
program around this RTM was worked out and the proposal was sent to the sponsoring agencies. The project received
the support of the University and the State of Rheinland-Pfalz and sometime later also of the federal agencies. In
the fall of 1978, the state minister was informed by the federal minister of research and technology (BMFT) that the
project had been discussed with the German Science Council, the Deutsche Forschungsgemeinschaft (DFG), the Max-
Planck-Gesellschaft, and members of the scientic community. As a result these representatives agreed to support the
proposal in order to I) demonstrate that also a large-scale research facility can be realized at a university, II) withstand
a further emigration of such research from the universities, and III) nd a constructive solution that could serve as
a model for university research. As a matter of fact such a solution was found in the following years. However, it has
to be said that the full nancial support would never have arrived if the RTM had not been designed stage by stage,
and each time delivered in perfect shape (often to the surprise of outside experts) by Helmut Herminghaus and his
crew of physicists and technicians. The rst stage of the RTM (14 MeV) went into operation already in May 1979, the
second stage (183 MeV) followed in 1983, and the last stage was ready for the experiments in the fall of 1990.
At present the microtron delivers a continuous beam of an intensity of about 100 A for unpolarized and 40 A for
polarized electrons with a polarization degree of about 80 %. Its energy close to 1 GeV provides the perfect resolution
to study the distributions of charge, magnetization, and strangeness inside the nucleon and light nuclei, the threshold
production of the Goldstone bosons pion and eta, the polarizabilities of nucleons and pions, and the excitation of
the most prominent nucleon resonance, the (1232). Since the physics with the rst two stages of the RTM was
summarized already at an earlier workshop ( Physics with MAMI A ), the present Symposium concentrates on the
achievements of the years with the 855 MeV stage (MAMI B). The organizers also decided to invite as speakers, with
a few exceptions, young colleagues who have made a career with their work at MAMI.
It remains to say thank you to many people and institutions for continuing support. We are grateful to all the
colleagues from the Institute, the postdocs, Ph.D. and younger students who contributed to the MAMI project.
VIII

Special thanks go to the people in the workshops and in the administration without whose eorts the project could
never have succeeded. We are grateful to the colleagues from the neighbouring Institut fur Physik for their work on
polarized beams and targets, for the TAPS detector brought to Mainz by the Gie en group, to the Bonn/Bochum
group for the polarized H2 -target, and to many other German institutions for active engagement and various detection
devices, notably Darmstadt, Erlangen, Gottingen, and Tubingen. Our thanks go to the foreign colleagues who have
participated in the project from the very beginning, notably to our Scottish colleagues who built the photon tagger
with the support of their SERC, the groups from Pavia sponsored by the INFN, from Saclay supported by the
CEA/DAPNIA and from Orsay supported by the CNRS. We appreciate common experimental and theoretical work
with physicists from various other places in Europe, e.g. Amsterdam (NIKHEF), Basel, Genova, Gent, Lljubljana,
Trento and several Russian universities and institutions, and from overseas, e.g., Jeerson Lab, MIT, Florida State
University, University of Nagoya, George Washington University, and TRIUMF. Finally, in view of the upgrade two
more collaborations have developed in recent years. The Crystal Ball Collaboration has shipped its detector from the
Brookhaven National Lab to Mainz, and the KAOS detector is being installed in Mainz with the help of the GSI
Darmstadt. Last but not least we are grateful to the members of the international Program Advisory Committee and
of numerous evaluation and expert committees for their invaluable scientic advice and moral uphold.
Concerning the institutions we rst and foremost thank our Physics Faculty, the Johannes Gutenberg-Universitat
and the State of Rhineland-Palatinate for continued and coherent support. We are extremely grateful to the state
and to the federal ministries (BMFT, BMBW, BMBF) who nanced the construction of the new accelerator and
experimental halls as well as the large spectrometers via the university construction program (HBFG). Our special
thanks go to the Deutsche Forschungsgemeinschaft that backed up the project by means of Collaborative Research
Centers (SFB 201, CRC 443) whose resources were of the utmost importance to sustain our postdoc and PhD program.
Finally, we received recent support by the European networking activities via the I3HP/Transnational Access program.
Last but not least the organizers are grateful to the speakers of this Symposium for summarizing the various
achievements with MAMI and related research, and for bringing back memories of the past. Though retirees enjoy
the latter aspects very much, there is no reason to engage in retrospection: The double-sided microtron is expected to
yield its 1.5 GeV electron beam later this year, and we wish our colleagues and their students all the success in the
years to come!

Mainz, April 1, 2006

Hartmuth Arenhovel
Hartmut Backe
Dieter Drechsel
Jorg Friedrich
Karl-Heinz Kaiser
Thomas Walcher
The Editors
Eur. Phys. J. A 28, s01, 1 5 (2006)
DOI: 10.1140/epja/i2006-09-001-x EPJ A direct
electronic only

The beauty of the electromagnetic probe

R.G. Milnera
MIT-Bates Linear Accelerator Center, Laboratory for Nuclear Science, Massachusetts Institute for Technology, Cambridge, MA
02139, USA

/
Published online: 15 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. Precision experiments using the electromagnetic probe have recently produced important new
data on fundamental properties of the nucleon, e.g. charge, magnetism, shape, polarizability, spin and
sea quark structure. These experiments have been made possible by a new generation of high duty factor
electron accelerators, advances in spin polarization technology (beams, targets and recoil polarimeters),
and the development of unique, optimized detector systems. In this contribution, the role of multiple
photon exchange in electron scattering from the proton and the role of sea quarks in nucleon structure are
highlighted.

PACS. 13.40.Gp Electromagnetic form factors 13.60.-r Photon and charged-lepton interactions with
hadrons 13.60.Fz Elastic and Compton scattering 14.20.Dh Protons and neutrons

1 Introduction form-factors have been carried out. In particular,

Understanding the structure of the nucleon in terms of
the fundamental constituents of the Standard Model, the
quarks and gluons of Quantum Chromodynamics (QCD),
is a major research area in Physics. The ultimate goal
is to test QCD with precision measurements and ab ini-
tio calculations. Over the last decade, experimentalists
have made substantial progress in determination of the
quark and gluon distributions at high energies (ECM
100 GeV) and measurement of fundamental properties of
the nucleon at low energies (ECM 1 GeV). Theorists are
starting to produce full QCD Monte Carlo simulations (al-
beit with heavy pion masses) of nucleon structure using
advanced computers [1].
The experimental study of the structure of the proton
and of atomic nuclei is best carried out using the point-
like electroweak probe, the best understood interaction in
Nature. Intense beams of highly polarized electrons have
become available at energies of 0.5 to 6 GeV at high duty
factor. Highly polarized proton, deuteron and 3 He tar-
gets have been developed as well as e cient polarimeters
for detection of recoil polarization. Optimized experiments
utilizing uniquely designed detectors have been carried
out. New data and insights have been obtained in mea-
surement of the following properties of the nucleon:
The proton and neutron charge and magnetism
through spin-dependent elastic electron scattering
at Mainz [2], Bates [3], NIKHEF [4] and JLab [5].
Precise measurements of all four of the nucleon elastic
a
e-mail: milner@mit.edu
the relatively small neutron electric form-factor has
been determined to better than 7% over the range
0.1 < Q2 < 2 (GeV/c)2 .
The shape of the proton through study of electroex-
citation of the 0 at the (1232)-resonance at low
Q2 0.1 (GeV/c)2 using out-of-plane detection at
Bates and Mainz [6]. It has been established that
the proton shape is slightly non-spherical. A chiral
extrapolation [7] of lattice QCD calculations [8] is in
good agreement with the data.
The electric and magnetic polarizabilities of the
proton through measurement of Virtual Compton
Scattering from the proton at Mainz [9] and JLab [10]
and using out-of-plane detection at Bates [11].
The quark and gluon contributions to the spin
structure of the proton using deep inelastic scat-
tering at HERMES/DESY [12], JLab [13], COM-
PASS/CERN [14] and RHIC-spin [15].
The role of strange quarks in the long distance mag-
netic and electric charge distribution of the proton at
Bates, Mainz and JLab [16,17]. There are hints of a
non-zero strange quark magnetic moment of the pro-
ton but these need to be conrmed by more precise
experiments.
Here I concentrate on two areas of research where im-
portant results have recently been obtained.
2 The European Physical Journal A
Fig. 1. The Jeerson Lab data [18] on the ratio GpE /GpM show-
ing the discrepancy between the recoil polarization (solid cir-
cles) and the Rosenbluth (other symbols) techniques.
2 Evidence for multiple photon eects in
elastic electron scattering from the proton
Essentially all electron scattering experiments to study
proton and nuclear structure to date have been analyzed
in terms of single photon exchange. The ne structure cou-
pling constant 1/137 is small enough that leading or-
der has been adequate. There are a few specic examples
where multiple photon exchange is known to be signif-
icant, e.g. in comparison of electron and positron scat-
tering in kinematics where the single photon exchange
cross-section is small, or in radiative processes. Thus, it
came as a surprise when the Jeerson Lab Hall A recoil
polarization measurements of electron-proton elastic scat-
tering at momentum transfers of about 2 (GeV/c)2 [18]
showed a substantial deviation from the data obtained
over several decades with the Rosenbluth technique [19],
which is based on precise cross-section measurements.
This discrepancy has been interpreted as the eect of mul-
tiple photon exchange in the elastic electron-proton cross-
section [20]. The cross section for elastic electron-proton
scattering in the one-photon exchange approximation can
be written in terms of the pointlike Mott cross-section, the
Sachs form factors GpE and GpM and the electron scattering
angle as
   
d d Gp2 + G p2
p2
= E M
+ 2 GM tan 2
, been studied extensively. The eects of sea quarks and
d d M ott 1+ 2 gluons are relatively poorly determined, in large part be-
where = Q2 /4M 2 . Figure 1 shows the recoil polarization
determination of GpE /GpM (solid circles) as a function of
momentum transfer Q2 . The Rosenbluth data (all other
data points) are believed to be uncorrected for the eects
Fig. 2. The quark and gluon momentum distributions at
Q2 = 10 (GeV/c)2 as a function of parton momentum x as de-
termined by the ZEUS experiment [23] at the HERA electron-
proton collider. Note that the sea quark momentum xS and the
gluon momentum xg distributions are divided by a factor of 20.
of multiple photon exchange and so give an incorrect de-
termination at higher Q2 , i.e. above about 1 (GeV/c)2 .
This multiple photon exchange contribution to elas-
tic electron-proton scattering can be conrmed by precise
comparison of electron-proton with positron proton elas-
tic scattering or by measurement of the asymmetry Ay in
scattering of unpolarized electrons from a vertically polar-
ized proton target [21]. If conrmed, this is a very signi-
cant result.
3 Role of sea quarks in nucleon structure
QCD tells us that the nucleon comprises three valence
quarks and a sea of quark-antiquark pairs. From the ear-
liest days of nuclear physics, these sea quarks in the form
of mesons, have been viewed as playing an important role
in the long distance structure of the nucleon e.g. the mag-
nitude and sign of the proton and neutron magnetic mo-
ments. In addition, the most successful hadronic theoreti-
cal descriptions of light nuclei incorporate meson exchange
between nucleons as an essential element of nuclear bind-
ing. This meson cloud structure to the nucleon has gen-
erally been accepted but has lacked both a rigorous the-
oretical underpinning and a denitive quantitative basis
from experiment.
The role of valence quarks in nucleon structure has
cause they require high center-of-mass energy, and are a
major focus of interest for the future [22]. One of the im-
portant contributions over the last decade has been the
experimental measurement of deep inelastic scattering at
high energies to determine the eects of the sea quarks and
 R.G. Milner: The beauty of the electromagnetic probe
Fig. 3. Comparison of the gluon and sea distributions from the
ZEUS-S NLO QCD t for various Q2 values [23] as measured
at the HERA electron-proton collider.
gluons. In particular, data taken by experiments at the
HERA electron-proton collider [23] have for the rst time
allowed a determination of the gluon momentum distribu-
tion in the proton, as shown in g. 2. The QCD evolution
of HERA data [23] shows a signicant sea contribution at
low Q2 , in contrast to the gluon contribution which van-
ishes, as seen in g. 3. This supports the point of view of
a strong role for sea quarks at low Q2 .
At low energies, electron scattering experiments deter-
mine the elastic electric and magnetic form factors of the
proton and neutron. Friedrich and Walcher have postu-
lated that the Q2 dependence of the elastic form factors
in the region 0.1 to 0.5 (GeV/c)2 may be sensitive to the
meson cloud structure of the nucleon and have produced
parameterizations of world data which suggest that there
may be experimental support for this ansatz [24]. They
t the measured four form factors with a parameteriza-
tion which consists of a smooth contribution and a bump
contribution. Figure 4 shows the worlds data for the pro-
ton elastic form factor plotted as a function of momentum
transfer Q2 , where the smooth contribution is subtracted.
3
Fig. 4. The proton charge elastic form-factor with the smooth
contribution subtracted in the parameterization of Friedrich
and Walcher [24].
A 2% dip in the parameterization is obvious at Q2 0.1
0.2 (GeV/c)2 , which coincides with the location of the
peak in the neutron charge elastic form-factor GnE . In the
absence of realistic QCD calculations, it is hard to deni-
tively state that this structure at low Q2 is due to the
meson cloud structure of the nucleon. However, it is a
physically plausible explanation.
4 BLAST Experiment at MIT-Bates
A new set of precision measurements of the low Q2 elastic
form factors of the proton and neutron have been carried
out using the South Hall Ring (SHR) at the MIT-Bates
Linear Accelerator Center. The Bates Large Acceptance
Spectrometer Toroid (BLAST) was constructed [25] to de-
tect scattered electrons, protons, neutrons and pions in the
scattering of longitudinally polarized electrons with an en-
ergy of 850 MeV from polarized targets of hydrogen and
deuterium. The polarized internal gas target technique of-
fers minimal systematic uncertainties and a high statistics
sample of data were taken by the BLAST experiment over
an eighteen month period from late 2003 to mid 2005.
The BLAST data are under analysis and will be able
to provide new and independent experimental constraints
of the Friedrich-Walcher ansatz.
The polarized protons and deuterons (both vector and
tensor) were produced using an Atomic Beam Source
(ABS) [26], which was located in the substantial and spa-
tially varying magnetic eld of the BLAST toroid. The
target spin state was alternated every ve minutes by
switching the nal RF transition immediately before the
target to ensure equal target densities for each of the three
states (vector +, vector , tensor ). The electrons scat-
tered from the polarized protons and deuterons in a cylin-
drical, windowless aluminum target tube 600 mm long,
15 mm in diameter and with a wall thickness of 50 m.
The polarized target was tuned and monitored using a
Breit-Rabi system which continuously sampled the atomic
polarization of a small fraction of the incoming beam from
the ABS. The vector polarizations of both the proton and
deuteron was typically 0.75. Data were taken with stored
electron beam intensities up to 225 mA.
4 The European Physical Journal A
Fig. 5. A schematic layout of the BLAST experiment at MIT-
Bates.
Fig. 6. The vector asymmetry AVed in quasielastic (e, e p) scat-
tering from vector polarized deuterium as a function of missing
momentum pm for 0.1 < Q2 < 0.2 (GeV/c)2 , as measured by
the BLAST experiment [27].
The polarized electron beam originated from a GaAs
polarized electron source and the storage ring was lled
with alternating electron polarizations approximately ev-
ery half hour. The longitudinal beam polarization at the
target was maintained using a Siberian Snake solenoid sys-
tem. The beam polarization was continuously monitored
using a laser Compton backscattering polarimeter, located
upstream of the injection point in the SHR. The average
beam polarization over the BLAST data taking period was
0.65.
The BLAST (see g. 5) consisted of eight copper coils
which provided a 0.4 Tesla toroidal magnetic eld. For
these measurements it was instrumented with symmet-
ric detectors in the horizontal plane: three drift chambers
for momentum, angle and position determination; plastic
scintillators for triggering and time of ight, and Cerenkov
detectors for pion rejection. In addition, large plastic scin-
tillators for neutron detection were arranged on one side.
The background rate for scattering from the target cell
was measured and found to be negligible.
Figure 6 shows a fraction of the BLAST data ac-
quired in quasielastic (e, e p) scattering from vector po-
larized deuterium [27]. The scattering asymmetry AVed is
plotted as a function of the missing momentum (pm ) of
the proton in the nucleon. The solid curve is the predic-
tion of a Monte Carlo simulation which uses Arenhovels
theory [28] as well as a realistic description of the exper-
iment. At low pm , the scattering is dominated by the S-
state in deuterium and the asymmetry is very close to
that for scattering from a free proton. These data can be
used to determine the product of beam and target vec-
tor polarization. At high pm , the scattering is dominated
by the D-state in the deuteron, where both proton and
neutron spins are anti-aligned with respect to the nuclear
spin. Thus, the scattering asymmetry changes sign. The
pm range of the data extend out to 500 MeV/c. BLAST
data on the four elastic form factors of the proton and
neutron are expected to be published in 2006. In addi-
tion, a sizable data set on electron scattering from tensor
polarized deuterium was acquired with BLAST.
5 Conclusion
The electromagnetic probe provides a beautiful and pre-
cise means to study strongly interacting matter. We are
fortunate to witness great advances in accelerator and ex-
perimental technology so that the full power of the electro-
magnetic probe can be exploited to study hadronic mat-
ter. The two examples discussed above indicate that new
insight into Nature is being provided by the elementary
elastic electron nucleon scattering reaction, particularly
with spin polarization techniques.
The role of the sea quarks/meson cloud in nucleon
structure continues to be a subject of signicant inter-
est. Precision determination of the elastic form factors at
low momentum transfers from BLAST may conrm the
ansatz of Friedrich and Walcher. Conrmation of a dip in
the proton electric and magnetic form factors as well as
the neutron magnetic form factor at Q2 0.15 (GeV/c)2
will not denitively quantify the role of the meson cloud
but it will demand of theorists a convincing explanation.
I note that the recent G0 data [29] on the linear combi-
nation of the electric and magnetic strange form factors
of the proton suggest a Q2 dependence at similar values
of Q2 to that of the dip. Is this signicant? Clearly, more
precise data are needed.
The determination of GnE as a function of Q2 by many
laboratories over a decade has clearly been a triumph
for the eld of electromagnetic nuclear physics. With the
BLAST data, it is expected that this quantity will be de-
termined to better than 5% at low momentum transfers.
It is anticipated that this will quantitatively constrain the
meson cloud contribution to the charge distribution of the
neutron.
The experimental and theoretical contributions at
MAMI, particularly by our ve distinguished colleagues
who are honored here, have been important to the signi-
cant progress made worldwide. It has been a pleasure and
a privilege to be part of this unique celebration. I con-
gratulate Profs. H. Arenhovel, H. Backe, D. Drechsel, J.
Freidrich, K.-H. Kasier, and Th. Walcher on their distin-
guished careers and I wish them every success in the next
phase of their lives.
 R.G. Milner: The beauty of the electromagnetic probe
The author would like to acknowledge discussions with A.M.
Bernstein, T.W. Donnelly, R. Miskamen, A.H. Mueller, J.W.
Negele, and C.N. Papanicolas. In addition, the author would
like to acknowledge that the BLAST experiment is the fruit of
a dedicated collaboration over an extended period of time. In
particular, an outstanding cohort of graduate students is play-
ing an essential role. The authors research is supported by the
United States Department of Energy under the Cooperative
Agreement DE-FG02-94ER40818.
References
1. R.G. Edwards et al., Phys. Rev. Lett. 96, 052001 (2006).
2. D. Rohe et al., Phys. Rev. Lett. 83, 4257 (1999).
3. T. Eden et al., Phys. Rev. C 50, R1749 (1994).
4. I. Passchier et al., Phys. Rev. Lett. 82, 4988 (1999).
5. R. Madey et al., Phys. Rev. Lett. 92, 122002 (2003).
6. N. Sparveris et al., Phys. Rev. Lett. 94, 122003 (2005).
7. V. Pascalutsa, M. Vanderhaeghen, Phys. Rev. Lett. 95,
232001 (2005).
8. C. Alexandrou et al., Phys. Rev. Lett. 94, 021601 (2005).
9. J. Roche et al., Phys. Rev. Lett. 85, 708 (2000).
10. G. Laveissiere et al., Phys. Rev. Lett. 93, 122001 (2004).
11. P. Bourgeois et al., submitted to Phys. Rev. Lett. (April
2006).
12. A. Airapetian et al., Phys. Rev. D 71, 012003 (2005).
13. X. Zheng et al., Phys. Rev. C 70, 065207 (2004).
5
14. E.S. Ageev et al., Phys. Lett. B 633, 25 (2006).
15. J. Kiryluk (MIT) for the STAR collaboration, Proceed-
ings of PANIC 2005, October 2005, Santa Fe, New Mex-
ico; K. Boyle (Stony Brook) for the PHENIX collabora-
tion, Proceedings of PANIC 2005, October 2005, Santa Fe,
New Mexico, to be published by the American Institute of
Physics.
16. S. Kowalski, these proceedings.
17. F. Maas, these proceedings.
18. V. Punjabi et al., Phys. Rev. C 71, 055202 (2005).
19. I.A. Qattan et al., Phys. Rev. Lett. 94 (142301) (2005).
20. P.A.M. Guichon, M. Vanderhaeghen, Phys. Rev. Lett. 91,
142303 (2003); P.G. Blunden, W. Melnitchouk, J.A. Tjon,
Phys. Rev. Lett. 91, 142304 (2003).
21. A.A. Afanasev et al., Phys. Rev. D 72, 013008 (2005).
22. A. Deshpande, R. Milner, R. Venugopalan, W. Vogelsang,
Annu. Rev. Nucl. Part. Sci. 55, 165 (2005).
23. S. Chekanov et al., Phys. Rev. D 67, 012007 (2002).
24. J. Friedrich, Th. Walcher, Eur. Phys. J. A 17, 607 (2003).
25. BLAST Technical Design Report August 10th, 1997.
26. D. Cheever et al., Nucl. Instrum. Methods A 556, 410
(2006).
27. A. Maschinot, MIT PhD Thesis 2005 (unpublished).
28. H. Arenhovel, W. Leidemann, E.L. Tomusiak, Phys. Rev.
C 52, 1232 (1995); 46, 455 (1992); Z. Phys. A 331, 123
(1988); 334, 363 (1989).
29. The G0 Collaboration, Phys. Rev. Lett. 95, 092001 (2005).
Eur. Phys. J. A 28, s01, 7 17 (2006)
DOI: 10.1140/epja/i2006-09-002-9 EPJ A direct
electronic only

Physics at the Thomas Jeerson National Accelerator Facility

L.S. Cardmana
Thomas Jeerson National Accelerator Facility, 12000 Jeerson Avenue, Newport News, VA 23606, USA and University of
Virginia, Department of Physics, 382 McCormick Rd., P.O. Box 400714, Charlottesville, VA 22904-4714, USA

/
Published online: 31 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. The Continuous Electron Accelerator Facility, CEBAF, located at the Thomas Jeerson Na-
tional Accelerator Facility, is devoted to the investigation of the electromagnetic structure of mesons,
nucleons, and nuclei using high energy, high duty-cycle electron and photon beams. Selected experimental
results of particular interest to the MAMI community are presented.
PACS. 29.17.+w Electrostatic, collective, and linear accelerators 25.20.-x Photonuclear reactions
25.30.Bf Elastic electron scattering 25.30.Dh Inelastic electron scattering to specic states

1 Personal Comments
It is an honor and a pleasure to be here to celebrate the
achievements of MAMI and the distinguished careers of
Professors Arenhovel, Backe, Drechsel, Friedrich, Kaiser,
and Walcher.
We are all deeply aware of the extent to which the
science we do builds on the achievements of those who
have gone before us, and on the insights and hard work
of our colleagues working in the eld today. One of my
very earliest memories as a scientist, dating from the days
when I was a young graduate student, is that of attending
Photonuclear Physics Boot Camp (otherwise known as
the Photonuclear Gordon Conference) and learning from
(and with) many of those retiring today.
Thomas (Walcher) was one of the very rst scientists
I ever knew beyond the boundaries of my own laboratory.
He came to visit us (at Yale), and I went and visited him
and his colleagues at Darmstadt. It has been a great plea-
sure to follow his distinguished career in science, from low-
Q2 electron scattering to hadronic physics at CERN and
beyond, and nally to the leadership role he has played at
MAMI for many years.
Hartmuth (Arenhovel) has been the keeper of the ame
of all knowledge about the deuteron, and a worthy succes-
sor to Gregory Breit. You should know that I was a grad-
uate student at Yale, and it was one of Professor Breits
missions in life to convince any and all who would listen
that the deuteron was the essence of nuclear physics, and
that until we understood the deuteron, we did not un-
derstand anything. I think it is fair to call Hartmuth the
Gregory Breit of my generation; he has made so many
contributions.
a
e-mail: cardman@jlab.org
Dieter (Drechsel) has always been one of those peo-
ple I have looked to as the source of the big picture in
nuclear physics. He has provided us with deep insights, a
sense of direction, and an understanding of what is really
important. He has also been an inspiring example here at
Mainz of the tremendous benets to everyone of having a
close collaboration between theory and experiment.
Jorg (Friedrich) has taught us all how to analyze and
interpret electron scattering data with minimal prejudice
(and, therefore, maximal honesty). It is a delight to see
the same rigorous approach that was so successful in the
study of nuclei and their excited states now being applied
to nucleon structure.
Karl-Heinz (Kaiser) and his mentor, Helmut Herming-
haus, taught the world how to build superb continuous-
wave (cw) electron accelerators eectively and e ciently.
Karl-Heinz, in particular, through the design and con-
struction of the double-sided microtron, is leaving the In-
stitute well positioned for another generation of superb
experiments.
In conclusion, on behalf of so many people I have
worked with in nuclear physics, I want to thank each of you
for your many contributions to our eld, and to express
the hope we all share that for each of you retirement is
a formality, not a reality, and that you will continue to be
active for years to come.
2 Research at Jeerson Laboratory
The Thomas Jeerson National Accelerator Facility, also
called Jeerson Lab (or JLab), operates the Continu-
ous Electron Beam Accelerator Facility (CEBAF). CE-
BAF is a cw electron accelerator capable of delivering
three electron beams for simultaneous experiments in the
8 The European Physical Journal A

1.5
three experimental areas. Originally designed for 4 GeV,
GE
p

its present maximum energy is 5.7 GeV.

GMp 0.10
The CEBAF user community consists of about 2000 1.0

MIT-Bates: 2H(e,en)

physicists; more than half of them are actively involved GE

n 0.08 Mainz A3: H(e,en)

2
NIKHEF: H(e,en)

in the experimental program. In addition to its main mis- 0.5

0.06

sion, JLab contributes to the development and use of Free 0.04

Electron Lasers, to medical imaging, and to community 0.02

Galster

outreach programs. 0.0

0 2 4 6
0
0 1 2
The intellectual and technical foundations for the con- Q2 (GeV2) Q2(GeV2)
1.2
struction of CEBAF were provided by the scientic suc- p
GM
Borkowski
Sill
Bosted
n
GM Selected World Data
Lung
Rock
Bartel
Walker
1.4
cesses of earlier electron accelerators (the generation that pGD 1.0 Andivahis
nGD
Arnold
Jourdan1
Jourdan2
1.2
included Saclay, MIT-Bates, NIKHEF, and, to some ex-
Gao
Xu

tent, SLAC), and by the enhanced research opportunities
provided by cw electron beams as demonstrated at fa-
cilities such as MAMI. CEBAF is a large, international
laboratory with a broad research program; it has been in
operation for some seven and a half year now.
What are the goals of CEBAFs research program? Ba-
sically, we aim to understand strongly-interacting matter.
How are the hadrons constructed from the quarks and glu-
ons of QCD, and how does the nucleon-nucleon force arise
from the strong interaction? We further aim to identify
the limits of our understanding of nuclear structure by us-
ing the high precision attainable with the electromagnetic
probe and the possibiltiy of extending investigations to
very small distance scales. A specic issue that motivated
the construction of CEBAF was our desire to gain insight
into the question of where the description of nuclei based
on nucleon and meson degrees-of-freedom fails and the un-
derlying quark degrees-of-freedom must be taken into ac-
count. One can ultimately characterize all of this as trying
to understand QCD, not in the perturbative regime acces-
sible at very high energies and very short distance scales,
but in the strong interaction regime relevant to most of
the visible matter in the Universe. To make progress in
these areas, there are other critical issues that must be
addressed, such as the mechanism of connement, the dy-
namics of the quark interaction, and how chiral symmetry
breaking occurs.
To provides some shape and structure to the discussion
of the experiments, the CEBAF program can be organized
into half a dozen broad thrusts. This presentation will
concentrate on two of them:
How are the nucleons made from quarks and glue?
Where are the limits of our understanding of nuclear
structure
3 How are the nucleons made from quarks
and glue?
Among the most interesting puzzles in physics today are:
why there is this eective degree-of-freedom in QCD, the
nucleon; and how something as complicated as the resid-
ual QCD interaction between quarks in nucleons can be
characterized by a rather simple N-N potential? To pro-
vide experimental insights that will help us solve the rst
of these puzzles, the Jeerson Lab research community has
mounted an array of investigations in three broad areas:
0.8 1.0
0.8
0.6
0.6
0.4 0.4
0 10
2
20
2
30 0 1 2 3 4
Q (GeV ) Q2 (GeV2)
Fig. 1. Nucleon form factor data available before the start of
experiments using CEBAF. Top 2 panels: electric form factors,
bottom panels: magnetic form factors. Left 2 panels: proton
form factors, right panels: neutron form factors (adapted from
ref. [1]).
What are the spatial distributions of the u, d, and s
quarks in the hadrons?
What is the excited state spectrum of the hadrons,
and what does it reveal about the underlying degrees-
of-freedom?
What is the QCD basis for the spin structure of the
hadrons?
3.1 What are the spatial distributions of the u, d, and
s quarks in the hadrons?
Elastic electron scattering has provided most of our infor-
mation on the spatial distributions of the quarks in the
nucleons. The data on the four electromagnetic structure
functions of the nucleon, GE and GM for both the proton
and the neutron, available just prior to the start of exper-
iments at CEBAF is shown in g. 1. The magnetic form
factors of the proton and the neutron were known rea-
sonably well, but the electric form factors were not. The
electric form factor of the proton had not been determined
accurately enough to distinguish between a wide range of
theories based on rather dierent physics. First results on
the electric form factor of the neutron were available from
Bates, Mainz, and NIKHEF, but these data were limited
to moderate momentum transfers and, therefore, not sen-
sitive to the details of the distribution of charge inside the
neutron. The measured form factor was consistent with
the r.m.s. radius derived from neutron-electron scattering.
The present status of the nucleon form factors is
shown in g. 2. The measurements of the polarization
transfer from the incident electron to the elastically
recoiling proton have shown that the electric and mag-
netic form factors for the proton dier substantially. The
systematic dierences between the polarization transfer
data and the Rosenbluth results for GE /GM are likely
 L.S. Cardman: Physics at the Thomas Jeerson National Accelerator Facility 9

1.5

GE
p

GMp 1.0
0.10 JLab E93-038:
2
MIT-Bates: H(e,en)

2
H(e,en) JLab E93-026: 2H(e,en)
3
Mainz A1: He(e,en)

n 0.08 Mainz A3: 2H(e,en)
GE 3
Mainz A3: He(e,en)
2
NIKHEF: H(e,en)
0.06 Schiavilla & Sick:GQ

0.5
0.04

0.02
Galster New Fit
0.0 0
0 2 4 6 0 1 2
2
Q (GeV ) 2 Q2(GeV2)
1.2 Borkowski
p
GM Sill
Bosted GM
n this measurement
Lung
Rock
(with statistical errors) Bartel
Walker
1.4
pGD 1.0 Andivahis
n G D
Arnold
Jourdan1
Jourdan2
1.2 Gao
Xu

0.8 1.0

0.8
0.6 grey: estimated systematic
0.6 error

0.4 0.4
0 10 20 30 0 1 2 3 4
2 2
Q (GeV ) Q (GeV2)
2

Fig. 2. Present status of the nucleon form factor data including the CEBAF data (adapted from ref. [1]).

1.5 0.10
due to two-photon exchange eects modifying the results. pGE
p
n
Theoretical estimates suggest that the modications are GMp
GE
1.0
much smaller for the polarization transfer data than 0.05

for the Rosenbluth data, so the former are likely to

0.5
be more directly interpretable in terms of the nucleon
0.00
form factors. The electric form factor of the neutron has 0.0
now been measured up to a Q2 of 1.5 (GeV/c)2 using 0 2 4 6 8
Q2 (GeV2)
10 0 2 4
Q2 (GeV2)
polarization transfer techniques, and the data taken with 1.5 1.2
p n
dierent methods agree quite well. GM GM
pGD nGD1.0
The theoretical interpretation of the data is summa-
rized in g. 3. The theories that describe the data reason- 1.0
0.8
ably well reveal two key aspects of nucleon structure: the
importance of the pion cloud, and the importance of in- 0.6

corporating the relativistic motion of the quarks into the 0 2 4 6 8 10 0 2 4 6 8 10

theoretical description of the nucleon. Q2 (GeV2) Q2 (GeV2)

When one looks at these form factors in a phenomeno- Bijker (VMD)

Holzwarth (soliton)
Miller (rel. QCM + bag)
F2/F1 ln2(Q2/2)/Q2

logical way with minimum prejudice [2], what emerges is Hammer (VMD + disp. rel.)

some of the clearest evidence we have for the nucleons
pion cloud (see g. 4). Similar results have been obtained
using a dierent approach to model-independent analy-
sis [3] of nucleon form factors.
We plan to extend the proton form factor data to
9 (GeV/c)2 , where we may see evidence for a diraction
minimum. The neutron form factor will also be extended
to 5 (GeV/c)2 . Further extensions of a factor of two are
planned with the 12 GeV Upgrade. Such extensions have
Fig. 3. Theoretical descriptions of the nucleon form factor
data [1].
historically proven to be important, and we expect these
data will provide further insight and sensitivity for com-
pleting our understanding of how to construct nucleons
from quarks and gluons.
10 The European Physical Journal A

0.03

r 2 ( r; GEn ) / fm-1

bn p0
0.02 (1- bn)n0 G0
total HAPPEx
0.01
A4

0.00

-0.01

-0.02
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
r / fm

0.03
+ Fig. 5. Nucleon strange form factor data from parity-violating
r ( r; GEppol ) / fm-1

0.02
-bpp0
bpn0
electron scattering at forward angles as a function of Q2 .
total

0.01
G0
(ex A4 SAMPLE with
0.00
tra GA calculation
po
lat
ed
-0.01
0.1 HA )
PP
2

E
GsE
X-
H
-0.02 ('0
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4) 3
95% C.L.
r / fm HAPPEX-4He ('04) 2
0 1
+
Fig. 4. Neutron radial charge distribution (top) and the polar-
ization term contribution to the proton radial charge distribu-
tion (bottom) as inferred from an analysis using phenomeno-
logical models [2] of the world nucleon form factor data. The
-0.1
green solid lines show the pion cloud contribution. Note that Leinweber
r2 (r) is plotted, emphasizing the distribution at large radii. et al. (2005)

Q2 = 0.1 GeV2

GsM
-1 0 1
The strange quark form factors have become an in-
teresting area of study that is both analogous and com-
plementary to the classical electromagnetic form factors. Fig. 6. Separated strange nucleon form factors at Q2 =
By using the weak component of the electro-weak interac- 0.1 (GeV/c)2 [10].
tion we access the weak neutral current form factor, which
can be interpreted very elegantly in terms of the strange
quark distribution. Because there are no valence strange separating them spatially. Even at the highest momentum
quarks, this measurement provides a unique window on transfers reached experimentally we are averaging over a
the sea quark distribution. The strange form factors can distance scale that is roughly the size of the nucleon, so it
also be expected to provide us with interesting experi- is not too surprising that the result is small. There is an
mental insights into nucleon structure: by combining the intriguing suggestion in the data for something that one
electromagnetic and the weak neutral current form factors would call vaguely pion-cloud like behavior, but it is fair
we should be able to separate the spatial distribution of to say that the statistical signicance of this eect is not
the u, d, and s quarks. very high.
Figure 5 shows the worlds data on the strange proton The data taken at forward angles includes a mixture of
form factor taken at forward angles as a function of Q2 . electric and magnetic form factors. At Q2 = 0.1 (GeV/c)2
One sees data from the A4 experiment at Mainz [6,7,8], we have data at both forward and backward scattering
from HAPPEx I and II (JLab Hall A) [9,10,11], and from angles, so we can separate these eects (see g. 6). The
G0 (JLab Hall C) [13]. These di cult experiments would data favor a positive value of GsM , which is at variance
be impossible without highly polarized electron beams with most of the theoretical models. Experiments are in
from magnicently stable accelerators. The fact that the progress that will reduce the size of the error ellipse at
data from dierent laboratories lie roughly on a smooth this Q2 value by a about a factor of 3, and additional
curve gives one condence that the experimenters are do- experiments planned at both MAMI and JLab will permit
ing it right. separations at other Q2 values. A broad, world-wide eort
The rst thing that strikes you about the data is that will provide the results we want.
the form factor is rather small. This is to be expected, as Another interesting experiment is the measurement of
all of the strange quarks emerge as quark-antiquark pairs the pion form factor. The pion is the simplest QCD bound
popping in and out of the vacuum, and to get a nite system, the positronium of QCD. One expects that the
form factor there must be some kind of a polarizing eect pion form factor will provide us with evidence for the
 L.S. Cardman: Physics at the Thomas Jeerson National Accelerator Facility
0.6
Amendolia +e elastics
+
Previous p(e,e )n
Q F
JLab F-1 Results
2
BSE+DSE
0.4 QCD Sum Rule
IMIN ARYCQM
0.2
PREL
Perturbative QCD
JLab F-2 Results
0 Fig. 8. W-dependence of the scattered electron rate for the
0 1 2 3 4 5 6 7
2 2
Q (GeV/c)
Fig. 7. Pion form factor data [14].
transition of the strong interaction from the perturbative
(QED-like) to the strong (connement) regime at the low-
est possible momentum transfer. These data also constrain
phenomenological models of the pion.
Measuring the pion form factor is not simple. At low
Q2 , one can scatter pions o atomic electrons, but a bowl-
ing ball does not transfer energy to a ping-pong ball e -
ciently, and even with very high energy pions this exper-
iment cannot reach high momentum transfers. To reach
higher momentum transfer in the absence of a free pion
target, one must scatter electrons o virtual pions inside
a proton and extrapolate the data to the pion pole. The
worlds data (see g. 7) is beginning to distinguish be-
tween dierent theoretical approaches. With the 12 GeV
Upgrade, we expect to extend the data out to a momen-
tum transfer of 6 (GeV/c)2 and to be able to infer the
distance scale for the onset of perturbative behavior.
3.2 What is the excited state spectrum of the
hadrons?
If one looks at several decades worth of data on nucleon
resonances and tries to use a simple quark model to clas-
sify the states in terms of the excitation in units of h
and the angular momentum of the three quarks, the states
that have been identied so far t nicely into this scheme,
but there are many states that have been predicted but
have not been found. It is an interesting fact that one
can explain all of the states that have been seen so far
by assuming that the nucleon and its excited states are a
diquark-quark system. Since most of the data have been
obtained from pion-induced reactions, and many of the
missing states are predicted not to couple to pions, it is
also possible that the missing states may have been over-
looked for experimental reasons.
In atomic spectroscopy the line spacing is large com-
pared to the line width, and measuring the complete
spectrum is relatively straightforward. In nucleon spec-
troscopy, the strong interaction causes the width to be
comparable to the spacing. Identifying weak states and
11
8 p(e, e )X reaction. CLAS data taken at 4 GeV primary beam
energy. The energies of the known excited states are shown in
black, while those of the states missing in the simple quark
model description are shown in red.
extracting the internal quark structure from the measured
cross sections is a di cult task. The problem can be seen
easily in g. 8, which shows the inclusive electron scatter-
ing cross spectrum from the proton for a 4 GeV electron
beam. With a modern electron accelerator and a large ac-
ceptance detector one can obtain data on the transition
form factors over a large Q2 -range [1 4 (GeV/c)2 ] in a
single shot. There is plenty of cross section in the region
where the missing states (shown in red) have been pre-
dicted, but extracting their individual strengths from the
data is a real challenge.
The combination of cw electron beams and modern,
large solid angle detectors provides important advantages
for addressing this problem experimentally. If one looks
at the same data set of g. 8 but uses the information on
the energy and momentum of the nal state proton mea-
sured in coincidence with the inelastically scattered elec-
tron, it is straightforward to infer the missing mass asso-
ciated with the decay of the excited state (see g. 9). One
can see clearly from the raw missing mass spectrum that
the missing states do not couple to pions, but rather
to the and . With the further information on the an-
gular correlations of those decay particles relative to the
momentum transfer axis one nally has the information
necessary to decompose the spectrum of g. 8, and learn
just what is there.
This eort naturally begins with the (1232), which
decays predominantly into pion and nucleon. Figure 10
shows a comparison of separated structure functions from
CLAS data for the p(e, e p) o reaction with theoretical ts
and results from previous experiments.
The N-transition is characterized by three mul-
tipoles: the electric quadrupole E, the magnetic dipole M ,
and the scalar multipole S. As we examine this transi-
tion as a function of momentum transfer we expect that
dierent aspects of the excitation will become apparent.
At large distance scales (corresponding to low momentum
transfers) we should see the eect of the pion cloud, while
at large momentum transfer (corresponding to short dis-
tances) we will eventually reach the limit given by pQCD
12 The European Physical Journal A

Missing States
W(GeV)

2 thresh.

N(1680)

N(1520)

(1232)

Mx(GeV)

Fig. 9. W vs. missing-mass Mx for the same CLAS data shown

in g. 8 [15]. Fig. 11. Ratios REM and RSM as a function of Q2 for the
N -transition [17].

80

)
A 1/2 (10 -3GeV -1/2)

-1/2
150
60

S1/2 (10 GeV

100
40

-3
50 20
0 0

-50 -20
-40
-100
P11(1440) -60
P11(1440)
-150
0 1 2 3 4 0 1 2 3 4
Q2 (GeV2) Q2 (GeV2)
Fig. 12. A 1 (Q2 ) and S 1 (Q2 ) multipoles for the P11 (1440)
2 2
Roper resonance [18].

resonance. According to the constituent quark model the

Fig. 10. Angular distributions of the separated structure func-
tions for the p(e, e p) o reaction in the Delta region [16].
where REM = E/M 1, and we further expect that the
S/M ratio RSM will become constant.
Results from an early experiment at JLab and data
from MAMI and Bates, all in the low-Q2 regime show the
eect of the pion cloud clearly (see g. 11). As a function
of Q2 , REM remains small and negative at high Q2 with a
trend toward 0 and a possible sign change. RSM continues
to rise in magnitude with Q2 . No trend is seen towards Q2 -
independence. We can only conclude that even at Q2 of
10 (GeV/c)2 we are far from the pQCD regime. Pion cloud
models describe the data well (tted to low and high-Q2
points). Unquenched Lattice QCD gives the correct signs
and approximate magnitudes.
One of the most interesting examples of the impact of
the pion cloud and of the value of measuring the tran-
sition form factors for nucleon excitation is the Roper
N (1440)P11 state is an N = 2 radial excitation of the nu-
cleon. However, the properties of this state such as its mass
and photocouplings are not well described by this model.
The new CLAS data (see g. 12) seem to explain this
puzzle. At low momentum transfer, what one is measur-
ing is dominated by the pion cloud. As you start squeezing
down the distance scale, what emerges is the underlying
quark structure of the Roper, which is, in fact, roughly
consistent with a radial excitation.
Investigation of nucleon excitation through the mea-
surement of the transition form factors is now slowly mov-
ing up in excitation energy. Most of this analysis is at a
preliminary stage, and what is really needed is a coher-
ent study of many channels at many values of momentum
transfer in a consistent (and comprehensive) analysis. It
will be a long time before we have all the answers.
As we search through this data, we are coming across
intriguing evidence for states that have been missing .
For example, there is evidence for a possible new N*
state near 1840 MeV visible in the photo- and electro-
production data. In the forward hemisphere, one sees a
nice peak from a known N* state at 1.7 GeV; in the
 L.S. Cardman: Physics at the Thomas Jeerson National Accelerator Facility 13

2 2 1.0 1.2
-1. < Cos(K) < 0., Q = 0.7 (GeV/c)

at al.)
JLab E99-117 (3He)

C
al. HH
0.15 HF perturbed QM
T + L L E142 (3He)

pQCD

et ith
mas
World Data parm Q2 = 10 GeV2

)
de fit w
0.8 E154 (3He) 1.0

(Tho
Symmetric Q Wave function
n p

(L CD
r
Helicity 3/2 suppression
A1 HERMES (3He) A1

pQ
ea
model
Spin 3/2 suppression

)
al.

r)
gu
et
Bag
0.6

(Is
el
0.8

eig

M
CQ
(W
C
HH

n
to
o
w/ l.)

oli
lS
fit et a
0.1 0.4 SU(6)

ira
CD der 0.6

Ch
Q
p ea
(L
b/sr

Statistical model
0.2 (Soffer et al.)
0.4
CLAS-EB1b Q2 = 1.4 - 4.52 GeV2
HERMES
0 SMC
0.2
0.05 SLAC - E155
SLAC - E143
-0.2
0
known New
-0.4
N* N* ? 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
x x
1.6 1.7 1.8 1.9 2 2.1
W (GeV) Fig. 14. (Left) Spin structure function of the neutron, An 1,
derived from 3 He data [19]; and (Right) spin structure function
Fig. 13. W-dependence of the cross section for the p(e, e K + ) of the proton Ap1 [20].
reaction integrated over backward-going K + [15].

backward hemisphere (see g. 13), one sees an additional

unexpected structure. A detailed analysis shows that the
angular distribution can be t nicely with the addition of a
new P11 state at 1840 MeV with a width = 140 MeV to
the known D13 (1870) and D13 (2170) states. Intriguingly,
a P11 state at 1840 MeV is consistent with the symmetric
quark model and SU (6) O(3) symmetry, but is incon-
sistent with diquark-quark symmetry.
I feel obliged to bring you up to date on the penta-
quark (or lack thereof). There was a great deal of excite-
ment for a while about what appeared to be evidence for
a 5-quark state. There have three experiments at JLab
pushing to substantially higher statistics, both in the p
and the d channel, and for virtual photons as well. No
evidence for a 5-quark state has been found in the rst
analyses of these new data.
3.3 What is the QCD basis for the spin structure of
the hadrons?
In addition to the investigation of the spatial distributions
of charge and magnetization in the nucleon and its excited
state spectrum, the third important experimental focus is
the nucleons spin structure. The rst thing to look at is
the spin structure function of the valence quarks at high-
x. The data for the proton was reasonable; the new CLAS
data with somewhat tighter error bars are conrming the
old results and improving the overall accuracy. There were
no data of any statistical signicance for the neutron above
an x of 0.3. The 3 He experiment at JLab has provided
three new data points (see g. 14). The new data, when
folded into a global analysis of the parton distribution
functions (PDF), show that the theoretical prejudices used
in earlier analyses were wrong; in particular we now know
that d/d stays negative at high x.
One can make predictions with a minimum of theoret-
ical prejudice for the integrals of the spin structure func-
tions at the two extremes of distance scales. In the limit
of extremely small distances (i.e. for Q2 ), assuming
only isospin symmetry and current algebra (or the opera-
tor product expansion within QCD), Bjorken showed that
Fig. 15. GDH integral as a function of the upper limit
max [21].
the dierence between the proton and the neutron inte-
grals is related to the neutron -decay coupling constant,
with a small Q2 dependent correction due to the running
of the coupling constant.
For very large distance scales (i.e. for Q2 0), there
is a slightly less rigorous set of assumptions (Lorentz in-
variance, gauge invariance, unitarity, and the dispersion
relation applied to the forward Compton amplitude) that
can be used to show that the dierence between the helic-
ity 23 and 12 total cross sections is related to the nucleon
anomalous magnetic moment (this is the GDH sum rule).
There has been a lovely set of data taken at ELSA
and MAMI that have determined the GDH integral as
a function of the upper photon energy integration limit
(see g. 15). The experiments were technically challeng-
ing [21], requiring the combination of polarized electrons,
a polarized target, and large-acceptance detectors. Theo-
retical analysis and interpretation of these data show that
the GDH sum rule is satised at the 5% level. The eort
has also provided us with a better understanding of the
physics of the reactions contributing to the integral. These
data, and the precision with which they have dened the
GDH integral at the photon point, provide the foundation
for our studies of the Q2 evolution of the moment of the
nucleons spin structure functions.
As one looks at the evolution of the moment of the
proton spin structure function with Q2 , one expects to see
14 The European Physical Journal A

s,g1 / JLab
0.16 Burkert-Ioffe
SLAC E143 2 s,g1 / world data GDH constraint
CLAS EG1a
1 p (no elastic)

0.14 s,F3 / pQCD evol. eq.

CLAS EG1b s, / OPAL
1
0.12
HERMES
0.1 0.5

(Q)/
0.08

s
0.06

0.04 0.1

0.02
0.05 -1
10 1 10
0
Q (GeV)
-0.02 Fig. 18. Eective strong-coupling constant as a function of Q2
0 0.5 1 1.5 2 2.5 3 derived from the Q2 -dependence of the Bjorken integral [23].
Q (GeV )
2 2

Fig. 16. Integral of the spin structure function of the proton of very low momentum transfer with high precision to test
as a function of Q2 [22]. the predictions of PT.
The rst signicant measurement of the Q2 -depen-
dence of the Bjorken integral (see g. 17) was made for
Q2 = (0.05 2.5) (GeV/c)2 . Remarkably, pQCD-based
0.20
Q2 evolution matches the data down to a Q2 of about
0.7 (GeV/c)2 . Deur et al. [23] have made an interesting in-
p-n
1

terpretation of the Q2 -dependence of the Bjorken integral

0.15 in terms of an eective strong-coupling constant ef f (Q2 )
(see g. 18). Again, there is evidence for a transition oc-
curring around Q2 = 1 (GeV/c)2 .
0.10

4 Explore the limits of our understanding of

0.05
nuclear structure
As described above, experiments at Jeerson Lab are pro-
viding essential new insights into nucleon structure. In a
very similar way, the precision, spatial resolution, and in-
0 0.5 1.0 1.5 2.0 2.5 3.0
2 2
terpretability of experiments performed using electromag-
Q (GeV/c) netic probes are being used to address long-standing issues
Fig. 17. Bjorken integral (proton-neutron dierence) as a
in nuclear physics, including specically nucleon-nucleon
function of Q2 . The grey band shows the evolution of ef f (Q2 ) correlations and the identication of the limits of our un-
predicted by pQCD [22]. derstanding of nite nuclei.

4.1 Correlations in nuclei

the anomalous magnetic moment of the proton in the long
wavelength limit, whereas at innite Q2 the Bjorken sum
rule is valid. In the regime close to the long wavelength
limit, chiral perturbation theory (PT) allows us to make
predictions. The transition between the two extremes is
an important piece of information on how the nucleon is
put together, and how nucleon structure emerges from the
parton soup.
We have data now, mainly from JLab, on the evolu-
tion of the structure functions integral for the proton (see
g. 16) and the neutron approaching the GDH sum rule
limit at Q2 = 0, and approaching the Bjorken limit at a
surprisingly low momentum transfer of about 1 (GeV/c)2 .
Several experiments at JLab are investigating the region
Nucleon-nucleon correlations have been a subject of great
interest since the beginnings of the eld. In his fabled
bible on nuclear physics, Hans Bethe estimated that
these correlations should be of scale a third of what one
observes in nuclear physics, and indeed they are. However,
nding clear, interpretable evidence for these correlations
has been a real challenge to experimentalists.
The previous generation of (e, e p) experiments car-
ried out at Saclay, NIKHEF, and Mainz explored the
spectral function strength for low-lying shells. Only about
2/3 of the strength anticipated from a simple shell model
was found. However, the interpretability of these measure-
ments was limited by the uncertainties introduced by the
 L.S. Cardman: Physics at the Thomas Jeerson National Accelerator Facility 15

3 a) 1e-09
r(4He/3He)

Experiment
Correlated Basis Function

S(Em,pm) [MeV sr ]
Theory

-1
2 1e-10

-4
1 1e-11
4 0.250

pm
b)
r(12C/3He)

(G
3

eV
1e-12 0.410

c)/
2
0.570
1 1e-13
6 c)
0 0.1 0.2 0.3
Em (GeV)
r(56Fe/3He)

4 Fig. 20. Spectral function for (e, e p) at high momentum trans-

fer and high missing energy [25].
2

1 1.5 2 2.5
xB

Fig. 19. Ratio of inelastic scattering cross sections o nuclei

relative to 3 He as a function of Bjorken-x [24].

corrections necessary for the nal-state interactions of the

knocked-out protons.
A new approach to nucleon-nucleon correlations avoids
this problem by comparing the ratio of inelastic electron
scattering o 4 He, 12 C, and 56 Fe to 3 He in a kinematical
regime where the scattering is basically from the quarks
within the nucleons, and the scattering from the nucleons
as coherent objects is highly suppressed. These data (see
g. 19) tell us that at any given moment the number of cor-
related nucleons in 4 He, 12 C, and 56 Fe is 0.3, 1.2, and
6.7, respectively. So about 10% of the time a nucleon is
involved in a nucleon-nucleon correlation. The measure-
ments further show that three-nucleon correlations are
clearly present (at x > 2), and about an order of mag-
nitude smaller than two-nucleon correlations.
Another approach [25] to the study of correlations is
to search explicitly for the strength that was identied as
missing in the last generation of (e, e p) experiments. We
are using the (e, e p) reaction at high momentum transfers
and high missing energies, a region that was simply not
accessible at the lower-energy, high duty-factor facilities
previously available. The missing strength was, indeed,
found (see g. 20), and agrees rougly with the predictions
of Correlated Basis Function theory (although the mo-
mentum distribution is not described correctly in detail).
In a third study correlated pairs have been measured
directly in the 3 He(e, e pp)n reaction. In this experiment,
the absorption of the virtual photon kicks out a proton,
and the opening angle of the remaining pair shows a back-
to-back peak. One can infer from the data the shape of the
pair momentum distribution.
Similar, though somewhat less direct, information can
be obtained from examining the 3 He(e, e p)X reaction at
very high missing momentum. Signicant strength above
what is predicted by PWIA has been observed (see g. 21).
The quantitative understanding of the results is work in
progress.
Fig. 21. Eective nucleon density for the 3 He(e, e p)X reaction
as a function of missing momentum. 2bbu stands for two-body
breakup [26].
4.2 The limits of our understanding of nite nuclei
One of the key issues that motivated the construction
of CEBAF was our desire to gain insight into the ques-
tion of where the description of nuclei based on nucleon
and meson degrees of freedom fails and the underlying
quark degrees-of-freedom must be taken into account.
Data on the elastic scattering from the deuteron and high-
energy photodisintegration, together with accurate theo-
retical calculations, are providing the answers.
We begin with the elastic scattering form factors for
the deuteron. The theory is in an advanced state: we
use the best ab initio calculation of the structure of the
deuteron with a potential VN N determined from a t to
N-N phase shifts, and then add exchange currents and rel-
ativistic corrections. The data set for the deuteron elastic
form factors demonstrate the technical accomplishments
of modern accelerators and equipment: elastic e-D scat-
tering has been measured down to cross sections charac-
teristic of -scattering!
The data for the electric and the magnetic form fac-
tors, and for the tensor polarization (see g. 22) demon-
strate that conventional nuclear theory works up to Q2 of
16 The European Physical Journal A

100 10-1
previous data previous data
Conventional
10-1 Hall A MMD, S2, D
10-2 Nuclear Theory
Hall C MMD, 0, D
MMD, S2, D Forest and Schiavilla, IA
10-2
MMD, 0, D 10-3 Hall A
Forest and Schiavilla, IA
10-3 Forest and Schiavilla, IA+pair
10-4
10-4
2
A(Q ) 10-5
10-5
10-6
B(Q2)
10-6

10-7
10-7

10-8 10-8

10-9 10-9
0 2 4 6 0.0 0.5 1.0 1.5 2.0 2.5
Q2 (GeV 2 ) Q2 (GeV 2 )

1.0
0.5
t20(Q2)
0.0
VEPP 1985
Bates 1984
-0.5 VEPP 1990
Bates 1991
NIKHEF 1995
NIKHEF 1996
-1.0 JLab/POLDER
MMD, S2, D
MMD, 0, D

-1.5 Forest and Schiavilla, IA

Forest and Schiavilla,
IA+pair Fig. 23. Cross sections for deuteron photodisintegration. The
0.0 0.5 1.0
2
1.5 2.0 energies associated with a transverse momentum of 1.37 GeV/c
Q (GeV 2 )
are indicated with a blue arrow in each panel [29].
Fig. 22. Electric and magnetic form factors of the deuteron
(top panels) and the deuteron tensor polarization (lower left)
together with the intrinsic shape of the deuteron inferred from 5 Summary
these data (lower right) (adapted from [27]).
The CEBAF accelerator at JLab is fullling its scientic
mission to understand how hadrons are constructed from
the quarks and gluons of QCD, to understand the QCD
about 2 (GeV/c)2 , i.e. the nucleon-based picture is still basis for the nucleon-nucleon force, and to explore the
valid at distance scales of about one half the size of a nu- transition from the nucleon-meson to a QCD description.
cleon. Why, we do not know; none of us expected it to Its success is based on the rm foundation of exper-
work that well before the experiments were undertaken. imental and theoretical techniques developed world-wide
The shape of the deuteron derived from the form fac- over the past few decades, on complementary data pro-
tor data is also shown in g. 22; one can see clearly that vided by essential lower-energy facilities, such as MAMI,
the nucleon spins are aligned end-to-end (resulting in a and on the many insights provided by the scientists we are
dumbell -shaped distribution) rather than anti-parallel gathered here to honor.
(which would have yielded a donut shape).
It is a pleasure to acknowledge the assistance of Bernhard
The photodisintegration of the deuteron was one of Mecking in the development of this article, and thoughtful com-
the rst experiments done in nuclear physics (at ener- ments on the manuscript from Volker Burkert, Kees de Jager,
gies of only a few MeV) and also one of the most re- John Domingo, and Rolf Ent. This work has been supported
cent ones (now at energies approaching 6 GeV). The reac- through The Southeastern Universities Research Association,
tion probes internal nucleon momenta well beyond those Inc., which operates the Thomas Jeerson National Accelera-
accessible in electron scattering because of the momen- tor Facility under Contract No. DE-AC05-84150 with the U.S.
tum mismatch between the photon and the nucleon. In Department of Energy.
a parton-based description of the reaction, one expects
the cross section to scale like s11 , where s is the CM
energy squared. The data (see g. 23) demonstrate that References
s11 scaling of the cross section is reached at photon en-
ergies which change with the proton center-of-mass angle. 1. C. Hyde-Wright, C.W. de Jager, Annu. Rev. Nucl. Part.
The transition occurs consistently at a transverse momen- Sci. 54, 217 (2004) and references therein.
tum of about (1.0 1.3) GeV/c, which shows that below 2. J. Friedrich, T. Walcher, Eur. Phys. J. A 17, 607 (2003).
0.2 fm the nucleon-meson description of the deuteron 3. J.J. Kelly, AIP Conf. Proc. 698, 393 (2004).
is no longer valid, and a parton-based description is more 4. T.M. Ito et al., Phys. Rev. Lett. 92, 102003 (2004).
appropriate. A more recent experiment [28] using CLAS 5. D.T. Spayde et al., Phys. Lett. B 583, 79 (2004).
has extended these data to include angular distributions 6. F.E. Maas et al., Phys. Rev. Lett. 94, 152001 (2005).
for a broad range of energies; the data is described by a 7. F.E. Maas et al., Phys. Rev. Lett. 95, 022002 (2004).
quark-gluon string model. 8. F. Maas, these proceedings.
 L.S. Cardman: Physics at the Thomas Jeerson National Accelerator Facility
9. K.A. Aniol et al., Phys. Rev. Lett. 96, 022003 (2006).
10. K.A. Aniol et al., Phys. Lett. B 635, 275 (2006).
11. K.A. Aniol et al., Phys. Lett. B 509, 211 (2001).
12. K.A. Aniol et al., Phys. Rev. C 69, 065501 (2004).
13. D.S. Armstrong et al., Phys. Rev. Lett. 95, 092001 (2005).
14. T. Horn, private communication for the F collaboration.
15. V.D. Burkert, Eur. Phys. J. A 17, 303 (2003).
16. L.C. Smith, Invited Talk, Japan-US Workshop on Electro-
magnetic Meson Production and Chiral Dynamics, Osaka,
Japan (April 2005).
17. CLAS Collaboration (V.D. Burkert), Int. J. Mod. Phys A
20, 1531 (2005).
18. I. Aznauryan talk at N*2005, Tallahasse, FL (October
2005), to be published.
19. X. Zheng et al., Phy. Rev. Lett. 92, 012004 (2004).
17
20. CLAS Collaboration (K.V. Dharmawardane et al.), sub-
mitted to Phys. Rev. Lett. (2006).
21. A. Thomas, these proceedings.
22. A. Deur, 13th Int. workshop on Deep Inelastic Scattering
(DIS2005), AIP Conf. Proc. 792, 969 (2005).
23. A. Deur, V. Burkert, J.P. Chen, W. Korsch, arXiv:hep-
ph/0509113.
24. K. Egiyan et al., Phys. Rev. Lett. 96, 082501 (2006).
25. D. Rohe et al., Phys. Rev. Lett. 93, 182501 (2004).
26. F. Benmokhtar et al., Phys. Rev. Lett. 94, 082305 (2005).
27. M. Garcon, J.W. Van Orden, Adv. Nucl. Phys. 26, 293
(2001) and references therein.
28. M. Mirazita et al., Phys. Rev. C 70, 014005 (2004).
29. E.L. Schulte et al., Phys. Rev. Lett. 87, 102302 (2001).
Eur. Phys. J. A 28, s01, 19 27 (2006)
DOI: 10.1140/epja/i2006-09-003-8 EPJ A direct
electronic only

Few-nucleon systems at MAMI and beyond

W.U. Boeglina
Physics Department, Florida International University, University Park, Miami, FL 33199, USA

/
Published online: 10 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. Few-body systems provide a testing ground for models of the NN interaction, reaction mecha-
nisms and for models of nuclei. An overview of results of coincidence experiments on the deuteron, 3 He
and 4 He obtained in the last 20 years at MAMI and at other facilities, covering a wide range of momentum
and energy transfers, is presented.

PACS. 25.10.+s Nuclear reactions involving few nucleon systems 25.30.-c Lepton-induced reactions
25.30.Fj Inelastic electron scattering to continuum

1 Introduction PWIA FSI

q H
pf q H
Few-body systems are ideal to investigate fundamental  p
problems in nuclear physics such as the ground state and  f
continuum wave functions, the importance of correlations H H
and the structure of the electromagnetic current operator. pi pi H H
H H pm pi
In addition, interaction eects such as meson exchange
currents (MEC), and isobar congurations (IC) can be pm = pi
studied. At large momentum transfers, one hopes to be
able to explore the transition from the regime where ob-
servables are best described by nucleon/meson degrees of
freedom to the regime where quark/gluon degrees of free- MEC
dom are the most e cient, in describing the interaction. H IC
p q
The few-body systems presented here consist of the q  f H
p
deuteron and the 3 He and 4 He nuclei. They range from  f
a loosely bound system, such as the deuteron, to a very H
pi H H H
tightly bound one such as 4 He. In contrast to complex pm pi pi H H
nuclei, the structure of few-body systems can nowadays pm pi
be calculated with high precision using realistic nucleon-
nucleon interaction potentials. Very successful methods Fig. 1. Various reaction mechanisms contributing to the
for the calculation of bound and continuum state wave (e, e N) reaction. Plane-wave impulse approximation (PWIA),
functions include the solution of Faddeev-Yakubovsky [1] distorted-wave impulse approximation (FSI), meson exchange
equations, Variational Monte Carlo [2], and other Monte- currents (MEC) and isobar congurations (IC).
Carlo based calculations [3,4].
Many laboratories have contributed to the study of
few-body systems in the last twenty years (in alphabet- explore new, previously inaccessible kinematical regions
ical order): ALS (Saclay, France), ELSA (Bonn, Ger- with very high statistical precision. I will therefore focus
many), Jeerson Lab or JLAB (Newport News, VA, on the study of few-body systems using knock-out reac-
USA), MAMI (Mainz, Germany), MIT-Bates (Middle- tions such as (e, e N) and (e, e NN).
ton, USA), NIKHEF (Amsterdam, The Netherlands), and
SLAC (Stanford, CA, USA). Recent electron accelerators
such as the Mainz Microtron and Jeerson Lab (CEBAF) 2 Short overview of the (e, eN) reaction
provide very high-intensity, continuous wave (CW) beams.
These have made coincidence experiments possible that Treating the incoming and scattered electrons as plane
waves, and applying the one-photon exchange approxima-
a
e-mail: boeglinw@fiu.edu tion, the (e, e p) reaction can be viewed schematically as
20 The European Physical Journal A

pf 10
-4

-5
e 10
exp

e CC Calculation with CD Bonn

pq 10
-6

d /ddedp [fm /MeV sr ]

2
q 10
-7

2
e 10
-8

pm 10
-9

5
-10
10
Fig. 2. Geometry and kinematical variables for the (e, e N)
reaction. 10
-11

shown in g. 1. In the case of the Plane-Wave Impulse Ap-
proximation (PWIA) the virtual photon is absorbed by a
bound nucleon having a certain initial momentum pi . The
struck proton subsequently leaves the nucleus with a nal
momentum pf . The residual system may remain in its
ground or in an excited state and has a recoil momentum
pm . In the following, the term missing momentum will be
used synonymously with recoil momentum. Within PWIA
the following relation between initial and missing momen-
tum is valid: pm = pi . The transferred energy is di-
vided between the kinetic energy of the ejected nucleon, its
separation energy, and the kinetic and, possibly, excitation
energy of the residual system. The missing momentum pm
and missing energy Em are dened as follows:
Momentum conservation : q = pf + pm ,
Energy conservation : E m = T p Tr .
-12
10 0 200 400 600 800 1000
pm [MeV/c]
Fig. 3. D(e, e p)n cross section measured at MAMI [11] com-
pared to calculations by H. Arenhovel [13].
tion of the (e, e N) cross section into an elementary (o-
shell) electron nucleon cross section [8] and the spectral
function describing the probability of nding a nucleon
with a given initial momentum and missing energy. In-
tegrating the spectral function over the missing energy
leads to the momentum distribution. Final state interac-
tions (FSI), MEC, and IC remove this simple relation and
therefore pi = pm (g. 1).

Here Tp is the kinetic energy of the ejected nucleon, and
Tr is the kinetic energy of the recoiling system, calculated
from pm under the assumption that the undetected (A-
1)-system remains in its ground state.
Figure 2 shows the electron scattering plane, dened
by the incoming and scattered electron momenta, and the
reaction plane, dened by the nal nucleon momentum
and the momentum transfer. The cross section in the one
photon exchange limit can be written as [5,6,7]
d5
= M ott (vL RL + vT RT +
dde dp
+ vLT RLT cos + vT T RT T cos 2),
where Ri are the response functions containing matrix ele-
ments of the charge and current operators. These, in turn,
provide the nuclear structure information. The vi are kine-
matical factors depending on the electron kinematics only,
and M ott is the Mott cross section describing the scat-
tering of relativistic electrons by a point charge.
If one neglects the interaction of the outgoing nucleon
with the recoiling system, one obtains the plane-wave im-
pulse approximation (PWIA) which permits a factoriza-
3 Studies of the deuteron
Early (e, e p) experiments were limited in luminosity by
the duty factor of the available electron accelerators. Cross
sections could be measured for large missing momenta
(pm 0.5 (GeV/c)) only at relatively small momentum
transfer (Q2 0.1 (GeV/c)2 ) or for large Q2 only at rel-
atively small (pm < 0.2 (GeV/c)).
Experiments have been carried out at all facilities men-
tioned above. More recent experiments, carried out in the
last ten years, beneted from the availability of high duty
cycle beams at Jeerson Lab, MAMI, NIKHEF(AmPS),
and at MIT-Bates(SHR). In general, the various exper-
iments can be separated into those that explored the
D(e, e p)n cross section over a large range of missing mo-
menta and those that extracted individual response func-
tions.
3.1 Cross section measurements at low Q2
These experiments explored the D(e, e p)n cross section
over a wide range of missing momenta at small to medium
 W.U. Boeglin: Few-nucleon systems at MAMI and beyond 21

120 8
Saclay 7
110 Bates NIKHEF

f00exp/Aren (%)
6 Bates
100 5

f00(fm)
1.2 4
90 3
2
80
exp/NMIR

1
1 70 0
-100 -50 0 50 100 20 30 40 50 60 70
120 8
Saclay 7
110 Bates NIKHEF

f11exp/Aren (%)
0.8 6 Bates
100 5

f11(fm)
4
90 3
0.6 2
80
1
70 0
0 50 100 150 200 250 300 -100 -50 0 50 100 20 30 40 50 60 70
pmiss (MeV/c) pm (MeV/c) pm (MeV/c)

Fig. 4. Ratio of experiment to theory for low missing mo-
menta. Data points: blue diamonds [11], red squares [14], vi-
olet circles [12], turquoise triangles down [9], green triangles
left [15], orange triangles up [16].
Fig. 5. Comparison of RL (f00 ),RT (f11 ) measurements to
one another and theory. Left: Ratio (in %) of the experimental
response functions from Saclay [14] and Bates [15] to the cal-
culation by H. Arenhovel [19], [20]. Right: Response functions
from NIKHEF [17, 18] and Bates [15].

momentum transfers [9,10,11,12]. The focus of these mea-

surements was the exploration of the momentum distribu-
tion within the plane-wave impulse approximation. It has
been found, however, that with increasing recoil momen-
tum FSI and, related to the corresponding energy transfer,
MEC and IC contributions increase dramatically. Figure 3
shows the D(e, e p)n cross section measured at MAMI [11]
and H. Arenhovels calculation that includes FSI, MEC,
and IC [13]. One can see that the cross section is well re-
produced up to pm = 350 MeV/c. At higher pm there are
signicant discrepancies between experiment and theory.
This occurs in a kinematical region where large virtual
delta excitation contributions are expected.
Since many experiments have measured the D(e, e p)n
cross section at missing momenta below 300 MeV/c it is in-
teresting to compare how well these results agree with each
other in order to learn how accurately the D(e, e p)n reac-
tion is known experimentally. As the various experiments
have been carried out at dierent kinematical settings, I
used Arenhovels calculation [13] as a reference to take into
account FSI. MEC and IC are also included, however they
tend to contribute less than FSI. The result is shown in
g. 4 where the ratio between the experimental and the
theoretical cross sections is shown. It is interesting to note
that the various experiments agree quite well among each
other, while the experimental cross sections seem to be
systematically smaller than the calculated cross sections
by about 10%. The reason for this discrepancy needs fur-
ther investigation but one has to keep in mind that most
experiments quote systematic errors around 5%.
3.2 Structure function separations
Experiments to extract various response functions of the
D(e, e p)n reaction have been carried out at most elec-
tron accelerators. The major results of these experiments
will be presented below. In general, all published re-
sponse function separations have been limited to miss-
ing momenta below 200 MeV/c. RL and RT have been
determined at NIKHEF [17,18], Saclay [14], and at MIT-
Bates [15].
Figure 5 shows a comparison of the various results
for overlapping kinematics. The left panel shows the ra-
tio to H. Arenhovels calculation [19,20]. The longitudi-
nal response has been found to deviate up to 20% from
the calculation depending on the missing momentum. The
transverse response, depending on the missing momen-
tum, deviates up to 10%. For missing momenta below
50 MeV/c the longitudinal response has been found to be
about 20% smaller than the calculation in both, the Saclay
and the Bates experiments. The transverse response has
been found to be about 4% smaller than the calculation
for the Saclay data and in agreement (within the error
bars) for the Bates experiment. In contrast the longitudi-
nal response found at NIKHEF is much larger than the
MIT-Bates result, while the transverse responses are in
agreement.
At MAMI RL and RT have been extracted for missing
momenta up to 350 MeV/c [21]. An example of the result
of this experiment is shown in g. 6.
Unfortunately at small recoil momenta problems with
the target lead to uncertainties in determining the abso-
lute cross section making it impossible to compare the
results to the low pm results discussed above.
To summarize, the experimental knowledge of RL and
RT is limited to low recoil momenta and there are dis-
agreements among dierent experiments and also when
compared to modern calculations. Currently there exists
no experimental program to address these problems.
The response function RLT is sensitive to relativistic
contributions to the electromagnetic current operator [22]
and FSI and has the advantage that it is particularly
easy to extract. This is due to the fact that the elec-
tron kinematics are xed and only the proton scattering
22 The European Physical Journal A

PWBA
N and has the additional advantage that the absolute cross
N + MEC + IC
10 N + MEC + IC + REL section normalization cancels in the ratio. An overview of
Data
experimental results is shown in g. 7 together with the
1 result of a determination of ALT at MAMI [23]. Other
recent ALT measurements have been published in refer-
RL / fm

ences [16] and [24].

0.1
A determination of RT T requires proton detection out
of the electron scattering plane. This has been achieved
0.01
at MIT-Bates using the Out-Of-Plane (OOPS) spectrom-
eter [24] system and at NIKHEF [28] using the HADRON
0.001
0 50 100 150 200 250 300 350 400
detectors. For an overview of results see [25].

pm / (MeV/c) An additional response function, RLT , can be obtained
PWBA
using out-of-plane detection of the proton and measuring
N + MEC + IC
N the helicity dependence of the cross section with polarized
10 N + MEC + IC + REL
Data electrons. This has been carried out at MIT-Bates [29]
using OOPS.
1
RT / fm

0.1 3.3 Cross section measurements at high Q2

0.01 D(e, e p)n cross sections have been obtained at SLAC for
high Q2 but low recoil momenta (pm < 0.2 GeV/c) [30].
0.001 Recently, experiments have been carried out at Jeerson
0 50 100 150 200 250 300 350 400
pm / (MeV/c) Lab in Hall A (experiment E01-020) as well as in Hall B
using CLAS (experiment E94-019). The goal of the Hall
Fig. 6. Result of a RL -RT separation at MAMI [21] for miss-
A experiment is to test the Generalized Eikonal Approx-
ing momenta up to 350 MeV/c at a momentum transfer of
imation (GEA) description of FSI [31] in the D(e, e p)n
450 MeV/c. Calculations are by H. Arenhovel [20] (PWBA: no
FSI but scattering o the neutron and observing the recoiling
reaction while the goal of the Hall B experiment is to
proton has been included). use the GEA description of the D(e, e p)n reaction in
the search for evidence of color transparency. Within the
GEA, FSI are described by a series of small-angle scatter-
0.4

exp
ings of the outgoing nucleon. This approximation, which
0.2 N+MEC+IC+REL
PWIA is typically valid for nucleon energies of 1 GeV and above,
0 has been successfully applied in high-energy nucleon scat-
tering. However it has never been tested for the D(e, e p)n
ALT

-0.2
A

-0.4
-0.6
-0.8
0 100
Pmiss (MeV/c)
Fig. 7. Left: an overview of measurements of ALT from var-
ious experiments [25] where 0.15 Q2 0.22 (GeV/c)2 .
Calculations are from H. Arenhovel et al. [26] (dash-dot:
N+MEC+IC, solid: N+MEC+IC+RC) and from E. Hummel
et al. [27] (dashed: PWBA), the dotted curve corresponds to
PWIA (cc1 ). 0.15 to 0.22. Right: Determination of ALT at
MAMI for Q2 = 0.33 (GeV/c)2 [23]. The calculation is by
H. Arenhovel [20] including FSI, MEC, IC and Relativistic
corrections (RC).
angle is changed (in the electron scattering plane) in such
a way, that the reaction plane varies between = 0 and
= 180 (see g. 2). The cross section dierence obtained
from these two measurements is then proportional to RLT .
A quantity closely related to RLT is the left-right asym-
metry
0 tion. Clearly the predicted angular dependence has been
ALT = 180
180 + 0
reaction. Another goal of the Hall A experiment is the
determination RLT for missing momenta up to 0.5 GeV/c
where relativistic eects are expected to be very large and
200
300 RLT is sensitive to details of the current operator.
The GEA predicts a characteristic dependence of the
strength of FSI on the angle of the recoiling neutron with
respect to the momentum transfer and on the value of the
missing momentum. For angles around 80 FSI eects are
predicted to be maximal. For pm = 0.2 GeV/c a reduction
of the cross section by about 30 40% is predicted and for
pm = 0.4 GeV/c and pm = 0.5 GeV/c an increase of the
cross section by more than a factor of two is predicted. The
location of the extremum of the rescattering contributions
give additional information about the details of the rescat-
tering process such as the importance of the Fermi motion
of the bound nucleons.
In order to address these questions in Hall A, the
D(e, e p)n cross section has been measured for Q2 =
0.8, 2.1 and 3.5 (GeV/c)2 and missing momenta between
pm = 0 and pm = 0.5 GeV/c. A very preliminary result is
shown in g. 8 where the observed yield is compared to a
Monte Carlo calculation using the PWIA for the cross sec-
observed. A detailed analysis is currently in progress.
 W.U. Boeglin: Few-nucleon systems at MAMI and beyond 23

5
FSI / PWIA

4.5
4 pm=0.5 GeV/c exp.

3.5
3 pm=0.5 GeV/c + MEC
pm= 0.5 GeV/c no MEC
2.5
pm= 0.2 GeV/c exp.
2
1.5
1
0.5 pm= 0.2 GeV/c no MEC
0
0 20 40 60 80 100 120 140 160 180
nq

Fig. 8. Preliminary ratio of the measured yield to the cal-

culated one using PWIA for the D(e, e p)n reaction at Q2 =
3.5 (GeV/c)2 . Data for pm = 0.5 (GeV/c) show a strong en-
hancement of the cross section at about 70 while the data for Fig. 9. The missing energy spectrum for the 3 He(e, e p) re-
pm = 0.2 (GeV/c) show a reduction at a similar angle. The action on the quasi-free peak. The 2-body breakup peak is
bars indicate the uncertainty from the preliminary status of clearly separated from the 3-body continuum. All strength
the analysis. The nal errors will be of the order of 10%. The above about 25 MeV is entirely due to the radiative tail [38].
calculation by J. M. Laget [32] reproduces the trend of the data
quite well down to angles of about 40 . The behavior of the
data at angles below 40 is unexpected but given the status if
the analysis no further conclusions can be drawn at this time.

4 3 He and 4 He studies
The breakup of the 3,4 He nuclei can lead to a 2-body -
nal state like in the deuteron or it can lead to a 3- and
even 4-body nal state for 3 He and 4 He, respectively. Only
recently, with the advent of high computing power and ef-
cient computational techniques, can the continuum nal
state be calculated accurately. These nuclei are the sim- Fig. 10. The asymmetry ALT and the response function RLT
plest systems in which to study short range correlations. measured at MAMI for the 4 He(e, e p)3 H reaction [39] together
with calculations with and without the inclusion of MEC [40].

4.1 Low Q2 experiments

 3
Similar to deuterium, early (e, e p) experiments on He
and 4 He explored the cross section with the goal to ob-
tain information on the momentum distribution [33,34].
The same problems as described in section 3.1 are en-
countered here. In addition to examining the momentum
distribution one has also studied the missing-energy spec-
trum. Early experiments on 3 He by Marchand et al. [33]
showed a structure in the missing energy spectrum that
is shifting with increasing recoil momentum in agreement
with the kinematics of scattering o a nucleon pair. After
the absorption of the virtual photon by one member of the
pair, the struck nucleon is observed and the other partner
of the pair recoils with the negative initial momentum of
the struck nucleon.
To investigate the nucleon knock-out reaction in de-
tail on these nuclei, a high-precision measurement of the
(e, e p) cross section for 3,4 He has been carried out at
MAMI. The goal was to determine the longitudinal and
the transverse response in parallel kinematics close to the
quasi-free peak and for missing energies up to 70 MeV/c.
Results of these measurements can be found in refer-
ences [35,36]. As an example, g. 9 shows the missing
energy spectrum obtained on the quasi-free peak. No ad-
ditional strength can be observed above a missing energy
of about 25 MeV and the dependence of the cross section
on the polarization of the virtual photon is the same as
the one within PWIA. Hence, besides an overall reduction
of the experimental cross section that is most likely due
to FSI, no additional eects have been observed. By con-
trast, an RL /RT separation carried out at Saclay [37] for
missing momenta above 250 MeV/c found very large de-
viations of the data from the calculations for the 2-body
breakup, even when FSI and MEC are included.
Cross sections have also been obtained for the 3-
body breakup region. Again no additional dependence on
the virtual photon polarization has been found beyond
PWIA [36].
For the 2-body breakup of 4 He the interference re-
sponse function, RLT has been extracted at MAMI [39]
for missing momenta from 150 MeV/c up to 300 MeV/c
at a momentum transfer Q2 0.33 (GeV/c)2 . The results
24 The European Physical Journal A

-6 0.75
10
PWIA
-7 0.50 GLB
10 PWIA
GLB+MEC
GLB
-8 GLB+MEC 0.25
10

(fm /MeV/sr )
2
-9
10 0.00

ALT
10
-10 -0.25

2
10
-11 -0.50

10
-12
-0.75

-13
10 -1.00

-14
10 -1.25
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
pm(MeV/c) pm(MeV/c)

Fig. 12. As in g. 9 but calculations by Schiavilla et al. [43].

MEC contribute very little to the cross section while a sizeable
contribution to ALT for 300 pm is observed. At missing
momenta above 800 (MeV/c) double rescattering dominates
the cross section.
Fig. 11. Left: The cross section for the 2-body breakup of 3 He
for Q2 = 1.55 (GeV/c)2 [42] (pm < 0 corresponds to = 0
and pm > 0 corresponds to = 180 ) Right: the extracted 10-2
ALT ratio. The calculations are Glauber-based by J.M. Laget
(see [42]).
10-3

d d p [ b/MeVsr2]
have been compared to calculations by R. Schiavilla et 10-4
al. [40,41], which show the need for MEC to improve the
agreement with the experiment (g. 10).
10-5

4.2 High-Q2 experiments

10-6
A detailed study of the 3,4 He electrodisintegration in
d 5 /d

quasi-free kinematics has been performed in Hall A at 10-7

Jeerson Lab. The (e, e p) cross section has been mea-
sured in parallel kinematics to allow for a RL /RT separa-
tion and in perpendicular kinematics in order to determine
the ALT asymmetry. The energy and 3-momentum trans-
fers have been kept constant at = 0.84 (GeV) and at
q = 1.502 (GeV/c) (Q2 = 1.55 (GeV/c)2 ) and cross sec-
tions have been measured up to pm = 1 GeV/c. At these
large momentum and energy transfers Glauber based cal-
culations are expected to be valid.
In g. 11 the experimental cross sections for the 2-body
breakup are shown. Negative values of pm correspond to
cross sections measured at = 0 while positive pm -values
are cross sections measured at = 180 . Full Glauber-
based calculations with modern 3-body wave functions
from realistic potentials provide an excellent descrip-
tion of the experimental data up to missing momenta of
150 MeV/c. The same calculations give a good description
of the observed cross sections up to missing momenta of
750 MeV/c. Some deviations between experiment and the
calculation can be observed for = 0 between 250 and
500 MeV/c. It is also evident that FSI play a major role
for recoil momenta above 300 MeV/c. In Lagets calcula-
tion MEC and IC are found to contribute at most 25%
which is in agreement with the expectation that these con-
tributions diminish with increasing momentum transfer.
Another recent Glauber-based calculation by R. Schi-
avilla et al. [43], where the full spin and isospin dependence
of the underlying NN amplitudes is retained, reproduces
the experimental 2-body breakup cross section very well
(g. 12). Rescattering eects have been found to be im-
portant over the full range of recoil momenta studied and,
10-8
10-9
0 100 200 300 400 500 600 700
pm [MeV/c]
Fig. 13. The cross section for the 2-body breakup of 3 He for
Q2 = 1.55 (GeV/c)2 [42] compared to a calculation by Kap-
tari et al. [44].
especially, double rescattering is important to improve the
agreement between experiment and theory for recoil mo-
menta above 750 MeV/c.
A third calculation by Kaptari et al. [44], based on
GEA, can also reproduce the measured cross sections up to
pm = 700 MeV/c reasonably well. In this calculation MEC
and IC contributions have not been included (g. 13). This
again supports the expectation that with increasing mo-
mentum transfer, MEC and IC contributions to the cross
section decrease.
In addition to 2-body breakup, the 3-body breakup
reaction has also been observed at missing energies up to
a 140 MeV [45]. As in the Saclay experiment at low mo-
mentum transfer, a broad bump has been observed whose
location shifts to increasing missing energy with increas-
ing missing momentum (g. 14). The peak location is kine-
matically in agreement with the breakup of a nucleon pair.
The observed nucleon momentum is the sum of the mo-
mentum transfer and the initial momentum in the pair and
 W.U. Boeglin: Few-nucleon systems at MAMI and beyond
Fig. 14. Missing energy spectra for the 3-body beakup of 3 He
in Hall A at JLAB [45], the kinematics is the same as in ref-
erence [42]. The arrows indicate the expected location of the
peak for the breakup of a nucleon pair. The width is a conse-
quence of the center-of-mass motion of the pair. The solid line
includes FSI and MEC. The calculations are by J.M. Laget
(see [45]).
the recoiling nucleon is the partner. The third nucleon in
helium is a spectator.
Calculations by J. M. Laget [46] show that FSI within
the active nucleon pair contribute strongly to the observed
structure while nal state interactions between the pair
nucleons and the spectator nucleon seem to be much less
important. Integrating over missing energy and dividing
by the electron-nucleon cross section [8] results in the
(PWIA) eective momentum distribution for the 3-body
breakup. This can then be compared to the eective mo-
mentum distribution from the 2-body breakup channel.
The result is shown in g. 15. With increasing recoil mo-
menta the momentum distribution is dominated by the
3-body breakup process. However at this point one can-
not experimentally distinguish contributions due to FSI
within the nucleon pair from contributions of shortrange
correlations. Maybe separated response functions will give
more information to address this question.
4.3 Search for short-range correlations in 3 He
The detection of nucleon pairs should allow one to
study nucleon correlations in a direct way. These triple-
25
Fig. 15. Proton eective momentum distributions in 3 He ex-
tracted from the 3-body breakup reaction (open circles) com-
pared to the 2-body contribution (triangles) [45]. The integra-
tion in missing energy ranges from threshold up to 140 MeV.
Calculations are by J. M. Laget [46, 32].
coincidence experiments can only be carried out with CW
electron beams. 3 He(e, e NN) experiments have been car-
ried out at NIKHEF with the pulse stretcher ring AmPS,
at MAMI and at Jeerson Lab. Again, one of the fun-
damental problems is to isolate the various reaction pro-
cesses that can lead to the emission of nucleon pairs in
addition to initial state correlations. The most prominent
processes are nal state interactions and 2-body currents
such as MEC and IC. In an experiment carried out at
NIKHEF [47,48], 3 He(e, e pp)n cross sections were mea-
sured for momentum transfers around 400 MeV/c and at
an energy transfer of about 220 MeV. The experimental
results were compared to continuum Faddeev calculations
performed with various modern potentials including one-
and two-body currents. The relative momenta of pair nu-
cleons was between 500 and 800 MeV/c. It was found that
calculations performed with only a one-body hadronic cur-
rent operator show a fair agreement with the data for small
missing (neutron) momenta. This can be interpreted as a
direct knock-out of a proton pair. With increasing miss-
ing momentum, contributions from MEC and IC also in-
crease. FSI between the pair nucleons depend on the rel-
ative emission angle. Decreasing relative emission angle
leads to increasing FSI.
At MAMI 3 He(e, e pn)p cross sections have been mea-
sured and are currently being analyzed.
Another study of the 3 He(e, e pp)n reaction has been
carried out with CLAS at Jeerson Lab [49]. Via a Daliz-
plot of the kinetic energies of the observed nucleons di-
vided by the energy transfer, various reaction mechanisms
could be selected. The location where both nucleons have
the smallest energy fraction corresponds to the process
where the virtual photon is absorbed by the uncorrelated
nucleon. The observed (small energy fraction) nucleons are
then the pair nucleons. This interpretation is supported by
26 The European Physical Journal A

1
Tp1 (a) (b) Data
150 Cut Data
Phase Space

Counts
100
0.5

50

0 0
0 0.5 1 -1 -0.5 0 0.5 1
Tp2
cos( )
pp

Fig. 16. a) Dalitz plot for the lab frame Tp1 / versus Tp2 /
for events with pN > 0.25 GeV/c [49]. b) The cosine of the
p-p lab frame opening angle. Open circles for events with small Fig. 18. Polarization transfer experiments on 4 He from
proton momenta (Tp < 0.2 ) and closed circles show all data. Mainz [55] and JLAB [56]. The best agreement is obtained for a
modied nucleon form factor within the quark-meson coupling
e model [57, 58].
P y
e pf
P x 1.05

q P z
1.00

e
(P x /P z )/(Px /Pz )PWIA

0.95
Fig. 17. Denition of polarization variables for (e, e N) polar-
ization transfer experiments. e represents the polarized elec-
tron beam, e the scattered electron and e the electron scatter-
ing angle. q the virtual photon, pf the nal nucleon momentum 0.90
and Px , Py , Pz the polarization of the ejected nucleon.
4
He EXP
1
0.85 H EXP
OPT(no CH EX)
the relative angular distribution of the observed nucleon OPT
pair that shows a pronounced peak at 180 (in the Lab OPT+MEC

frame) corresponding to nucleon emission back-to-back 0.80

(g. 16). As before [47,48] one nds strong FSI within
the nucleon pair and only small contributions due to two-
body currents and rescattering of the struck nucleon with
the pair.
5 Polarization transfer experiments
Spin degrees of freedom open up a new, large set of observ-
ables that make it possible to address a variety of dier-
ent questions in nuclear physics. Spin observables lead to
interference terms between dierent reaction amplitudes.
This in turn makes it possible to study very small am-
plitudes when their eect is enhanced by a large one. An
important application of this is the determination of the
nucleon form factors where polarized electrons are scat-
tered o an unpolarized target and the polarization of the
struck nucleon is determined in a polarimeter [50].
For the (
e , e N ) reaction on a free nucleon one obtains
(see g. 17)
GE P  (Ei + Ef )
= x tan e /2 ,
GM Pz 2M
0.0 0.5 1.0 1.5 2.0 2.5 3.0
2 2
Q [(GeV/c) ]
Fig. 19. Comparison of the 4 He polarization transfer exper-
iment [56] and the calculation by Schiavilla et al. [59]. No
nucleon form factor modications have been included. OPT:
Only one-body currents included and p3 H FSI described by
an optical potential with and without (no CH-EX) charge ex-
change. OPT+MEC: Full optical potential including one- and
two-body currents. The error bars in the calculation are due
to the Monte Carlo method used and are similar for all calcu-
lations shown.
where GE and GM are the electric and magnetic Sachs
form factors, Px and Pz are the nucleon polarization, Ei
and Ef the incident and scattered (or nal) electron en-
ergy, M the nucleon mass and e the electron scattering
angle.
This method has been used extensively to determine
the neutron form factor and the high-Q2 behavior of the
ratio of the electric to the magnetic form factor of the
proton [51,52,53,54] (see contributions by D. Rohe and
M. Ostrick). The same process can also be measured in
nuclei. Experiments carried out at MAMI [55] and at
 W.U. Boeglin: Few-nucleon systems at MAMI and beyond
JLAB [56] on 4 He showed that the ratio
(Px /Pz )He
R=
(Px /Pz )H
is reduced by about 10% (g. 18). Some calculations based
on the quark meson coupling model [57,58] suggest a pos-
sible modication of the nucleon form factors inside nuclei
to account for this observation. Another calculation by
Schiavilla et al. [59] uses realistic wave functions for the
bound state that include correlation eects. One- and two-
body currents are included and spin and isospin depen-
dences in the nal state interaction including charge ex-
change have been taken into account. The calculation can
reproduce the experimental data without the need of form
factor modications as can be seen in g. 19. New high-
precision experiments on 4 He are planned at Jeerson Lab
in Hall A to improve and extend the available data.
6 Summary and conclusion
In the last 20 years much progress has been made in the
knowledge of the structure of few-body systems. The avail-
ability of high quality CW beams made it possible to mea-
sure coincidence cross sections over a wide range of kine-
matical variables which were inaccessible before. In paral-
lel, theoretical progress together with increasing compu-
tational power has resulted in sophisticated models that
agree very well with the data.
Coincidence data on 3 He have enjoyed a lot of atten-
tion and the analysis of recent data taken at Jeerson
Lab is still in progress. The importance of a detailed un-
derstanding of nal state interactions, MEC, and IC is
crucial in order to extract information on the short-range
structure of light nuclei.
While many new single-arm data on the deuteron have
been obtained, available coincidence data and especially
the lack of new, high-precision response function determi-
nations are disappointing.
Triple coincidence experiments on the He nuclei are
expected to provide new data on the structure of corre-
lations. However, these experiments are very complex to
carry out, analyze, and interpret. Several experiments are
being currently analyzed at MAMI and JLAB. The (up-
graded ) Mainz Microtron will continue to play a leading
role in nuclear physics.
I would like to thank H. Arenhovel, H. Backe, D. Drechsel,
J. Friedrich, K-H. Kaiser and Th. Walcher for making MAMI
such a success, for their contributions to our eld and for giving
me the opportunity to carry out research at MAMI. I wish you
all the best in the future. This work was supported in part by
the Department of Energy, DOE grant DE-FG02-99ER41065.
References
1. A. Nogga, H. Kamada, W. Glockle, Nucl. Phys. A 689,
357 (2001).
27
2. R. Schiavilla, V.R. Pandharipande, R.B. Wiringa, Nucl.
Phys. A 449, 219 (1986).
3. J. Carlson, Phys. Rev. C 36, 2026 (1987).
4. B.S. Pudliner et al., Phys. Rev. Lett. 74, 4396 (1995).
5. A.E.L. Dieperink, T. de Forest, Annu. Rev. Nucl. Sci. 25,
1 (1975).
6. S. Frullani, J. Mougey, Adv. Nucl. Phys. 14, 1 (1984).
7. S. Bo , C. Giusti, F.D. Pacati, Phys. Rep. 226, 1 (1993).
8. T. de Forest, Nucl. Phys. A 392, 232 (1983).
9. M. Bernheim et al., Nucl. Phys. A 365, 349 (1981).
10. S. Turck-Chieze et al., Phys. Lett. B 142, 145 (1984).
11. K.I. Blomqvist et al., Phys. Lett. B 429, 33 (1998).
12. P.E. Ulmer et al., Phys. Rev. Lett. 89, 062301 1 (2002).
13. H. Arenhovel, W. Leidemann, L. Tomusiak, Phys. Rev. C
52, 1232 (1995).
14. J.E. Ducret et al., Phys. Rev. C 49, 1783 (1994).
15. D. Jordan et al., Phys. Rev. Lett. 76, 1579 (1996).
16. Kasdorp et al., Few-Body Syst. 25, 115 (1997).
17. M. van der Schaar et al., Phys. Rev. Lett. 66, 2855 (1991).
18. M. van der Schaar et al., Phys. Rev. Lett. 68, 776 (1992).
19. W. Fabian, H. Arenhovel, Nucl. Phys. A 314, 253 (1979).
20. H. Arenhovel, private communication (2001).
21. R. Boehm, Thesis, University of Mainz (2001).
22. S. Jeschonnek, J.W. Van Orden, Phys. Rev. C 62, 044613
(2000).
23. W.U. Boeglin, private communication (2005).
24. Z.-L. Zhou et al., Phys. Rev. Lett. 87, 172301 (2001).
25. Z.-L. Zhou et al., Proceedings of the MIT-Bates Workshop
(1998).
26. F. Ritz, H. Goller, Th. Wilbois, H. Arenhovel, Phys. Rev.
C 52, 1232 (1995).
27. E. Hummel, J.A. Tjon, Phys. Rev. C 49, 21 (1994).
28. A. Pellegrino et al., Phys. Rev. Lett. 78, 4011 (1997).
29. S.M. Dolni et al., Phys. Rev. C 60, 064622 (1999).
30. H.J. Bulten et al., Phys. Rev. Lett. 74, 4775 (1995).
31. L.L. Frankfurt, M.M. Sargsian, M.I. Strikman, Phys. Rev.
C 56, 1124 (1997).
32. J.M. Laget, Phys. Lett. B 609, 49 (2005).
33. C. Marchand et al., Phys. Rev. Lett. 60, 1703 (1988).
34. E. Jans et al., Nucl. Phys. A 475, 687 (1987).
35. R.E.J. Florizone et al., Phys. Rev. Lett. 83, 2308 (1999).
36. A. Kozlov et al., Phys. Rev. Lett. 93, 132301 (2004).
37. J.M. Le Go et al., Phys. Rev. C 55, 1600 (1997).
38. R.E.J. Florizone, Thesis, MIT (1999).
39. K. Aniol et al., Eur. Phys. J. A 22, 449 (2004).
40. R. Schiavilla et al., Phys. Rev. Lett. 54, 835 (1990).
41. J. Forest et al., Phys. Rev. C 54, 646 (1996).
42. M. Rvachev et al., Phys. Rev. Lett. 94, 192302 (2005).
43. R. Schiavilla et al., Phys. Rev. C 72, 064003 (2005).
44. L.P. Kaptari, C. Cio degli Atti, Phys. Rev. C 71, 024005
(2005).
45. F. Benmokhtar et al., Phys. Rev. Lett. 94, 082305 (2005).
46. J.M. Laget, Few-Body Syst., Suppl. 15, 171 (2003).
47. D.L. Groep et al., Phys. Rev. Lett. 83, 5443 (1999).
48. D.L. Groep et al., Phys. Rev. C 63, 014005 (2000).
49. R.A. Niyazov et al., Phys. Rev. Lett. 92, 52303 (2004).
50. R. Arnold et al., Phys. Rev. C 23, 363 (1981).
51. M.K. Jones et al., Phys. Rev. Lett. 84, 1389 (2000).
52. O. Gayou et al., Phys. Rev. C 64, 038202 (2001).
53. O. Gayou et al., Phys. Rev. Lett. 88, 092301 (2002).
54. T. Pospischil et al., Eur. Phys. J. A 12, 125 (2001).
55. S. Dietrich et al., Phys. Lett. B 500, 47 (2001).
56. S. Strauch et al., Phys. Rev. Lett. 91, 052301 (2003).
57. D.H. Lu et al., Phys. Lett. B 417, 217 (1998).
58. D.H. Lu et al., Phys. Rev. C 60, 068201 (1999).
59. R. Schiavilla et al., Phys. Rev. Lett. 94, 072303 (2005).
Eur. Phys. J. A 28, s01, 2938 (2006)
DOI: 10.1140/epja/i2006-09-004-7 EPJ A direct
electronic only

Experiments with polarized 3He at MAMI

A1 and A3 Collaboration
D. Rohea
Departement fur Physik und Astronomie, Universitat Basel, Klingelbergstr. 82, 4056 Basel, Switzerland

/
Published online: 11 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. Experiments with polarized 3 He at MAMI have already a long tradition. The A3 Collaboration
started in 1993 with the aim to measure the electric form factor of the neutron. At this time MAMI was
the second accelerator where experiments with 3 He were possible. Some years before this pilot experiment
the development of the apparatus to polarize 3 He in Mainz started. There are two techniques which allow
for
to polarize sucient large quantities of 3 He. Both techniques will be compared and the benet of 3 He
at MAMI will be given
nuclear physics will be discussed. A review of the experiments done so far with 3 He
and the progress in the target development, the detector setup and the electron beam performance will be
pointed out.
PACS. 13.40.Gp Electromagnetic form factors 13.88.+e Polarization in interactions and scattering
25.70.Bc Elastic and quasielastic scattering 29.25.Pj Polarized and other targets

1 Introduction  Final-state interactions (FSI) and me-

 has gained increasing interest due to its
Polarized 3 He
special spin structure described below, but also due to
the fact that the Schrodinger equation for the three-body
system can be exactly solved by means of the Faddeev
formalism [1,2]. Furthermore it is the only polarized tar-
get which tolerates currents of several A compared to
100 nA for a ND  3 target. This helps to compensate
the smaller thickness of the gas target. The gas target has
the advantage that it is almost not diluted by unpolarized
 3 target.
carrier material as it is the case for the ND
With the availability of highly polarized 3 He of sev-
eral bars and the delivery of polarized continuous electron
beams of high intensity, spin-dependent quantities can be
studied, which show a large sensitivity to the underlying
nuclear structure and reaction mechanism. Since in 3 He
the protons reside with high probability in the S-state, the
spin of 3 He is essentially carried by the neutron [3]. This
property of the 3 He-spin structure can be best exploited
 e, e n) with restriction to
in the quasielastic reaction 3 He(
small missing momenta as well as in inclusive 3 He( e, e )
near the top of the quasielastic peak. In such kinematics

the 3 He-target has been used extensively as polarized neu-
tron target to measure the magnetic [4,5,6] and electric [7,
8,9,10,11] form factors of the neutron, Gmn and Gen .
Combining the theoretical calculation with the data
gives insight into the three-body system and the nuclear
a
e-mail: Daniela.Rohe@unibas.ch
structure of 3 He.
son exchange currents (MEC) can be probed and studied
under dierent kinematical conditions. There are also re-
actions and kinematics where 3 He does not appear as neu-
   appears
tron target. In the He(e, e p)d reaction, e.g., 3 He
3
as a polarized proton target [12]. Such a measurement will
also be presented in these proceedings.
The usual Faddeev calculations include FSI and MEC,
but they are carried out non-relativistically. It was shown
in [13] that in particular relativistic kinematics plays
an important role already at Q2 = 0.67 (GeV/c)2 (see
sect. 3.2). Less important is the relativistic treatment
of the current operator and of the 3 He ground state.
A relativistic ground-state wave function became only
recently available with the development of a Lorentz
boosted nucleon-nucleon potential. It was constructed
with the condition to give the same NN phase shifts
with the relativistic Lippmann-Schwinger equation as the
non-relativistic potential when used with the Schrodinger
equation [14]. The problem of the relativistic version of
the Faddeev calculation is that it can treat only the in-
teraction between the two nucleons which are not directly
involved in the reaction (= spectators). We hope that fur-
ther ongoing theoretical work will lead to a full relativistic
treatment of the three-body system. Experimental data
will support such an eort.
On occasion of the symposium this contribution aims

at giving a review of experiments performed with 3 He
at MAMI in the last 20 years. The huge progress made
30 The European Physical Journal A
during this time in the development of the target, the
performance of the polarized electron beam and the im-
provement of the detector setup will be demonstrated. The
dierent objectives for the experiments will be presented.
Finally upcoming experiments at MAMI in the near future
are briey presented.
2 Polarization methods
For nuclear target applications two methods are in use,
metastable-exchange optical pumping (MEOP) [15] and
spin-exchange optical pumping (SEOP) [16]. Both meth-
ods were already developed in the 60ies but became
only ecient in use with the development of laser light
sources of sucient power and proper frequency band
width. MEOP is used for the Mainz target whereas at,
e.g., Jeerson Lab the SEOP technique is applied. Both
techniques will be shortly explained and the advantages of
each method discussed. It should be mentioned that there
is a third method to polarize 3 He. Here high magnetic
elds and low temperatures are needed which leads to po-
larizations of 38% in solid 3 He [17]. Due to the low heat
conductivity of the solid 3 He this method is not suitable
for nuclear-physics experiments with electron beams.
In MEOP as well as in SEOP angular momentum is
transferred to the atomic electrons by resonant absorption
of circularly polarized light and subsequent re-emission of
unpolarized light. A magnetic eld of 530 G denes the
quantization axis. In MEOP an atomic transition in 3 He
is pumped whose lower level is the metastable 23 S1 state.
It is reached by a weak gas discharge (a fraction of 106
atoms is excited). Therefore this method works only at low
pressures of about 1 mbar which also guarantees a su-
ciently long lifetime of the 23 S1 state. With moderate laser
power of about 1020 W and for large gas quantities of 20
liter at 1 mbar a polarization up to 80% can be reached in
a minute. The formerly used LNA-Laser ( = 1083 nm,
10 W) is nowadays replaced by two ytterbium ber
lasers (15 W each). Due to hyperne coupling the elec-
tronic polarization results in a corresponding alignment
of the nuclear spin. Subsequent collisions between polar-
ized 3 He -atoms in the rst excited metastable state and
unpolarized 3 He-atoms in the ground state transfer the
nuclear polarization to the ground state 3 He. The process
of metastable-exchange collisions is fast and has a large
cross section (1015 cm2 ). Therefore this method is quite
ecient. The drawback of this method is that it can only
be applied at low pressures of about 1 mbar.
In SEOP an alkali-vapor (usually Rb) is optically
pumped by the circularly polarized light provided by a Ti-
sapphire laser or by diode lasers tuned to the D1 -resonance
line of 795 nm. Once the Rb is polarized, the polarization
is transferred to the 3 He via spin-exchange collisions. The
spin-exchange mechanism proceeds via the hyperne in-
teraction between the 3 He nucleus and the Rb valence
electron. This can induce both species to ip their spin.
Because this interaction is weak the cross section for this
process is small (1024 cm2 ). Therefore optically thick Rb
vapor and large laser power (> 40 W) are needed to po-
larize the gas in a target cell of 10 bar within 20 h to 50%.
The advantage is that no further compression stage is re-
quired and a compact design is possible. To avoid radiation
trapping in the optical thick Rb vapor which occurs when
unpolarized resonant uorescence light is emitted and af-
terwards reabsorbed, 50100 mbar nitrogen is added. The
addition of a fraction of 102 N2 leads to 5 (10)% con-
tribution to the scattering rate from a proton (neutron)
and therefore to an eective dilution of the polarization
observables.
Except for experiments in a storage ring the mass
density of a few mbar of polarized 3 He  from MEOP is
too low for a nuclear physics experiment. Therefore one
or two mechanical compression stages1 are necessary to
reach pressures of up to 6 bars. Up to now three dierent
polarizers were in use for nuclear physics experiments at
MAMI. The rst one was the Toepler compressor which
uses 17.6 kg mercury as a piston. The pressure achieved
in the 100 cm3 target cell was 1 bar and the polarization
could be increased from 38% to 49% from 1993 to 1995.
The target cell was lled in a continuous ow (0.1 bar l/h)
with polarized gas and the polarization loss from the low
pressure pumping cell to the target was 30%. The increase
of the polarization was achieved by coating the target cell
with cesium to reduce the relaxation of the polarization
due to collisions with the container material (glass). The
Toepler compressor was developed for the rst measure-
ment of the electric form factor of the neutron Gen which
is described below. Nowadays the compression stage is re-
placed by one titanium piston which allows a production
rate of 1.5 bar l/h. The polarization losses are negligible
and the target cell is lled with 5 bar and 75% polariza-
tion. This is a great improvement and increases the per-
formance of the nuclear physics experiment signicantly.
3 Experiments with polarized 3 He
3.1 The electric form factor of the neutron
3.1.1 Motivation and techniques
Form factors describe the contribution from the inner
structure of a scatterer to the interaction. For spin-1/2
particles there are two form factors determining the
electromagnetic response, the magnetic and the electric
form factor. They are related via a Fourier transforma-
tion to the magnetic and to the charge distribution (see
sect. 3.1.5), respectively. A form factor independent of
the momentum transfer q to the particle would indicate
a point-like distribution, hence any q-dependence points
to an underlying substructure. The electric form factor
Gen of the neutron is particularly sensitive to its internal
structure because it is not obscured by the total charge
as in the case for the proton. The substructure of the nu-
cleon is determined by the (sea-)quarks and the gluons.
1
In the Hermes experiment the cell was cooled down to 25 K
to achieve a compression factor of 3.5.
 A1 and A3 Collaboration (D. Rohe): Experiments with polarized 3 He at MAMI
0.08
0.06
Gen
0.04
0.02
0 two-body currents is relatively small and the sensitivity
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
2 2
Q (GeV/c)
Fig. 1. The data points have been determined from elastic
electron-deuteron scattering [19] using the Paris potential. The
curves show how much the extracted Gen -values would vary
if the data had been analyzed by other than the Paris NN
potential (Nijmegen, Argonne V14, Paris, Reid Soft Core from
top to bottom) and demonstrate the model dependence of the
resulting Gen .
Therefore Gen is a particularly suitable test case for our
understanding of the quark degrees of freedom and a con-
straint for models. QCD would be the rst choice to calcu-
late form factors but it is still limited due to the computer
power available. Often approximations (quenched lattice)
are applied to avoid the computationally expensive part.
Recently a full lattice QCD could reproduce the trend of
the data [18]. With the extension of the data base in the
last few years the theoretical interest also increases and a
large variety of models and model-based ts were devel-
oped. This was not the case in 1987 when the rst Gen
measurement at MAMI was planned. The data base was
scarce and in particular the error bars exceeded 100%. The
reason: Gen is dicult to measure, as its value is small,
roughly a factor 10 smaller than the magnetic form fac-
tor Gmn . The nucleon form factors enter the (quasi)elastic
cross section quadratically, so the magnetic scattering am-
plitude dominates by far. Therefore an LT separation in
the reaction (e, e n) leads to unreasonably large errors. A
further complication comes from the fact that there exists
no free neutron target of sucient density. Thus the con-
tribution of the neutron to scattering o the deuteron or
3
He have been employed. The extraction of this contribu-
tion, however, requires to account for the nuclear structure
and, for elastic scattering o the deuteron, for the large
contribution of the proton electric form factor.
In 1990 the best data were measured by Platchkov and
collaborators [19] using elastic scattering on the deuteron.
An LT separation gives the longitudinal and the trans-
verse structure functions A(Q2 ) and B(Q2 ). A(Q2 ) de-
pends quadratically on the charge and quadrupole form
factors of the deuteron. The charge form factor dominates
for small Q2 (< 0.4 (GeV/c)2 ). It is proportional to (Gep
+ Gen )2 and therefore contains an interference term Gep
times Gen which increases the sensitivity to Gen . On the
other hand the contribution from Gep 2 had to be removed
which increases the uncertainty in Gen . The main draw-
back is that the extraction of Gen from A(Q2 ) requires
the removal of the contribution from the deuteron struc-
ture via a calculation, which depends on the chosen NN
potential. This introduces a large model dependence of
about 50% as shown in g. 1 [19]. Analyzing the data
31
with the modern NN potentials would lead to a smaller
model uncertainty. The treatment of MEC, which makes
a signicant correction, introduces further uncertainties.
The systematic errors described above can be signi-
cantly reduced by exploiting the quadrupole form factor
of the deuteron instead of A(Q2 ). The contribution from
0.9 1
of Gen to the chosen NN potential is reduced [20]. How-
ever, at low Q2 the statistical error of FC2 is large be-
cause the monopole form factor FC0 dominates the T20
data. Thus, the analysis using A(Q2 ) becomes superior
for Q2 < 0.4 (GeV/c)2 .
A method which is much less model dependent exploits
the observables measurable in a double-polarization exper-
iment. In exclusive reactions it is a sensitive tool to mea-
sure Gen . Here the longitudinally polarized electron beam
scatters quasi-elastically on deuterons or 3 He, which are
either polarized or where the polarization of the knocked-
out neutron is detected [21] (see contribution of M. Ostrick
to this symposium). The asymmetry with respect to the
electron helicity contains an interference term Gen times
Gmn which amplies Gen by Gmn . The sensitivity to Gen
is largest in the perpendicular asymmetry A , where the
direction of the target spin is perpendicular to the momen-
tum transfer (or the polarization of the scattered neutron
is perpendicular to its momentum, respectively). In con-
trast the parallel asymmetry A does not depend on form
factors (for Gen small) and therefore can serve as nor-
malization. Measuring the asymmetry has the advantage
that no absolute cross section measurements are required
which avoids the eort (and systematic errors) of deter-
mining absolute eciencies, solid angles and luminosity.
The electron-target asymmetry is obtained via
N + /L+ N /L
Aexp = , (1)
N + /L+ + N /L
where L+ (L ) are the integrated charge and N + (N )
the number of events for positive (negative) electron helic-
ity. The electron helicity is ipped every second randomly.
In general, the asymmetry A can be decomposed ac-
cording to the direction of the target spin which is given
by the angles S and S with respect to the momentum
transfer q and the scattering plane
A = A sin S cos S + A cos S . (2)
Before the asymmetry obtained in the experiment can be
compared to theory it has to be corrected for the polar-
ization of the electron beam Pe and the target PT as well
as for a dilution factor V :
1
A= Aexp . (3)
Pe PT V
The dilution factor V can come from the scattering on
unpolarized carrier material in the target or scattering on
the target container (background). Also charge exchange
p n in the shielding in front of the hadron detector con-
tributes to V because the protons in 3 He are almost un-
polarized. The corrected asymmetry A contains the elec-
tromagnetic form factors but also depends on the reaction
32 The European Physical Journal A

mechanism involved (see below). For scattering on a free

neutron with polarization Pn one has

1 2 (1 + ) tan(/2)Gen Gmn
A = , (4)
Pe Pn G2en + G2mn ( + 2 (1 + ) tan(/2))

1 1 + + (1+ )2 tan2 (/2) tan(/2)G2mn
A = 2 .
Pe Pn G2en + G2mn ( + 2 (1 + )) tan(/2)

Gen is determined best from the ratio of the asymmetries

A and A ,
A Gen
, (5)
A Gmn
instead of A alone. This has several advantages. The
product of the polarizations, Pe PT in eq. (3), drops out
in the ratio, thereby the systematic error introduced with
the two measurements of absolute polarizations is consid-
erably reduced. Furthermore the dilution factor V cancels.
In addition, theoretical corrections accounting for the nu-
 are reduced in the ratio. Some cor-
clear structure in 3 He
rections like the polarization of the neutron, Pn , are af- Fig. 2. Setup of the Gen experiment in the A3 hall at MAMI.
fecting both asymmetries in the same way and therefore Electrons are detected in the segmented lead glass detector in
also cancel in the ratio. For a bound, moving neutron Pn is coincidence with neutrons in the plastic scintillator array.
 measured.
usually smaller than the polarization PT of 3 He
It is a function of the initial momentum of the neutron and 0.08
3
therefore its mean value depends on the detector accep- He(e,e)
D(e,en)
0.06
tances.
Gen

0.04

0.02 3
3.1.2 Form factor measurements in the experimental hall He(e,en)
A3 at MAMI 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
2 2
Q (GeV/c)
The setup for the Gen experiment in the A3 experimen-
Fig. 3. (Uncorrected) result from the rst measurement of Gen
tal hall is shown in g. 2. Electrons were detected in [7]. Two double-polarization experiments
a calorimeter of 256 closely packed lead glass counters at MAMI using 3 He
performed at Bates in the same time period are also shown [24,
(4 4 cm) with a solid angle of 100 msr in a distance
25, 26]. For comparison the data already shown in g. 1 are
of 1.9 m. The shower produced by an electron extends
displayed as well.
over about 10 modules. Therefore the energy summed
over clusters of detectors was used leading to an energy
resolution E/E = 20% FWHM. This moderate energy
In this setup the already mentioned Toepler compres-
resolution was sucient to separate the inelastic contri-  with polarizations in-
bution from the quasielastically scattered events. The in- sor was used to produce 1 bar 3 He
elastic events, mainly resulting from -production in the creasing from 38% to 49% from 1993 to 1995. At the same
-resonance, have vanishing asymmetry. In case of an ad- time the electron polarization could be increased from 30
mixture this would dilute the asymmetry. In front of the to 50% by changing the cathode from a bulk to a strained
calorimeter a focusing air Cerenkov detector was placed layer GaAsP. Keeping in mind that the statistical error of
which suppresses background from electrons scattered on the asymmetry decreases with (Pe PT T )1 (T : measure-
the exit or entrance windows of the target cell or the beam ment time) both improvements enhance the performance
line. Further it serves to discriminate photons and pions of the experiment signicantly.
from electrons. With this setup Gen was measured at Q2 =
The neutrons were detected in a plastic scintillator ar- 0.31 (GeV/c)2 [7,8]. The pilot experiment of Meyerho et
ray which covers 250 msr and therefore the entire Fermi al. in 1993 [7] did only use a quarter of the detector setup
cone. It consisted of two walls and could also be used as shown in g. 2. Its result is shown in g. 3 together with
a neutron polarimeter for the Gen -measurement using the other double-polarization experiments performed at Bates
reaction D(e, en) [22,23]. The overall detector thickness  e, e ) [24,25] and D(e, en) [26].
at the same time using 3 He(
of 40 cm yields a neutron detection eciency of n = 32%. In g. 3 the uncorrected results are shown. In the mid of
The neutron detector was shielded with 5 cm lead on the the 90ies it was not clear that the measured value for Gen
front and surrounded by 1 m concrete against electromag-  as polarized neutron target needs a large cor-
using 3 He
netic background. rection accounting for FSI. No exact Faddeev calculation
 A1 and A3 Collaboration (D. Rohe): Experiments with polarized 3 He at MAMI 33

was available and the diagrammatic approach of Laget [27,

28] indicated a negligible correction. Later it was shown
by the Bochum-Krakow group that a correction of about
30% had to be applied on Gen at Q2 = 0.35 (GeV/c)2 [9].

3.1.3 Non-PWIA contributions

An experimental measure for non-PWIA contributions is

the target analyzing power Ay . For coplanar scattering
Ay is identical to zero in PWIA due to the combination
of time reversal invariance and hermiticity of the transi-
tion matrix [29]. Thus, a non-zero value of Ay signals FSI
and MEC eects and its measurement provides a sensitive
check of the calculation of these eects. For an unpolar-
ized beam and the target spin aligned perpendicular to
the scattering plane the target analyzing power can be
measured
1 N N
Aoy = , (6)
PT N + N
where N , (N ) are the normalized 3 He(e, e N ) events for
target spin aligned parallel (antiparallel) to the normal
of the scattering plane. This quantity was measured for

the reactions 3 He(e, 
e p) and 3 He(e, e n) at Q2 = 0.37 and
2 an early result from Laget [27]. Clearly it underestimates
0.67 (GeV/c) [11] using the experimental setup in the A1
spectrometer hall at MAMI described below. For this the
target spin, aligned perpendicular to the scattering plane,
was reversed every 2 minutes. Contrary to the determina-
tion of Gen , dilution eects do not cancel for Ay in eq. (6)
and have to be determined. The main contribution for

3 He(e, e n) comes from charge exchange in the 2 cm lead
shielding in front of the hadron detector. This factor was
determined using hydrogen as target. Then the recoil pro-
ton was tagged with the elastically scattered electrons in
the spectrometer and the number of neutrons detected in
the scintillator were counted. The correction from charge
exchange in the lead shielding is 10 to 15% for (e, e n).
The corrected experimental result [11] for the reaction
 ment was extended to higher Q2 . A pilot experiment was
3 He(e, e n) is shown in g. 4 together with the data point
measured at NIKHEF [30]. Furthermore the calculation
0.4
Ay
Fig. 5. Electron momentum spectrum measured in spectrom-
eter A for two electron helicities. The target spin was aligned
parallel to q. The solid line corresponds to a Monte Carlo sim-
ulation.
from the Bochum-Krakow group is shown including FSI
(dashed) and FSI plus MEC contributions (solid). The
eect from MEC is small as expected for quasifree kine-
matics. A more detailed examination revealed that the
large contribution from FSI at small Q2 comes mainly
from (e, e p) followed by charge exchange. The full calcu-
lation is in good agreement with the data. Also shown is
the eect from non-PWIA reaction mechanisms. Figure 4
conrms that the FSI contribution and thus the theoreti-
cal correction to Gen gets smaller with increasing Q2 . This
is expected from simple arguments like the decreasing of
charge-exchange cross section with Q2 and the shorten-
ing of the interaction time during the reaction at higher
momentum transfer.
3.1.4 Form factor measurements in the A1 spectrometer
hall at MAMI
To avoid a large theoretical correction the Gen measure-
performed already in 1997 in the A1 spectrometer hall at
a Q2 of 0.67 (GeV/c)2 followed by a second experiment in
2000 to double the statistics. Using the same setup data
on Ay were taken (see sect. 3.1.3). Both the target and
detector setup were considerably improved compared to
the experiment in the A3 hall.
The scattered electrons were detected in the magnetic

spectrometer A which has a focal plane detector consisting

0.2 of two drift chambers, a scintillator array and a Cerenkov
detector. It has a momentum acceptance of 20% and a
0 solid angle of 28 msr. Due to its high resolution and the
0 0.2 0.4 0.6 0.8 ecient pion rejection in the Cerenkov the inelastic con-
2 2
Q (GeV/c) tribution can be well separated from the quasielastic re-
Fig. 4. Target asymmetry in the reaction 3 He(e,
e n) as gion. Further the spectrometer serves to determine the
2
a function of Q measured at MAMI [11] (circles) and at direction of the momentum transfer q with good preci-
NIKHEF [30] (square). Shown is the result of the non- sion. The angle between q and the target spin direction
relativistic Faddeev calculation including FSI and MEC (solid) has to be precisely known for the extraction of Gen from
and FSI only (dashed). The dot-dashed line (in green, close to A . A momentum spectrum of the scattered electrons
zero) is obtained when neglecting charge-exchange by setting with an incident energy of 855 MeV is shown in g. 5.
Gep = Gen = 0 in the calculation. The dotted curve represents The two spectra belong to dierent electron helicities and
the result from a diagrammatic approach [27]. target spin parallel to the momentum transfer. The thick
34 The European Physical Journal A
solid line represents a Monte Carlo simulation using sim-
ple kinematical relations valid in Born approximation. The
good agreement supports that the contribution from non-
quasielastic events is negligible.
The hadron detector was placed in the direction of q.
It consists of four layers with ve scintillator bars each. In
front of the detector two layers of E detectors discrim-
inate protons and neutrons. In 160 cm distance from the
target the detector covers a solid angle of about 100 msr.
The entire detector was shielded with 10 cm lead except
for an opening towards the target where a reduced shield
of 2 cm was used. In addition a lead collimator in front
of the detector helps to suppress background produced in
the downstream beam line.
The data are taken from refs. [33,
and 3 He.
 electrons from D, D
The entire 3 He-target was enclosed in a rectangular
box of 2 mm thick -metal and iron except for a cut-
out towards the opening angle of the spectrometer. The
box served as an eective shield for the stray eld of the
magnetic spectrometers. Three independent pairs of coils
inside the box provided a homogeneous magnetic guiding
eld of 4 104 T. With additional correction coils a rel-
ative eld gradient of less than 5104 cm1 was achieved.
The setup also allowed for an independent rotation by re-
mote control of the target spin in any desired direction
with an accuracy of 0.1 .
The polarization of 3 He was monitored with Adia-
batic Fast Passage (AFP) using the technique described
in ref. [31] which measures the magnetic eld of the ori-
ented spins. Since the AFP-technique destroys part of the
polarization ( 0.10.2%) and since it cannot be used
during data-taking due to spin-ipping, it is used only
about once in 4 h. Therefore Nuclear Magnetic Resonance
(NMR) monitored continuously ( every 10 min) the rela-
tive polarization and served mainly as online control of the
polarization. The systematic error of the absolute polar-
ization is estimated to be 4% and the uncertainty in the re-
laxation time is 2 h. In the rst (second) beam time in 1997
(2000) the averaged target polarization was 32% (36%).

The 3 He-target consisted of a spherical glass container
(diameter 9 cm) with two cylindrical extensions sealed
with oxygen-free 25 m Cu windows. The Cu windows
were positioned outside of the acceptance of the spec-
trometer ( 5 cm) and shielded with Pb blocks to min-
imize background from beam-window interactions. The
3 uncertainty [38]. For this t the dipole form is modied
He-target was polarized via metastable optical pumping
to a typical polarization of 0.5 and compressed to an op-
erating pressure of 5 bar with a two-stage titanium com-
pressor [32]. Then the target cell was transported in a
portable magnetic eld to the target pivot. The relaxation
time of the polarization due to contact with the surface
is increased by careful cleaning and coating with cesium
to 80 h. The relaxation time is reduced to about 40 h due
to the dipole-dipole interaction between the 3 He-atoms at
high pressure and due to ionization of 3 He by the electron
beam. The latter process leads to the creation of 3 He+
and loss of polarization by transfer of angular momentum
to the rotational degrees of freedom. An electron current
of 10 A was used with a polarization of 7580%; the lat-
ter was measured with a Moeller polarimeter installed a
0.08 D(e,en)
0.07
D(e,en)
3
He(e,en)
0.06
0.05
Gen
0.04
0.03
0.02
0.01
0
0 0.5 1 1.5 2
2
Q (GeV/c)
2
Fig. 6. Gen extracted from quasielastic scattering of polarized
22, 23, 8, 34, 11, 35, 36, 37]. The dashed line represents the Gal-
ster t [38] and the solid curve the result of [39]. For some of
the experimental data points the correction due to the reaction
mechanism beyond PWIA is indicated by the size of the arrows.
few meters upstream of the target pivot in the A1 three-
spectrometer hall. Nowadays currents of more than 20 A
are routinely provided.
Combining the measurement of Ay (sect. 3.1.3) and
the Faddeev calculation of the Bochum-Krakow group one
estimates a correction to Gen at Q2 of 0.67 (GeV/c)2 of
(3.41.7)%. At this Q2 a relativistic calculation is already
needed (see sect. 3.2). The corrected Gen -value is shown
in g. 6 together with all published results from double-
polarization experiments (apart from the very rst ones
which may be regarded as results of feasibility studies).
Measurements using polarized deuterium instead of 3 He 
(indicated by a circle in g. 6) or detecting the polariza-
tion of the knocked-out neutron in a polarimeter (squares)
were performed at NIKHEF, Jlab and MAMI. Also indi-
cated in g. 6 are the theoretical corrections applied to the
Gen -value extracted from the data. Clearly the correction
decreases with increasing Q2 .
The dashed line in g. 6 is the so called Galster t,
determined from the data available in the 70ies. Most of
these data were obtained using elastic electron-deuteron
scattering. As mentioned in sect. 3.1.1 this method implies
a large model dependence. The Galster t was obtained by
using data up to Q2 = 0.8 (GeV/c)2 with large statistical
in such a way that the slope at small Q2 known from
n-e scattering could be reproduced. Surprisingly this t
still gives a good description of the actual data set up to
Q2 = 0.8 (GeV/c)2 .
3.1.5 Charge distribution of the neutron
A new t to the present data set was provided by Friedrich
2 and Walcher [39] using a phenomenological model of the
nucleon. In this model a superposition of two dipoles
for the smooth part and two Gaussians to account for
a possible bump is used as tting function. Their re-
sult is shown by the solid line in g. 6. Here a bump
 A1 and A3 Collaboration (D. Rohe): Experiments with polarized 3 He at MAMI
Fig. 7. Charge distribution (weighted with the radius squared)
of the neutron decomposed into the contributions from the bare
proton po , the pion cloud and the bare neutron no (picture
taken from [39]).
at Q2 0.2 (GeV/c)2 describes the data set best, but
more precise data are needed for a rm conrmation. It
is remarkable that also in Gmn , Gep and Gmp this bump
appears in the same Q2 region. Due to the large magnetic
moments of the nucleons and the charge of the proton it
is only visible if the form factor data are divided by the
dipole form factor.
A physically motivated t decomposes the neutron into
a bare neutron no and a polarization part:
n = (1 bn )no + bn (po + ). (7)
The bare neutron consists of three quarks with form fac-
tors assumed to be of the dipole form. In the polarization
part the neutron exists as a bare proton po surrounded by
a pion cloud. The form factor of the pion is constructed
from the spatial distribution of the harmonic oscillator
wave function in a p-state. With six free parameters a good
description of both Gen and Gep is achieved. According to
this t the neutron exists to 90% as bare neutron and to
bn 10% as proton with pion cloud.
From such a t the charge distribution of the neutron
can be obtained by a Fourier transformation,
 tial [41] instead of the AV18 NN potential [42]. The dif-
1 sin(Qr) 2
(r) = Gen (Q2 ) Q dQ, (8)
2 2 0 Qr
where Q is the momentum transfer in the Breit frame,
i.e. a frame with no energy transfer to the nucleon. The
behavior at high Q2 mainly determines the charge distri-
bution deep inside the nucleon at small radii r. In g. 7 the
result for the charge distributions (r) r2 of the neutron
(black line) as well as for the three components in eq. (7)
are shown. It is remarkable that the pion contribution ex-
tends as far as 2 fm (maximum at 1.5 fm). In contrast,
the authors of ref. [40] separated the contribution of the
two-pion continuum and found a peak at a distance of only
0.3 fm. The maximum of the pion cloud in the model of
Friedrich and Walcher corresponds to the Compton wave-
length of the pion ( = 1.43 fm) determining the range
of the nuclear force in the Yukawa model. This conrms
that in this model only one pion is taken into account by
construction. This consideration might resolve part of the
disagreement because in ref. [40] two-pion contributions
are considered.
35
3.2 Test of the theory at Q2 = 0.67 (GeV/c)2
The Faddeev calculation mentioned so far is fully non-
relativistic and it was not clear at which Q2 relativistic
eects would become non-negligible. On the other hand,
the contribution from non-PWIA reaction mechanisms to
 e, e n) are small at high Q2 as
the asymmetries in 3 He(
shown in sect. 3.1.4. More sensitive are the asymmetries
 e, e p). These asymmetries
A and A in the reaction 3 He(
are expected to be small because the two protons are most
of the time in the S-state. In this case the asymmetries
vanish, unless one resolves the dierent exit channels (see
sect. 3.3). Therefore comparing the experimental result to
the theory provides a sensitive test to eects from reaction
mechanisms as well as from the 3 He structure. Both might
need a relativistic treatment.
There are several ingredients in the Faddeev calcula-
tion which might be treated relativistically or non-relati-
vistically. This includes the 1-body current operator, the
T -matrix element describing the FSI, the kinematics and
 ground state wave function. It should be men-
the 3 He
tioned that up to now in the relativistic description only
the interaction between the spectator nucleons, i.e. the
ones which are not involved in the primary reaction, can
be included. This is called FSI23 or rescattering term of
rst order. At the moment there are no exact calculations
available for 3 He which can treat MEC and full FSI at high
 ground state be-
Q2 . The relativistic treatment of the 3 He
came only recently available with the development of a
Lorentz boosted NN potential. In ref. [14] such a poten-
tial was obtained and used in a relativistic 3N-Faddeev
equation for the bound state to calculate the triton bind-
ing energy. The results presented below are still based on
an exact but non-relativistic 3 He ground state. A newer
calculation prepared for a recent proposal to measure Gen
at Q2 of 1.5 (GeV/c)2 shows that the dierence is small.
The dependence on the NN interaction was studied
with a calculation which employs the CD-Bonn NN poten-
ference in the result is negligible. It should be mentioned
that the potential approach is not strictly valid when the
center-of-mass energy of the 3N-system, E3N , is well above
the pion production threshold. E3N can be obtained for
the 3-body breakup via

E3N = (MHe + )2 |q|2 2Mp Mn . (9)
However, in quasi-elastic kinematics the focus is mostly
on the region of phase space, where one of the nucleons is
struck with a high energy and momentum and leaves the
remaining two-nucleon system with a rather small internal
energy. Thus this approximation, which has to be made
also in other calculations for 3 He and deuterium [43,44],
might not be too serious.
 e, e p) were taken si-
The data on the reaction 3 He(
multaneously to the measurement of Gen at Q2 =
0.67 (GeV/c)2 . Protons were selected in the hadron de-
tector by requiring hits in two consecutive E detectors.
The background in the coincidence time spectrum, de-
termined from the time dierence between the rst bar
36 The European Physical Journal A
Fig. 8. Experimental results of A (left) and A (right) for the central region of the quasi-elastic peak as a function of the
scattering angle of the knocked-out proton. The result of the full (PWIA) calculation is shown with solid (dashed) line. The
results of three full calculations, however with a non-relativistic current (dot), or with only a (v/c)2 correction (dot-dot-dashed),
or with non-relativistic kinematics (dot-dash) are also shown.
in the hadron detector and the scintillator plane of spec-
trometer A, was negligible. In order to study the eect
of FSI on the asymmetries in dierent kinematic regions,
the quasi-elastic peak is divided into two regions of . One
region covers the peak and therefore emphasizes low nu-
cleon momenta whereas the other region covers the low
tail sensitive preferentially to high nucleon momenta. The
events in each of the two regions are summed over the en-
tire acceptance of the out-of-plane angle of electron and
proton and over the electron scattering angle in a range
from 75.8 to 81.8 .
In g. 8 the parallel and perpendicular asymmetries
in the central region of the quasielastic peak are shown
as a function of the scattering angle of the proton. They
are compared to the theory which contains the two-body
(2BB) and three-body breakup (3BB). The 3BB channel
is integrated over the rst 26 MeV. As can be seen from
the gures the PWIA calculation (dashed line) clearly dis-
agrees with the data. From the calculations which include
FSI23 only, the one with non-relativistic kinematics (dot-
dashed line) cannot describe the experimental results. Rel-
ativistic (solid line) or non-relativistic (dots) treatment
of the current operator does not make a large dierence.
The calculation taking into account relativistic kinematics
and FSI23 provides a good description of the data. Both
ingredients are important to achieve agreement with the
experimental results.

3.3 Structure of 3 He
In the experiment described in the previous section it was
not possible to separate the 2BB and 3BB channel due
to the limited resolution of the hadron detector. For a
 spectrome-
better understanding of the structure of 3 He
ter B was taken for proton detection. The kinematics was
limited to the central region of the quasielastic momen-
tum distribution at Q2 of 0.31 (GeV/c)2 . Each hour the
target spin was turned to measure the parallel, perpendic-
ular, antiparallel and antiperpendicular asymmetry alter-
nately with the purpose to reduce the systematic errors.
The target cell was of the same kind as already used for
the Gen measurement (see sect. 3.1.4). A new polarizer
was used consisting of one-stage titanium compressor [45]
with eectively no polarization loss during the transfer
from the low pressure gas reservoir to the target container.
The 3 He was optically pumped with two Ytterbium ber
lasers each providing 15 W on the resonance transition
(1083 nm). With this setup an initial target polarization
of 70 to 75% could be achieved. Averaged over the beam
time period and accounting for relaxation a target polar-
ization PT of (49.8 0.3(stat.) 2(syst.))% was obtained.
From the measured kinematic variables in the two
spectrometers, the missing energy is reconstructed accord-
ing to
Em = E Ee Tp TR . (10)
Here, E (Ee ) is the initial (nal) electron energy and Tp
is the kinetic energy of the outgoing proton. TR is the ki-
netic energy of the (undetected) recoiling (A1)-system,
which is reconstructed from the missing momentum un-
der the assumption of 2BB. The resulting Em distribution
reconstructed from the data is shown in g. 9 as thick
solid line. The resolution is limited mainly by the prop-
erties of the target cell and not by the resolution of the
spectrometers. The FWHM of 1 MeV allows a clear sep-
aration of the Em -regions where only 2BB or 2BB and
3BB contribute. The Em -region from 4.0 to 6.5 MeV is in-
terpreted as pure 2BB. This cut was chosen to avoid any
contribution from the 3BB-channel (starting at 7.7 MeV)
considering the experimental Em resolution. In agreement
with ref. [46], the yield of the 3BB is negligible beyond
25 MeV. Therefore the cut for the 3BB-channel was made
from 7.5 to 25.5 MeV in the Em spectrum. Because the
3BB resides on the radiation tail of the 2BB, the latter
has to be accounted for in the analysis of the 3BB-region
of the measured spectrum. To this end, the tail was cal-
culated in a Monte Carlo simulation which accounts for
 A1 and A3 Collaboration (D. Rohe): Experiments with polarized 3 He at MAMI
2
x 10
counts per bin
1600
1400
1200
1000
800
600
400
200
0 ||
5 10 15 20 25
Em (MeV)
Fig. 9. Experimental Em distribution (thick line) and the sim-
ulation of the 2BB (thin red line). The dierence is shown as
thick black line 3BB.
internal and external bremsstrahlung, ionization loss and
experimental energy resolution adjusted to the experimen-
tal distribution. The simulated 2BB distribution is shown
as thin red line in g. 9. Subtracting this from the data
leads to the distribution belonging to the 3BB channel
which is also shown in g. 9.
The ratio of the Monte Carlo simulation of the 2BB to
the experimental data in the region of the 3BB is denoted
by a23 . For the region 7.5 < Em < 25.5 MeV it amounts to
a23 = 0.434 0.002(stat.) 0.015(sys.). Then the asym-
metry A3BB for the 3BB-channel is extracted from the
asymmetry A2+3BB in the 3BB region by accounting for
the contribution from the radiation tail
A2+3BB A2BB a23
A3BB = . (11)
1 a23
All asymmetries are corrected for target and electron po-
larization. In g. 10 the parallel and perpendicular asym-
metries A3BB and A2BB are compared to two calculations
of the Bochum-Krakow group. One uses PWIA only (dot-
dashed), the other accounts for full FSI and MEC (solid
line). The eect of MEC is negligible in this kinematics.
The data integrated over the total detector acceptance are
in good agreement with the calculation including FSI.
The calculation shows that the FSI contribution is
small in the 2BB while it is large in 3BB. This suggests
that the main contribution of FSI results from the rescat-
tering term which does not exist in the 2BB, and not from
direct FSI. This was also conrmed by further examina-
tion of the theoretical result by Golak [47].
In the 2BB channel the spins of the neutron and proton
in the recoiling deuteron are coupled to one, therefore they
are parallel. Consequently, in a simplied picture, the spin
of the second (knocked-out) proton must be antiparallel
 Correct
to the deuteron spin and thus to the spin of 3 He.
coupling of the spins 1 and 1/2 to 1/2 leads to 33% polar-
ization of the knocked-out proton relative to that of the

polarized 3 He-target. This is precisely what is observed as
37
0.2
2BB
0 3BB
asymmetry
-0.2
3BB
-0.4
|| ||
target spin direction
Fig. 10. Comparison of the data to the theoretical calcula-
tion for the 2BB and 3BB for the four target spin directions
(anti)parallel (, ; left panel) and (anti)perpendicular (,
; right panel). In addition the combined sum for the parallel
and perpendicular position is shown ( and , respectively). To
facilitate the comparison, all 2BB (3BB)data are shown with
positive (negative) sign. PWIA: dot-dashed lines. Full calcu-
lation including FSI and MEC: solid lines. Statistical errors
point up, systematic uncertainties point down. For the 2BB
the size of the error bars is smaller than the symbols.

A2BB . In the 2BB channel, the polarized 3 He-target can
thus be interpreted as a polarized proton target.
For the 3BB channel the situation is dierent. In
PWIA the asymmetry is almost zero for the 3BB which
reects the fact that the two protons, which are domi-
nantly in the S-state and thus have opposite spin orien-
tation, now contribute equally to the knock-out reaction.
The inclusion of FSI, however, leads to an asymmetry,
which is larger and opposite in sign compared to the 2BB.
The main eect comes from the np t-matrix (rescattering
term). Since dierent spin combinations of the singlet and
 target cannot be in-
triplet np t-matrix contribute, the 3 He
terpreted as a polarized proton target in the 3BB channel.
4 Summary and outlook
In this contribution a review of the experiments with po-
 performed at MAMI was given. The eort to
larized 3 He
build a machine to polarize 3 He started already in 1987.
The rst experiment at MAMI with 3 He  was performed
in the experimental hall A3 to measure the electric form
factor of the neutron. Experiments with the same purpose
at higher Q2 followed, using improved target and detector
setups in the three-spectrometer hall A1. With the new
detector setup a better discrimination of inelastic events
from the ones quasielastically scattered is possible. The
performance of the target was steadily improved due to
the development of new polarizers. This resulted in a more
dense target (5 bar) with higher polarization (PT = 50%).
In addition the electron source was improved using a
38 The European Physical Journal A
strained layer crystal. This led to nowadays available cur-
rents of 20 A with an electron polarization of 75%.
Parallel to the experiments the non-relativistic Fad-
deev calculation was developed by the Bochum-Krakow
group. One of the rst applications was the calculation
of the correction of Gen at Q2 = 0.35 (GeV/c)2 due to
FSI which leads to a deviation of the asymmetry mea-
 from that for a free neutron. The inuence
sured on 3 He
of FSI was also conrmed by measuring the target asym-
metry Ay where the beam is unpolarized and the target
spin perpendicular to the scattering plane. This quantity
is particularly sensitive to FSI and MEC contributions.
In PWIA it vanishes. Good agreement between data and
theory was found.
Another experiment concentrated on the question
when a relativistic calculation is needed and which ingre-
dients need to be treated relativistically. For this the reac-
 e, e p) was investigated at Q2 = 0.67 (GeV/c)2 .
tion 3 He(
It turned out that the kinematics has to be treated rel-
ativistically already at this Q2 . On the other hand, a
relativistic current operator is much less important. At
present a relativistic calculation is only possible in PWIA
and with FSI23 included.
 17. D.G. Haase et al., Nucl. Instrum. Methods A 402, 341
To become more sensitive to the inner structure of 3 He
3  
the 2BB and 3BB channels in the reaction He(e, e p) were
separated. Also here the theoretical calculation is in good
agreement with the data. It is interesting that in the 2BB
channel, which is almost unaected by FSI at the kine-
matics of the present experiment, the 3 He  target can be
considered as a polarized proton target with the proton
 By contrast, the 3BB chan-
spin opposite to that of 3 He.
nel is largely aected by FSI. In this case 3 He cannot be
interpreted as polarized proton target.
All these reactions considered so far were not sensi-
tive to MEC because the kinematics were chosen to cor-
respond to the top of the quasielastic peak and the Q2
was suciently high. At Q2 < 0.2 (GeV/c)2 MEC con-
tribute signicantly to the reaction and modify the asym-
metries. Since MEC are not so well understood as com-
pared to FSI it is planned to study kinematics which are
sensitive to MEC. The data taken to measure Gen at
Q2 = 0.25 (GeV/c)2 are aected by MEC in some kine-
matical regions covered by the detector acceptance.
With the upgrade of MAMI to MAMI-C the Gen mea-
surement will be pushed to Q2 = 1.5 (GeV/c)2 . For this a
new hadron detector is under construction which should
have a higher neutron detection eciency.
There are also plans to use polarized 3 He  with (po-
larized) photons in the A2 experimental hall. Then 3 He  41. R. Machleidt, F. Sammarruca, Y. Song, Phys. Rev. C 53,
would be used as a polarized neutron target to measure
the Gerasimov-Drell-Hearn sum rule. For this a new target
setup is needed which is already under consideration.
Finally I want to thank Karl-Heinz Kaiser for the excellent
beam quality at MAMI and for his eort to adjust and setup
the beam for our sensitive experiments. Then I want to thank
Jorg Friedrich and Thomas Walcher for their support and ad-
vice as well as Hartmuth Arenhovel, Hartmut Backe and Dieter
Drechsel for the good atmosphere in the institute.
References
1. J. Golak et al., Phys. Rev. C 65, 064004 (2002).
2. R.W. Schulze, P.U. Sauer, Phys. Rev. C 48, 38 (1993).
3. B. Blankleider, R.M. Woloshyn, Phys. Rev. C 29, 538
(1984).
4. H. Gao et al., Phys. Rev. C 50, R546 (1994).
5. W. Xu et al., Phys. Rev. Lett. 85, 2900 (2000).
6. W. Xu et al., Phys. Rev. C 67, 012201(R) (2003).
7. M. Meyerho et al., Phys. Lett. B 327, 201 (1994).
8. J. Becker et al., Eur. Phys. J. A 6, 329 (1999).
9. J. Golak et al., Phys. Rev. C 63, 034006 (2001).
10. D. Rohe et al., Phys. Rev. Lett. 83, 4257 (1999).
11. J. Bermuth et al., Phys. Lett. B 564, 199 (2003).
12. P. Achenbach et al., Eur. Phys. J. A 25, 177 (2005).
13. C. Carasco et al., Phys. Lett. B 599, 41 (2003).
14. H. Kamada, W. Glockle, J. Golak, Ch. Elster, Phys. Rev.
C 66, 044010 (2002).
15. G.K. Walters, F.D. Colgrove, L.D. Schearer, Phys. Rev.
Lett. 8, 439 (1962).
16. M.A. Bouchiat, T.R. Carver, C.M. Varnum, Phys. Rev.
Lett. 5, 373 (1960).
(1998).
18. M. Gockeler et al., Nucl. Phys. A 755, 537 (2005).
19. S. Platchkov et al., Nucl. Phys. A 510, 740 (1990).
20. R. Schiavilla, I. Sick, Phys. Rev. C 64, 041002(R) (2001).
21. R.G. Arnold, C.E. Carlson, F. Gross, Phys. Rev. C 23, 363
(1981).
22. C. Herberg et al., Eur. Phys. J. A 5, 131 (1999).
23. M. Ostrick et al., Phys. Rev. Lett. 83, 276 (1999).
24. C.E. Jones-Woodward et al., Phys. Rev. C 44, 571 (1991).
25. A.K. Thompson et al., Phys. Rev. Lett. 68, 2901 (1992).
26. T. Eden et al., Phys. Rev. C 50, R1749 (1994).
27. J.M. Laget, Phys. Lett. B 273, 367 (1991).
28. J.M. Laget, Phys. Lett. B 276, 398 (1992).
29. H.E. Conzett, Nucl. Phys. A 628, 81 (1998).
30. H.R. Poolman, PhD Thesis, Vrije Universiteit te Amster-
dam, 1999.
31. E. Wilms et al., Nucl. Instrum. Methods A 401, 491
(1997).
32. R. Surkau et al., Nucl. Instrum. Methods A 384, 444
(1997).
33. I. Passchier et al., Phys. Rev. Lett. 82, 4988 (1999).
34. H. Zhu et al., Phys. Rev. Lett. 87, 081801 (2001).
35. R. Madey et al., Phys. Rev. Lett. 91, 122002 (2003).
36. G. Warren et al., Phys. Rev. Lett. 92, 042301 (2004).
37. D.I. Glazier et al., Eur. Phys. J. A 24, 101 (2005).
38. S. Galster et al., Nucl. Phys. B 32, 221 (1971).
39. J. Friedrich, Th. Walcher, Eur. Phys. J. A 17, 607 (2003).
40. H.-W. Hammer, D. Drechsel, Ulf-G. Meiner, Phys. Lett.
B 586, 291 (2004).
1483 (1996).
42. R.B. Wiringa, V.G.J. Stoks, R. Schiavilla, Phys. Rev. C
51, 38 (1995).
43. A. Deltuva et al., Phys. Rev. C 70, 034004 (2004).
44. H. Arenhovel, W. Leidemann, E. Tomusiak, Phys. Rev. C
46, 455 (1992).
45. E.W. Otten, Europhys. News 35, 16 (2004).
46. R.E.J. Florizone et al., Phys. Rev. Lett. 83, 2308 (1999).
47. J. Golak, private communication, 2005.
Eur. Phys. J. A 28, s01, 39 48 (2006)
DOI: 10.1140/epja/i2006-09-005-6 EPJ A direct
electronic only

Few-nucleon systems (theory)

M. Schwamba
Institut fur Kernphysik, Johannes Gutenberg-Universitat, D-55099 Mainz, Germany

/
Published online: 9 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. An overview over present achievements and future challenges in the eld of few-nucleon systems
is presented. Special emphasis is laid on the construction of a unied approach to hadronic and electro-
magnetic reactions on few-nucleon systems, necessary for studying the borderline between quark-gluon and
eective descriptions.

PACS. 13.40.-f Electromagnetic processes and properties 21.45.+v Few-body systems 25.30.-c Lepton-
induced reactions 25.20.-x Photonuclear reactions

1 Introduction may serve as eective neutron targets so that a better

understanding of few-nucleon systems may also lead to a
One of the most challenging topics in modern physics deals better understanding of neutron properties. Concerning
with the structure of atomic nuclei and their constituents. the test of eective theories, electromagnetic (e. m.) reac-
Despite the large eorts in the last decades, our present tions have always been at the forefront in nuclear struc-
understanding of hadronic systems is still far from being ture investigations. The electromagnetic interaction is well
satisfactory. The non-Abelian gauge structure of the un- known from classical electrodynamics and is weak enough
derlying fundamental theory quantum chromodynam- to allow a perturbative treatment in terms of the ne
ics (QCD) leads to enormous complications in practical structure constant 1/137.
applications. Therefore, one uses in conventional nuclear In this work, selected examples of present achieve-
physics not the fundamental quarks and gluons of QCD, ments in the eld of few-nucleon systems are presented.
but nucleons, isobars and mesons as relevant degrees of We concentrate ourselves mainly on the two-nucleon sys-
freedom (d. o. f.). These so-called eective approaches tem which deserves special attention because it has the
are presently still the most promising ones for reaching same relevance in nuclear physics as the H-atom in atomic
a quantitative understanding of hadronic physics at low physics. However, also some recent progress in the descrip-
and intermediate energies below about 1 GeV excitation tion of more complex few-nucleon systems is presented.
energy. A well-known example for the success of this ef-
fective picture is the quantitative understanding of N N -
scattering data below pion threshold in terms of meson- 2 The two-nucleon system
exchange mechanisms between two interacting nucleons
(for a pedagogical introduction, see [1]). 2.1 Introduction
On the other side it is clear that the eective de- Although the two-nucleon system is the simplest few-nuc-
scription will break down at some su ciently high en- leon system, it is far from being trivial. Even if we restrict
ergy/momentum transfer. Moreover, it is presently not ourselves to energies below the two-pion threshold, this
clear whether a clear cut borderline exists or whether even becomes obvious by noting that quite a large number of
at relatively small energies quark and gluon degrees of dierent reactions is possible like
freedom manifest themselves in specic reactions and ob-
servables. N N -scattering NN NN,
It is obvious that for a detailed study of such funda- Compton scattering d d,
mental questions a profound understanding of few-nucleon e. m. deuteron breakup d N N, ed e N N ,
systems is inevitably necessary because the corresponding photopionproduction d d, d N N ,
theoretical treatment is naturally the most cleanest one. elastic electron scattering ed e d,
Moreover light nuclei, especially the deuteron and 3 He, Bremsstrahlung N N N N ,
pionic reactions d d, d  N N ,
a
e-mail: schwamb@kph.uni-mainz.de N N N N .
40 The European Physical Journal A

above-mentioned necessity for a unied approach to the

(a) d sector. Needless to say that such a consistent picture is in
A In a rst step, we may restrict ourselves to the two-
(b) d channels with at most one asymptotically free pion need
(c) d
Fig. 1. Diagrammatic illustration of possible destinations of
an electromagnetically produced pion. Further discussion in
the text.
Thus the two-nucleon system oers a great variety of
possible interaction mechanisms worthwhile to be studied.
A very important point for the forthcoming discussion is
the fact that these dierent reactions cannot be treated
independently. First of all, they are linked by unitarity as
becomes obvious by considering the corresponding optical
theorems like
Im T (N N N N ) tot (N N N N, d, N N, . . .), (1) dence non-Hermitean. Moreover, G0 (z) has poles beyond
Im T (d d) tot (d N N, d, N N, . . .), (2) from three-body scattering theory [2,3]. Intuitively, they
Im T (d d) tot (d N N, d, N N, . . .), (3) describe the possibility that beyond pion threshold the
where the left sides are understood to be evaluated in for-
ward direction. This means, for example, that the forward
Compton scattering amplitude is related to all possible re-
actions with a photon and a deuteron in the initial state.
If the restriction to energies below the two-pion threshold
is dropped, of course also additional channels like 2-, K-
and -production have to be considered.
Therefore, as a consequence of unitarity, a unied de-
scription of all possible reactions is necessary. Before we
outline such an approach in some detail, let us try to un-
derstand the connection of the dierent above-mentioned
reactions from a more intuitive point of view without refer-
ring to formal arguments based on unitarity. For that pur-
pose, let us consider the three diagrams depicted in g. 1.
In all of them, a photon is absorbed by a deuteron produc-
ing a real or virtual pion. The three diagrams dier with
respect to their nal state: In diagram (a) the pion leaves
the two-nucleon system, whereas in diagrams (b) and (c)
it is absorbed by one of the two outgoing nucleons. Despite
the close relationship of the three diagrams, their physi-
cal interpretation is completely dierent: diagram (a) is a
contribution to photopionproduction on the deuteron, dia-
gram (b) a part of the meson-exchange currents (MEC) to
deuteron photodisintegration, and diagram (c) contributes
to the anomalous magnetic moment of the hit nucleon.
This simple example illustrates that single-particle prop-
erties, pion production mechanisms and meson-exchange
currents are closely related. This fact again underlines the
dierent possible reactions in the one- and two-nucleon
principle also required for more complex nuclei.
nucleon system for energies up to the -region so that a
basically nonrelativistic treatment should be su cient and
solely to be studied. Despite these simple boundary con-
ditions, the construction of such a unied approach is far
from being trivial and in fact not successfully realized till
now. In most existing approaches, only one or two reac-
tions of interest are selected and the rest is just ignored.
Moreover, simplifying approximations are used in order
to reduce the numerical complexity. To be more precise,
let us return to the dierent diagrams in g. 1. In the
meson-exchange contribution (b), a proper description of
the propagation of the intermediate N N -system (cut A)
requires for a given invariant energy W of the system the
numerical evaluation of the exact free retarded propagator
1
G0 (z) = (z HN (1) HN (2) H ) ,z = W i,
(4)
where HN (i) and H describe the kinetic energy opera-
tors for nucleon i and the pion, respectively. Although
this expression looks quite simple, its structure is quite
nontrivial: It is nonlocal and due to its energy depen-
pion threshold leading to logarithmic singularities known
produced pion must not necessarily be reabsorbed by
one of the nucleons but may become onshell as indicated
in diagram (a). Therefore, the singularities link N N - to
N N -scattering as required by the optical theorems (1)
through (3) and their correct treatment is inevitably nec-
essary.
Due to these features of G0 , it is obvious that its nu-
merical implementation is rather involved. Therefore, in
most of the approaches an approximative treatment, the
so-called static limit is used by assuming that the nucle-
ons are innitely heavy during the meson exchange (cut A
in diagram (b) of g. 1) so that in consequence no energy
transfer occurs. The resulting static propagator
1
Gstat
0 = (5)
H
is local, energy independent and regular. Due to these nice
features, which lead to large numerical simplications, it is
even nowadays very popular and used for example in state-
of-the-art high precision N N -potentials like AV18 [4]
or CD-Bonn [5,6]. The static limit works well below pion
threshold but we will see that this approximation fails at
higher energies. Intuitively, this is not very surprising: Due
to the lack of singularities in (5), the pion is frozen inside
the hadronic system and therefore no longer a dynamic
degree of freedom.
Finally, let us make a comment on the treatment of
diagram (c) in g. 1. In conventional approaches, it is just
 M. Schwamb: Few-nucleon systems (theory)
N N N N
Fig. 2. Graphical illustration of meson-nucleon-nucleon ver-
tices VXN (left) and VN X (right) which serve as the basic in-
gredients for the hadronic interaction.
Fig. 3. Graphical illustration of the one-boson-exchange po-
tential (left) and mesonic loop contributions to the nucleon
self-energy. Both terms are generated by the second iteration
of the XN -vertices depicted in g. 1.
neglected by arguing that one uses physical nucleons as
relevant eective degrees of freedom which already con-
tain the correct anomalous magnetic moments so that the
additional consideration of diagram (c) would lead to dou-
ble counting. However, one has to take into account that
the loop in diagram (c) is energy dependent so that its
simulation by a current governed by a constant anomalous
magnetic moment might be a rather crude approximation.
Moreover, due to the occurrence of the retarded propaga-
tor (4), the loop can become complex so that its neglect
violates the optical theorem (3) beyond pion threshold.
In order to solve these problems, a careful distinction be-
tween so-called bare and physical nucleons is necessary.
This conceptual complication is in most cases just cir-
cumvented by neglecting diagram (c).
2.2 Model structure
In this section, we present the general structure of our ap-
proach developed within the past years. It is suitable to
study all hadronic and electromagnetic reactions on the
two-nucleon system for energies up to the -resonance re-
gion with at most one asymptotic free pion. In order to
keep the discussion as transparent as possible, technical
aspects are mostly avoided. The interested reader is re-
ferred to [7,8,9,10] concerning further details.
In order to treat a meson X as a dynamic degree of
freedom, one has to work within a Hilbert space H where
X is treated explicitly. Consequently, one has to allow for
transitions between the N N - and the XN N -sector. In
our approach, they are generated by conventional XN -
vertices VXN and VN X = (VXN ) (X {, , , , . . .}),
see g. 2, known from N N -potential theory [1]. They serve
as the basic ingredients of the hadronic interaction V in
the Schrodinger equation for the deuteron bound state and
41
the scattering equation for continuum states. The latter
reads for a given invariant energy W as follows:
() (P W )
| = iG(W i)| , (6)
(P W )
where | denotes a plane-wave state (i.e. either a
noninteracting N N -, d- or N N -system) and G the full
propagator
1 1
G(z) = = , (7)
zH z H0 V
containing the potential V and the kinetic energy operator
H0 . The full propagator can be rewritten in terms of the
scattering amplitude
1
T (z) = V + V G0 (z)T (z) with G0 (z) = (z H0 )
(8)
according to
G(z) = G0 (z) + G0 (z)T (z)G0 (z). (9)
It contains, therefore, the interaction V up to innite or-
der. The second order terms of (8), which are depicted in
g. 3, consist rst of all of a one-boson-exchange potential
(OBEP) of the type
V OBEP (z) = VN X (1)G0 (z)VXN (2) + (1 2), (10)
and a contribution to the nucleon self-energy, see equa-
tion (11) below.
Consequently, the possibility of meson production and
annihilation as well as the structure of the N N -force is
based on the same XN -vertex VXN . This allows there-
fore to construct the desired unied approach. The price
we have to pay is at least twofold. First of all, the N N -
interaction is more complex as conventional ones because
it has to be treated in the exact retarded, energy de-
pendent manner. In our explicit realization, we use the
parametrization of the Elster potential [11], which just
consists of the diagrams of g. 3 with inclusion of -, -,
-, -, - and -exchange. The free parameters of the cor-
responding vertices (cutos, coupling constants) are tted
to the N N -scattering phase shifts for energies up to the
pion threshold.
As a second complication, the mesonic loop diagrams
V self (z) = VN X (1)G0 (z)VXN (1) + (1 2) (11)
appear, depicted on the right-hand side of g. 3. In or-
der to avoid any double counting, we have to distinguish
therefore bare from physical nucleons. Whereas the rst
ones are the basic d. o. f. of our Hilbert space, the latter
contain, among other things, the loop contributions (11).
This distinction requires a proper renormalization proce-
dure in order to formulate the model in a self-consistent
manner, see [8] for more details. Its neglect leads to a se-
vere violation of unitarity beyond pion threshold [11].
Next, we introduce the electromagnetic interaction. It
is done by using the canonical gauge invariance preserv-
ing method of minimal substitution, and typical proto-
types of resulting currents are depicted in g. 4. More
42 The European Physical Journal A
Fig. 4. Examples for current contributions in the one- and two-
nucleon sector: left: one-body current; middle: meson-exchange
current (MEC); right: electromagnetic loop correction.
of a spectator pion, and V N the N -interaction in the
Fig. 5. N -scattering in the P33 channel.
Fig. 6. Diagrammatic representation of the amplitude
T X (12).
details can be found in [9,10]. In practice, gauge invari-
ance (as well as unitarity) is not exactly fullled due to
some technical reasons: Whereas the Elster potential is
treated in a completely relativistic manner concerning the
vertices, the vertex structure in the corresponding MEC is
presently treated only nonrelativistically within a p/MN -
expansion. Moreover, MEC of at least fourth order in the
N -coupling constant are necessary to preserve gauge in-
variance exactly [9]. Their handling is technically very
complicated and, therefore, presently neglected. These vi-
olations of gauge invariance and unitarity occur fortu-
nately only at higher order in the 1/MN -expansion.
The model discussed so far is only suitable for ener-
gies below the pion threshold. In order to allow for higher
energies, nuclear resonances must necessarily be incorpo-
rated. In the present approach we restrict ourselves to the
which is again considered as a bare particle () with
vanishing decay width. Similar to [12], its coupling to the
N -system is generated by a suitable N -vertex V
whose parametrization is xed by studying N -scattering
in the P33 -channel, see g. 5. In a similar manner, the
electromagnetic transition N is xed once for all
by considering photopionproduction on the nucleon in the
M1+ (3/2)-multipole [9]. As next step, the has to be
introduced in the two-nucleon system. This is performed
nonperturbatively within a N N -N coupled-channel ap-
proach, see [8] for more details.
Last but not least, in addition the possibility of mutual
interactions within the N N -system needs to be consid-
ered. This can be tackled using standard three-body tech-
niques for the relevant amplitude T X depicted in g. 6.
NNinteraction FSI in pion production
X
Fig. 7. The amplitude T (12) contributes both to the N N -
interaction as well as to nal state interactions (FSI) in pion
production.
Neglecting three-body forces, T X has the form
 
T X (z) = V + V N + V + V N G0 (z)T X (z), (12)
where V describes the N N -interaction in the presence
presence of a spectator nucleon. As indicated in g. 7,
T X (z) contributes simultaneously to the N N -scattering
amplitude as well as to nal state interactions (FSI) in
pion production processes.
For technical reasons we intend to parametrize these
interactions in terms of suitable separable realizations [13,
14]. In the present realization, solely the so-called d-
channel, i.e. V in the 3 S1 /3 D1 N N -channel is consid-
ered [8]. A more complete treatment of T X (z) is under
construction. Moreover, the approach discussed so far is
presently only realized for N N -scattering [8] and elec-
tromagnetic deuteron breakup [9,10,15]. An extension to
photopionproduction as well as elastic d-scattering will
be available soon.
Our proposed model is denitely a very promising one
for studying simultaneously all possible hadronic and elec-
tromagnetic reactions up to the -region with at most one
asymptotic free pion. Concerning alternatives to our ap-
proach, we only mention here the presently most popular
one, namely eective eld theory (EFT) which is based on
the spontaneously broken chiral symmetry of QCD. EFT
starts from the most general eective Lagrangian which
is consistent with the symmetries of QCD and therefore
more involved than the Lagrangians used in our approach.
On the other hand one has to recognize that our approach
is nonperturbative whereas EFT performs a simultaneous
expansion in small external momenta and quark masses.
It is therefore a perturbative treatment in terms of an ex-
pansion parameter Q/ with Q m and 1 GeV,
the chiral symmetry breaking scale. In contrast to our ap-
proach, it is presently applicable only in quite a small
energy domain like N N -scattering up to pion threshold
(see [16,17] and references therein), low momentum elastic
electron deuteron scattering [18] or electropionproduction
near threshold [19].
2.3 Deuteron breakup in the -region
Next, we turn to the results of our approach for a se-
lected choice of reactions, starting with deuteron pho-
todisintegration. Despite its simplicity, this reaction has
posed severe problems for theoreticians until the middle
of the 90s. This becomes obvious from g. 8, where ex-
perimental data for the total cross section in the -region
 M. Schwamb: Few-nucleon systems (theory)
Fig. 8. The total cross section tot of deuteron photodisin-
tegration as a function of the photon energy. Results from
Tanabe and Ohta [20] (dotted), Wilhelm and Arenhovel [21]
(dashed) and Schwamb and Arenhovel [9, 10] (full). Experimen-
tal data from [22] (), [23] () and [24] ().
Fig. 9. Dierential cross sections of deuteron photodisintegra-
tion in the center-of-mass frame for a laboratory photon energy
of 260 MeV (left) and 440 MeV (right). Notation of the curves
as in g. 8.
is compared with the most sophisticated models available
at that time, namely the unitary three-body approach of
Tanabe and Ohta [20] as well as the model of Wilhelm and
Arenhovel [21]. Similar to our treatment, realistic N N -
interactions are used and a dynamical treatment of the
is incorporated. Moreover, a considerable conceptual im-
provement in comparison to earlier work was the fact that
no free parameters occur in the photodisintegration chan-
nel because similar as in our approach all of them
have been xed in advance by considering N - and N N -
scattering as well as photopionproduction on the nucleon.
From g. 8 it becomes obvious that the theory clearly fails
in describing the data. The predicted total cross sections
are too small and a dip structure around 90 occurs in
the dierential cross section at higher energies which is
not present in the data, see g. 9. These problems were
very severe ones because deuteron photodisintegration is
the simplest photonuclear reaction on a nucleus.
In the past decade, we have made considerable ef-
forts to solve this problem [7,9,10] reaching now an al-
most quantitative description of the total cross section in
the -region, see g. 8. Furthermore, also the description
of the dierential cross section is considerably improved.
43
This success turned out to be the combined result of var-
ious independent improvements compared to our start-
ing point [21]. Apart from the additional incorporation of
dissociation currents, the d-channel and conceptual im-
provements in the description of the N -transition,
retardation eects both in the hadronic interaction as well
as in the MEC turn out to be very important. The latter
have been partially neglected in [20,21] by using the static
Paris and Bonn-OBEPR potentials, respectively, and cor-
responding static MEC. This result clearly indicates that
even in breakup reactions of nuclei, where no asymptotic
free pions occur, the latter must be treated in a dynamic
manner for energies beyond pion threshold.
In a recent extension of this work, we have stud-
ied the role of retardation in deuteron electrodisintegra-
tion [15]. Neglecting polarization eects the dierential
cross section for this reaction in the one-photon-exchange
approximation is determined by four structure functions,
two diagonal ones fL and fT and two interference ones
fLT and fT T [25,26]. They are functions of the squared
three-momentum transfer q 2 , the nal state kinetic energy
Enp = W 2MN , and the angle between q and the pro-
ton momentum in the nal neutron-proton center-of-mass
system. It turns out that retardation leads to dramatic
changes in the structure functions fL and fLT for excita-
tion energies beyond the pion threshold whereas the other
structure functions fT and fT T are much less aected.
This is illustrated in g. 10 for a suitable kinematics in the
-region which has been studied at NIKHEF [27]. It turns
out that especially the recoil charge contribution (right
panel in g. 10) is very important. This mechanism is not
present in conventional static approaches due to an implic-
itly applied wave function renormalization procedure [28,
29] whose aim is to construct orthonormalized baryonic
wave functions. This concept breaks down beyond pion
threshold, where the pion can become onshell and must
be necessarily included in the hadronic wave functions.
This fact, already discussed in [9], clearly indicates that
a static treatment is only a poor approximation in reac-
tions on the deuteron beyond pion threshold. It would of
course be very important to perform experimental checks
of these predictions.
2.4 The deuteron as eective neutron target
The precise knowledge of elementary particle properties
is very important for a better understanding of their in-
ternal structure. With respect to the neutron as one of
the most important particles, its nite lifetime forces us
to consider few nucleon systems like the deuteron or 3 He
as alternative eective neutron targets. The basic ques-
tion is, whether for a specic neutron property of interest a
specic reaction on the deuteron exists where the neutron
contribution is dominant and nuclear background eects
from Fermi motion, MEC, FSI, etc. are small or at least
under control.
As a rst example, let us consider the neutron form
factors GEn and GM n . The magnetic form factor GM n
44 The European Physical Journal A

Fig. 10. Results for deuteron electrodisintegration taken from [15]. Left panel: the structure functions fL , fT , fLT and fT T
for the kinematics of the NIKHEF experiment [27], i.e. Enp = 280 MeV, q 2 = 2.47 fm 2 . Notation of the curves: dashed: static
approach; full: retarded approach. The additional dash-dotted curves represent the results of the retarded approach where the
Coulomb monopole contribution of the recoil charge operator, depicted on the right, is switched o.

}effect G En

Fig. 11. Relevant diagrams in deuteron electrodisintegration.

of the neutron can be determined from electron backscat-

tering o the deuteron in quasi-free neutron kinematics.
In this specic kinematics, the momentum of the virtual
photon in the laboratory frame is completely transferred
to the neutron whereas the spectator proton is at rest in
the nal state. These conditions lead to the rule of thumb
Enp /MeV = 10 q 2 /fm2 .
Compared to GM n , the electric neutron form factor
GEn is more di cult to measure. Various possibilities to
measure GEn have been discussed in [30]. It turned out
that the cleanest determination is obtained in double po-
larization observables in deuteron electrodisintegration,
i.e. d( e , e n)p or d ( e , e n)p. The relevant diagrams
contributing to this reaction are depicted in g. 11.
In the following, we restrict ourselves to the reaction
d( e , e n)p. In the Born approximation (PWBA), i.e. ne-
glecting FSI, MEC as well as isobars, and neglecting in
addition the D state of the deuteron, it turns out that in
quasi-free neutron kinematics the polarization component

Px in the scattering plane perpendicular to the photon mo-
mentum is directly proportional to GEn so that one has
a linear relation between the observable and the quantity
of interest:

Px GEn GM n . (13)

Fig. 12. The polarization Px of the outgoing neutron in the
scattering plane perpendicular to the photon momentum as a
function of the proton scattering angle for a squared photon
four-momentum of Q2 = 1 GeV2 , a squared three-momentum
transfer q 2 = 25.67 fm 2 and a kinetic energy of the outgoing
nucleons of Enp = 250 MeV. The neutron scattering angle n is
given by n = 180 . Notation of the curves: dotted: PWBA
with GEn = 0; dash-dotted: PWBA with GEn = 0; dashed: full
static calculation based on Bonn-OBEPR potential (GEn =
0); solid: full retarded calculation based on Elster potential
(GEn = 0). In quasi-free neutron kinematics ( = 180 ), one

readily recognizes the sensitivity of Px to GEn as well as its
insensitivity to nuclear structure eects like FSI, MEC and
resonance contributions.
Moreover, it turns out that in quasi-free neutron kinemat-
ics the role of the background eects of FSI, MEC and iso-
bars is under control and almost model independent, see
g. 12 as an illustrative example. This allows therefore a
very clean interpretation of the existing data (consider [31]
and the references therein), so that we can conclude that
the deuteron is a very e cient eective neutron target
with respect to the extraction of GEn .
 M. Schwamb: Few-nucleon systems (theory)
We now turn to a second example where the use of the
deuteron as an eective neutron target would be highly
desirable. It deals with the investigation of the Gerasimov-
Drell-Hearn sum rule (GDH) for various hadronic tar-
gets [32,33]. This sum rule links the anomalous magnetic
moment of a particle to the energy weighted integral over
the spin asymmetry of the absorption cross section. In de-
tail it reads for a particle of mass M , charge eQ, anoma-
lous magnetic moment and spin S
 d
d   P  
2 2 e
2 N
I GDH = ( ) A
( ) = 4 S,
0  M2
(14)
where P/A (  ) denote for a given photon momentum 
the total absorption cross sections for circularly polarized
photons on a target with spin parallel (P ) and antiparal-
lel (A) to the photon spin. This sum rule gives therefore a
very interesting relation between a ground state property
() of a particle and its whole excitation spectrum. Apart
from the general assumption that the integral in (14) con-
verges, its derivation is based solely on rst principles like
Lorentz and gauge invariance, unitarity, crossing symme-
try and causality of the Compton scattering amplitude of
a particle. Consequently, a check for various targets, both
from the experimental as well as from the theoretical point
of view, would be very important.
Inserting the known anomalous magnetic moments of
proton and neutron into (14), one obtains quite large GDH
sum rule values, i.e. IpGDH = 204.8 b for the proton and
InGDH = 233.2 b for the neutron. On the other side, the
deuteron has a small anomalous magnetic moment d =
0.143 n.m. resulting in a very small GDH sum rule value
of IpGDH = 0.65 b.
Whereas GDH measurements on proton targets can be
directly performed (consider [34] and references therein),
no free neutron target exists and one may try to extract
InGDH from deuteron measurements. In contrast to the
extraction of GEn , this task is however much more com-
plicated. First of all, let us recall that for the extraction
of the electric neutron form factor one specic reaction
(e.g. deuteron electrodisintegration) in one specic kine-
matics (the quasi-free one) is su cient. On the other hand,
concerning the GDH sum rule one has to determine total
inclusive cross sections, i.e. contributions in all possible
kinematics from very dierent reactions like
N N, N, N, . . . (15) by solving a system of coupled equations for the N N -
for the nucleon, and
d N N, N N, d, N N, N N, . . . (16)
for the deuteron have to taken into account.
These complications become even more serious if one
considers the sum of the proton and neutron value com-
pared to the deuteron value. If one assumes that the meson
production on the deuteron is dominated by the quasi-free
production on the nucleons bound in the deuteron, one
would expect that IdGDH should be roughly IpGDH +InGDH .
This assumption is however wrong by more than two or-
ders of magnitude. Consequently, concerning the GDH
45
t N
N
t N
N
d d T NN
N N
(a) (b)
N
t N
N
T
(c)
Fig. 13. Considered diagrams for single pion production.
(a) impulse approximation (IA), (b) incorporation of N N -nal
state interaction (N N -FSI), (c) incorporation of N -nal state
interaction (N -FSI).
sum rule the deuteron reaction cannot be considered just
as an incoherent sum of the proton and the neutron re-
action. In order to obtain the small deuteron GDH value,
strong anticorrelation eects between the dierent pos-
sible channels for the deuteron must occur which are not
present in the elementary case. This cancellation is a chal-
lenge for any theoretical framework since it requires the
above-mentioned unied consistent treatment of hadronic
and electromagnetic properties for the dierent possible
channels in a wide energy region.
In the past years, considerable eorts have been un-
dertaken in order to obtain a more quantitative under-
standing of the GDH sum rule on the deuteron [35,36,37].
In the presently most sophisticated approach [37], besides
deuteron photodisintegration also coherent and incoherent
single and double pion production as well as -production
are considered. At the moment, the aforementioned re-
tarded approach is only available for the breakup channel.
Concerning incoherent single pion production, the consid-
ered mechanisms in our present realization are depicted
in g. 13. For the elementary production operator, the
MAID model [38] is used, allowing one to extend the cal-
culation up to photon energies of 1.5 GeV. Moreover, nal
state interactions are perturbatively taken into account
up to the rst order in the corresponding N - and N N -
scattering amplitudes. For coherent pion production, the
model of [39] is used taking into account pion rescattering
, N - and N N -channels. It is partially similar to our
approach discussed in section 2.2. However, no retarda-
tion concerning the N N -interaction and the correspond-
ing MEC is presently taken into account. For double-pion
production the evaluation is based on a traditional eec-
tive Lagrangian approach similar to the one in [40]. It is
presented in great detail in [41].
Although this treatment of the GDH sum rule on the
deuteron is presently the most sophisticated one, we are
aware of specic shortcomings. The most serious one is the
use of dierent approaches for the dierent reactions. In
order to obtain a more unied picture, work is in progress
to adopt the discussed retarded approach not only to the
46 The European Physical Journal A

0
Bonn r-space
500
d 0np FSI eects are in general quite small. The same is true also
-100 retarded potential 400 d 0d
300
N 0N for -production. But FSI is nonnegligible for incoherent
() [b]

() [b]
-200
-300 d pn
200 neutral pion production due to the non-orthogonality of
the nal state wave in IA to the deuteron bound state wave
GDH

GDH
100
-400
Id

Id
0
-500
-100 d pn (x5)
function, see [42] for more details. Please note moreover
N N (x5)
-600 the signicant dierences between the deuteron and the
1 10 100 1000 500 1000 1500
[MeV] [MeV] corresponding nucleon values for I GDH (). This feature
0 20
occurs also in double-pion production where the largest
-20 0 contribution is coming from the + -channel. Here the
() [b]

() [b]

-40 -20 inclusion of FSI, where only N N -rescattering is presently

taken into account, is quite small.
GDH

GDH

-40
Id

Id

-60
d -pp (IA+FSI) d +nn (IA+FSI)
d -pp (IA)
-
n p
-60 d +nn (IA)
+
p n
The contributions of various channels to the nite
-80
500 1000 1500 500 1000 1500
GDH integral (17) for nucleon and deuteron are listed in
[MeV] [MeV]
20
table 1. While for the neutron the total sum is about 8
150 - +
d 0 0 np (IA+FSI) d np (IA+FSI) % lower than the sum rule value, it is too large by about
- +
d 0 0 np (IA) d np (IA)
- +
N 0 0 N N N
28 % for the proton. Concerning the deuteron, each of
() [b]

() [b]

100
10
the dierent channels (apart from -production) produces
GDH

GDH

50 very large contributions. Due to the large cancellation of

Id

Id

0 the photodisintegration and the meson production chan-

0
500 1000 1500 2000 500 1000 1500 2000 nels, the sum of all contributions is quite small (27.31 b).
[MeV] [MeV]
50 50 This is still somewhat too large compared to the theoret-
d 0 - pp (IA+FSI) d 0 + nn (IA+FSI)
40
-
d 0 pp (IA)
- 40
+
d 0 nn (IA)
+
ical value of 0.65 b. However, one should keep in mind
n 0 p p 0 n
that our approach still needs to be improved due to sev-
() [b]

() [b]

30 30

20 20
eral shortcomings as indicated above.
GDH

GDH

The strong cancellation between the regions at low

Id

Id

10 10

0 0 and high energies is a fascinating feature clearly demon-

500 1000 1500
[MeV]
2000
Fig. 14. Contributions of various channels to the nite GDH
integral (17) as a function of the upper integration limit
for deuteron disintegration, single- and double-pion and -
production on nucleon and deuteron. For the neutral charge
channels 0 , , 0 0 , and + , the nucleon integrals are the
sum of proton and neutron integrals. See [37] for more details.
breakup channel, but at least also to single-pion produc-
tion.
In order to present the results in a transparent way,
we introduce for convenience the nite GDH integral as
dened by
 ther progress in that eld.
d   P 
I GDH
() = 
( ) A (  ) , (17)
0
for which the results for photodisintegration, single and
double pion and -production are exhibited in g. 14.
With respect to the photodisintegration channel, at very
low energies a very large negative contribution arises from
the M 1 transition to the resonant 1 S0 state which can only
be reached if the spins of photon and deuteron are antipar-
allel. Sizeable dierences especially in the -region occur
between our retarded approach and an older static evalua-
tion [35] which was based on the Bonn-OBEPR potential.
Concerning single pion production, we show in g. 14
the results both in IA and with inclusion of nal state
interactions (labeled as IA+FSI) together with the corre-
sponding results for the elementary reactions. One notes
besides a positive contribution from the -resonance an-
other one above a photon energy of about 600 MeV from
D13 (1520) and F15 (1680). For charged pion production
500 1000 1500
[MeV]
2000 strating the decisive role of the pion as a manifestation
of chiral symmetry governing strong interaction dynam-
ics in these two dierent energy regions. With respect to
meson production channels on nucleon and deuteron, the
dierent behaviour of the corresponding spin asymmetries
indicates that a direct experimental access to the neutron
spin asymmetry from a deuteron measurement by sub-
tracting the one of the free proton is not possible. On
the other hand, the measurement of the spin asymmetry
for the dierent channels on the deuteron presents itself
a stringent test of our present theoretical understanding
of two-nucleon physics. Therefore, the experimental pro-
gram at facilities like MAMI and ELSA concerning the
GDH sum rule on the deuteron is very important for fur-
3 More complex few-nucleon systems
Till now, we have concentrated ourselves solely on the two-
nucleon system. The present situation in the three-nucleon
system is outlined in great detail in [43] and therefore
not discussed here. Concerning even more complex few-
nucleon systems, we want to present here merely some
recent highlights obtained with the Lorentz integral trans-
form method (LIT) [44]. The basic question in this context
is, up to which mass number A and energy/momentum
transfer precise microscopic calculations for the electro-
magnetic response can be performed. The most fundamen-
tal observable in this eld is denitely the total inclusive
cross section tot . In conventional scattering theory, an
economic method to calculate tot is to apply the optical
 M. Schwamb: Few-nucleon systems (theory) 47

Table 1. Contributions of various channels to the nite GDH integral (in b), integrated up to 0.8 GeV for photodisintegration,
1.5 GeV for single pion and -production and 2.2 GeV for double pion production on nucleon and deuteron, see [37] for further
details.

np Sum rule value
neutron 138.95 82.02 5.77 215.20 233.16
proton 176.38 93.93 8.77 261.54 204.78
deuteron 381.52 263.44 159.34 13.95 27.31 0.65

orem, the quantity  has to be treated as innitesimal

A A
...
Fig. 15. Top panel: diagrammatic representation of the di-
rect Compton scattering amplitude for an A-nucleon system.
Only one-body currents are depicted for the sake of simplicity.
Bottom panel: graphical illustration of a selected choice of con-
tributing mechanisms to the imaginary part of the Compton
scattering amplitude in (18).
theorem, here to Compton scattering (see g. 15)
tot (A X; W ) lim Im T (A A; W + i, = 0)
0
(18)
with W as invariant energy of the reaction. In order to ob-
tain the imaginary part, one has to know very precisely the
pole structure of the intermediate virtual states between
photon absorption and emission which requires a careful
numerical treatment of the occurring singularities. It is
obvious that with increasing mass number A and increas-
ing energy/momentum transfer this task becomes more
and more complicated and nally practically impossible.
An elegant solution of this problem has been proposed
about a decade ago by the Trento group [44]. The essential
idea is to perform rst of all an integral transform of tot
according to
 ular importance for testing present theoretical frameworks
tot (W )
L(R , I ) = dW (19)
(W )2
with = R + iI , where R , I can be treated as free
parameters. After some algebra, using the completeness
relation of the nal states, it turns out that L(R , I )
has the same formal structure as the optical theorem for
Compton scattering (18), i.e.
L(R , I ) Im T (A A; R + iI , = 0). (20)
The essential dierence between (18) and (20) lies in the
argument W + i versus R + iI . In the optical the-
small yielding in consequence the above-mentioned com-
plicated pole structure. Its counterpart in the LIT, I ,
is nite and at our disposal. It can, at least in principle,
be chosen arbitrarily. This has far reaching consequences,
because the pole structure in (19) vanishes for I nite.
This yields enormous numerical simplications, because
one needs only bound state techniques, avoiding in conse-
quence the calculation of A-body scattering states. In or-
der to obtain the desired inclusive cross section tot , one
has of course to perform a numerical inversion of the LIT.
Recently, a variety of dierent reliable inversion methods
has been presented [45] so that this problem is very well
under control. An important cross check for the inversion
is that the resulting cross section should be independent
of the parameter I so that the LIT method is completely
parameter free.
Due to these features, it is not very surprising that
the LIT has been applied with considerable success to mi-
croscopic calculations of quite a few electroweak cross sec-
tions of various nuclei ranging form A = 27 like inclusive
electron scattering (see e.g. [46,47]) and total photoab-
sorption cross sections (see e.g. [48,49,50]). In the mean-
time, it has also been extended to exclusive reactions [51,
52], photopionproduction on the deuteron [53,54] as well
as weak processes [55]. This list of applications shows that
the LIT approach constitutes an important progress open-
ing up the possibility to carry out ab initio microscopic
calculations not only for reactions on the classical few-
body systems (deuteron, three-body nuclei) but also for
reactions on more complex nuclei.
4 Summary and outlook
The study of reactions on few-nucleon systems is of partic-
in terms of eective degrees of freedom. Of specic interest
are electromagnetic reactions above pion threshold where
a unied approach needs to be constructed. Few-nucleon
systems are moreover of importance as eective neutron
targets, for example with respect to the extraction of the
electric neutron form factor GEn . The situation turns out
to be much more complicated with respect to the study of
the GDH sum rule on the neutron, where in contrast
to GEn no selection of the pure quasi-free kinematics is
possible and where many dierent reaction channels have
to be taken into account. Nevertheless, the planned mea-
surements of the GDH spin asymmetry on the deuteron
48 The European Physical Journal A

and 3 He at MAMI will lead to very stringent tests of our 20. H. Tanabe, K. Ohta, Phys. Rev. C 40, 1905 (1989).
present knowledge of nucleon and nuclear structure. Addi- 21. P. Wilhelm, H. Arenhovel, Phys. Lett. B 318, 410 (1993).
tional measurements are also desirable for electromagnetic 22. J. Arends et al., Nucl. Phys. A 412, 509 (1984).
reactions on more complex few-nucleon systems (A 4) 23. G. Blanpied et al., Phys. Rev. C 52, R455 (1995); Phys.
where nowadays for the rst time purely microscopic cal- Rev. C 61, 024604 (2000).
culations with the help of the Lorentz integral transform 24. R. Crawford et al., Nucl. Phys. A 603, 303 (1996).
method are possible. 25. W. Fabian, H. Arenhovel, Nucl. Phys. A 314, 253 (1979).
Summarizing, the study of few-nucleon systems is a 26. H. Arenhovel, W. Leidemann, E.L. Tomusiak, Phys. Rev.
very active eld both from the experimental as well as C 46, 455 (1992).
27. A. Pellegrino et al., Phys. Rev. Lett. 78, 4011 (1997).
theoretical point of view. The expected progress will be
28. M. Gari, H. Hyuga, Z. Phys. 277, 291 (1976).
very important for the future development of hadronic
29. H. Arenhovel, Czech. J. Phys. 43, 207 (1993).
physics in general. 30. H. Arenhovel, W. Leidemann, E.L. Tomusiak, Z. Phys. A
331, 123 (1988); 334, 363 (1989).
This is dedicated to the occasion of the retirement of 31. D.I. Glazier et al., Eur. Phys. J. A 24, 101 (2005).
H. Arenhovel, H. Backe, D. Drechsel, J. Friedrich, K-H. Kaiser 32. S.B. Gerasimov, Yad. Fiz. 2, 598 (1965) (Sov. J. Nucl.
and Th. Walcher. It has been supported by the Deutsche Phys. 2, 430 (1966)).
Forschungsgemeinschaft (SFB443). I would like to thank 33. S.D. Drell, A.C. Hearn, Phys. Rev. Lett. 16, 908 (1966).
H. Arenhovel for his careful reading of the manuscript and 34. H. Dutz et al., Phys. Rev. Lett. 93, 032003 (2004).
for various stimulating discussions. 35. H. Arenhovel, G. Kre , R. Schmidt, P. Wilhelm, Phys.
Lett. B 407, 1 (1997).
36. E.M. Darwish, H. Arenhovel, M. Schwamb, Eur. Phys. J.
A 17, 513 (2003).
References 37. H. Arenhovel, A. Fix, M. Schwamb, Phys. Rev. Lett. 93,
202301 (2004).
1. R. Machleidt, Adv. Nucl. Phys. 19, 189 (1989). 38. D. Drechsel, O. Hahnstein, S.S. Kamalow, L. Tiator, Nucl.
2. E.W. Schmid, H. Ziegelmann, The Quantum-Mechanical Phys. A 645, 145 (1999).
Three-Body Problem (Pergamon Press, Oxford and 39. P. Wilhelm, H. Arenhovel, Nucl. Phys. A 593, 435 (1995);
Friedrich Vieweg & Sohn, Braunschweig, 1974). 609, 469 (1996).
3. W. Glockle, The Quantum-Mechanical Few-Body Problem 40. J.A. Gomez Tejedor, E. Oset, Nucl. Phys. A 600, 413
(Springer Verlag, Berlin 1983). (1996).
4. R.B. Wiringa, V.G.J. Stoks, R. Schiavilla, Phys. Rev. C 41. A. Fix, H. Arenhovel, Eur. Phys. J. A 25, 115 (2005).
51, 38 (1995). 42. A. Fix, H. Arenhovel, Phys. Rev. C 72, 064005 (2005).
5. R. Machleidt, F. Sammarrucca, Y. Song, Phys. Rev. C 53, 43. J. Golak, R. Skibinski, H. Witala, W. Glockle, A. Nogga,
1483 (1996). H. Kamada, Phys. Rep. 415, 89 (2005).
6. R. Machleidt, Phys. Rev. C 63, 024001 (2001). 44. V.D. Efros, W. Leidemann, G. Orlandini, Phys. Lett. B
7. M. Schwamb, H. Arenhovel, P. Wilhelm, Th. Wilbois, 338, 130 (1994).
Phys. Lett. B 420, 255 (1998). 45. D. Andreasi, W. Leidemann, Ch. Reiss, M. Schwamb, Eur.
8. M. Schwamb, H. Arenhovel, Nucl. Phys. A 690, 647 Phys. J. A 24, 361 (2005).
(2001). 46. V.D. Efros, W. Leidemann, G. Orlandini, Phys. Rev. Lett.
9. M. Schwamb, H. Arenhovel, Nucl. Phys. A 690, 682 78, 432 (1997).
(2001). 47. V.D. Efros, W. Leidemann, G. Orlandini, E.L. Tomusiak,
10. M. Schwamb, H. Arenhovel, Nucl. Phys. A 696, 556 Phys. Rev. C 69, 044001 (2004).
(2001). 48. V.D. Efros, W. Leidemann, G. Orlandini, Phys. Rev. Lett.
11. Ch. Elster, W. Ferchlander, K. Holinde, D. Schutte, R. 78, 4015 (1997).
Machleidt, Phys. Rev. C 37, 1647 (1988). 49. S. Bacca, M. Marchisio, N. Barnea, W. Leidemann, G.
12. H. Popping, P.U. Sauer, X.-Z. Zhang, Nucl. Phys. A 474, Orlandini, Phys. Rev. Lett. 89, 052502 (2002).
557 (1987). 50. S. Bacca, H. Arenhovel, N. Barnea, W. Leidemann, G.
13. J. Haidenbauer, W. Plessas, Phys. Rev. C 30, 1822 (1984); Orlandini, Phys. Lett. B 603, 159 (2004).
Phys. Rev. C 32, 1424 (1985). 51. A. La Piana, W. Leidemann, Nucl. Phys. A 677, 423
14. S. Nozawa, B. Blankleider, T.-S.H. Lee, Nucl. Phys. A 513, (2000).
459 (1990). 52. S. Quaglioni, W. Leidemann, G. Orlandini, N. Barnea,
15. M. Schwamb, H. Arenhovel, Phys. Lett. B 588, 49 (2004). V.D. Efros, Phys. Rev. C 69, 044002 (2004).
16. R. Machleidt, D.R. Entem, J. Phys. G 31, S1235 (2005). 53. Ch. Reiss, W. Leidemann, G. Orlandini, E.L. Tomusiak,
17. E. Epelbaum, W. Glockle, U.-G. Meissner, Nucl. Phys. A Eur. Phys. J. A 17, 589 (2003).
747, 362 (2005). 54. Ch. Reiss, H. Arenhovel, M. Schwamb, Eur. Phys. J. A 25,
18. M. Walzl, U.-G. Meissner, Phys. Lett. B 513, 37 (2001). 171 (2005).
19. H. Krebs, V. Bernard, U.-G. Meissner, Eur. Phys. J. A 22, 55. D. Gazit, N. Barnea, Phys. Rev. C 70, 048801 (2004).
503 (2004).
Eur. Phys. J. A 28, s01, 49 57 (2006)
DOI: 10.1140/epja/i2006-09-006-5 EPJ A direct
electronic only

Nucleon form factors in dispersion theory

H.-W. Hammera
Helmholtz-Institut fur Strahlen- und Kernhysik (Theorie), Universitat Bonn, Nussallee 14-16, D-53115 Bonn, Germany

/
Published online: 11 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. Dispersion relations provide a powerful tool to analyse the electromagnetic form factors of the
nucleon in both the space-like and the time-like regions with constraints from other experiments, unitarity,
and perturbative QCD. We give a brief introduction into dispersion theory for nucleon form factors and
present rst results from our ongoing form factor analysis. We also calculate the two-pion continuum
contribution to the isovector spectral functions drawing upon the new high statistics measurements of the
pion form factor by the CMD-2, KLOE, and SND collaborations.
PACS. 11.55.Fv Dispersion relations 13.40.Gp Electromagnetic form factors 14.20.Dh Protons and
neutrons

1 Introduction the existence of the -resonance [15,16]. Despite this suc-

The electromagnetic form factors of the nucleon oer a
unique window on strong interaction dynamics over a wide
range of momentum transfers [1,2]. At small momentum
transfers, they are sensitive to the gross properties of the
nucleon like the charge and magnetic moment, while at
high momentum transfers they encode information on the
quark substructure of the nucleon as described by QCD.
Their detailed understanding is important for unravel-
ing aspects of perturbative and nonperturbative nucleon
structure. The form factors also contain important in-
formation on nucleon radii and vector meson coupling
constants. Moreover, they are an important ingredient in
a wide range of experiments from Lamb shift measure-
ments [3] to measurements of the strangeness content of
the nucleon [4].
With the advent of the new continuous beam elec-
tron accelerators such as CEBAF (Jeerson Lab.), ELSA
(Bonn), and MAMI (Mainz), a wealth of precise data for
space-like momentum transfers has become available [5].
Due to the di culty of the experiments, the time-like form
factors are less well known. While there is a fair amount
of information on the proton time-like form factors [6,7,
8,9,10], only one measurement of the neutron form factor
from the pioneering FENICE experiment [11] exists.
It has been known for a long time that the pion plays
an important role in the long-range structure of the nu-
cleon [12]. This connection was made more precise us-
ing dispersion theory in the 1950s [13,14]. Subsequently,
Frazer and Fulco have written down partial-wave dis-
persion relations that relate the nucleon electromagnetic
structure to pion-nucleon (N ) scattering and predicted
a
e-mail: hammer@phys.washington.edu
cess, the central role of the 2 continuum in the isovec-
tor spectral function has often been ignored. Hohler and
Pietarinen pointed out that this omission leads to a gross
underestimate of the isovector radii of the nucleon [17].
They rst performed a consistent dispersion analysis of the
electromagnetic form factors of the nucleon [18] including
the 2 continuum derived from the pion form factor and
N -scattering data [19]. In the mid-nineties, this analysis
has been updated by Mergell, Mei ner, and Drechsel [20]
and was later extended to include data in the time-like
region [21,22]. Recently, the new precise data for the neu-
tron electric form factor have been included as well [23].
Using chiral perturbation theory (ChPT), the long-
range pionic structure of the nucleon can be connected
to the Goldstone boson dynamics of QCD [24]. The non-
resonant part of the 2 continuum is in excellent agree-
ment with the phenomenological analysis [25] and the -
meson contribution can be included as well [26,27,28]. It
is well known that vector mesons play an important role
in the electromagnetic structure of the nucleon, see, e.g.,
refs. [15,29,30,31,32,33], and the remaining contributions
to the spectral function have usually been approximated
by vector meson resonances.
A new twist to this picture was recently given by Fried-
rich and Walcher [34]. They interpreted the form factor
data based on a phenomenological t with an ansatz for
the pion cloud using the idea that the proton can be
thought of as virtual neutron-positively charged pion pair.
A very long-range contribution to the charge distribution
in the Breit frame extending out to about 2 fm was found
and attributed to the pion cloud. This was shown to be
in conict with the phenomenologically known 2 contin-
uum and ChPT by Hammer, Drechsel, and Mei ner [35].
We will address this conundrum in more detail in sect. 9.
50 The European Physical Journal A
jem
p p
Fig. 1. The nucleon matrix element of the electromagnetic
current jem .
In this paper we give a brief introduction into disper-
sion theory for nucleon form factors and present prelim-
inary results from our ongoing form factor analysis. We
also calculate the two-pion continuum contribution to the
isovector spectral functions drawing upon the new high
statistics measurements of the pion form factor by the
CMD-2, KLOE, and SND collaborations. Finally we ad-
dress the question of the range of the pion cloud and give
an outlook on future work.
2 Denitions
The electromagnetic (em) structure of the nucleon is de-
termined by the matrix element of the current operator
jem between nucleon states as illustrated in g. 1.
Using Lorentz and gauge invariance, this matrix ele-
ment can be expressed in terms of two form factors,
 try. The matrix element can be expressed as
 em  F2 (t)
where F (t) is a generic form factor. In the case of the
electric and Dirac form factors of the neutron, GnE and
F1n , the expansion starts with the term linear in t and the
normalization factor F (0) is dropped.
3 Dispersion relations and spectral
decomposition
Based on unitarity and analyticity, dispersion relations
relate the real and imaginary parts of the electromagnetic
(em) nucleon form factors. Let F (t) be a generic symbol
for any one of the four independent nucleon form factors.
We write down an unsubtracted dispersion relation of the
form 
1 Im F (t )
F (t) = dt , (6)
t0 t t i
where t0 is the threshold of the lowest cut of F (t) (see
below) and the i denes the integral for values of t on the
cut.1 Equation (6) relates the em structure of the nucleon
to its absorptive behavior.
The imaginary part Im F entering eq. (6) can be ob-
tained from a spectral decomposition [13,14]. For this pur-
pose it is most convenient to consider the em current ma-
trix element in the time-like region (t > 0), which is re-
lated to the space-like region (t < 0) via crossing symme-
p |j |p = u(p ) F1 (t) + i
q u(p), (1)
2M

where M is the nucleon mass and t = (p p) the four- 2
momentum transfer. For data in the space-like region, it
is often convenient to use the variable Q2 = t > 0. The
functions F1 (t) and F2 (t) are the Dirac and Pauli form
factors, respectively. They are normalized at t = 0 as
F1p (0) = 1, F1n (0) = 0, F2p (0) = p , F2n (0) = n ,
with p = 1.79 and n = 1.91 the anomalous magnetic
moments of protons and neutrons in nuclear magnetons,
respectively.
J = N (p)N (p)|jem (0)|0 (7)
= u(p) F1 (t) + i
F2 (t)
(p + p) v(p),
2M
where p and p are the momenta of the nucleon and anti-
nucleon created by the current jem , respectively. The four-
momentum transfer in the time-like region is t = (p + p)2 .
(2)
Using the LSZ reduction formalism, the imaginary part
of the form factors is obtained by inserting a complete set
of intermediate states as [13,14]

It is convenient to work in the isospin basis and to Im J = (2)3/2 N p|JN (0)| (8)
decompose the form factors into isoscalar and isovector Z

parts, |jem (0)|0 v(p) 4 (p + p p ),
1 p 1 p
Fis = (F + Fin ), Fiv = (F Fin ), (3) where N is a nucleon spinor normalization factor, Z is
2 i 2 i
the nucleon wave function renormalization, and JN (x) =
where i = 1, 2.
J (x)0 with JN (x) a nucleon source. This decomposition
The experimental data are usually given for the Sachs
is illustrated in g. 2. It relates the spectral function to
form factors
on-shell matrix elements of other processes.
GE (t) = F1 (t) F2 (t), (4) The states | are asymptotic states of momentum p
GM (t) = F1 (t) + F2 (t), which are stable with respect to the strong interaction.
They must carry the same quantum numbers as the cur-
where = t/(4M 2 ). In the Breit frame, GE and GM may rent jem : I G (J P C ) = 0 (1 ) for the isoscalar current
be interpreted as the Fourier transforms of the charge and and I G (J P C ) = 1+ (1 ) for the isovector component of
magnetization distributions, respectively.
1
The nucleon radii can be dened from the low-t expan- The convergence of an unsubtracted dispersion relation for
sion of the form factors, the form factors has been assumed. We could have used a once

subtracted dispersion relation as well since the normalization
F (t) = F (0) 1 + tr2 /6 + . . . , (5) of the form factors is known.
 H.-W. Hammer: Nucleon form factors in dispersion theory
N 60
jem
N 20
Fig. 2. The spectral decomposition of the nucleon matrix el-
ement of the electromagnetic current jem .
jem . Furthermore, they have zero net baryon number. Be-
cause of G-parity, states with an odd number of pions only
contribute to the isoscalar part, while states with an even
number contribute to the isovector part. For the isoscalar
part the lowest mass states are: 3, 5, . . ., KK, KK, . . .;
for the isovector part they are: 2, 4, . . ..
Associated with each intermediate state is a cut start-
ing at the corresponding threshold in t and running to
innity. As a consequence, the spectral function Im F (t)
is dierent from zero along the cut from t0 to with
t0 = 4 (9) M2 for the isovector (isoscalar) case.
The spectral functions are the central quantities in
the dispersion-theoretical approach. Using eqs. (7,8), they
can in principle be constructed from experimental data.
In practice, this program can only be carried out for the
lightest two-particle intermediate states (2 and KK) [19,
36,37].
The longest-range, and therefore at low momentum
transfer most important pion cloud contribution comes
from the 2 intermediate state in the isovector form fac-
tors. A new calculation of this contribution will be dis-
cussed in the following section.
4 Two-pion continuum
In this section, we re-evaluate the 2 contribution in a
model-independent way [38] using the latest experimental
data for the pion form factor from CMD-2 [39], KLOE [40],
and SND [41].
We follow ref. [42] and express the 2 contribution
to the isovector spectral functions in terms of the pion
charge form factor F (t) and the P -wave N N am-
plitudes f1
(t). The 2 continuum is expected to be the
dominant contribution to the isovector spectral function
from threshold up to masses of about 1 GeV [42]. Here,
we use the expressions
qt3
Im GvE (t) = F (t) f+1
(t),
M t
q3
Im GvM (t) = t F (t) f
1 2
(t), (9)
2t
 also tabulated in ref. [42] contain assumptions about the pion
where qt = t/4 M2 . The imaginary parts of the Dirac
and Pauli Form factors can be obtained using eq. (4).
51
KLOE
CMD-2
50 SND
40
40
2
|F (t)|
30 30
0.5 0.6
10
0
0 0.2 0.4 0.6 0.8
2
t [GeV ]
Fig. 3. The pion electromagnetic form factor F (t) in the time-
like region as a function of the momentum transfer t. The dia-
monds, squares, and circles show the high statistics data from
the CMD-2 [39], KLOE [40], and SND [41] collaborations, re-
spectively. The dashed, solid, and dash-dotted lines are our
model parametrizations. The inset shows the discrepancy in
the resonance region in more detail.
The P -wave N N amplitudes f 1
(t) are tabulated
in ref. [42]. (See also ref. [43] for an unpublished update
that is consistent with ref. [42].) We stress that the rep-
resentation of eq. (9) gives the exact isovector spectral
functions for 4M2 t 16M2 , but in practice holds up
to t  50M2 . Since the contributions from 4 and higher
intermediate states is small up to t  50M2 , F (t) and
1
the f (t) share the same phase in this region and the two
quantities can be replaced by their absolute values.2
The updated pion form factor is shown in g. 3. The
diamonds, squares, and circles show the high statistics
data from the CMD-2 [39], KLOE [40], and SND [41]
collaborations, respectively. The dashed, solid, and dash-
dotted lines are our model parametrizations which are of
the Gounaris-Sakurai type [20,30]. The form factor shows
a pronounced - mixing in the vicinity of the -peak.
There are discrepancies between the three experimental
data sets for the pion form factor [41]. The discrepancies
in the -resonance region are shown in more detail in the
inset of g. 3. Since we are not in the position to settle this
experimental problem, we will take the three data sets at
face value. We will evaluate the 2 continuum given by
eq. (9) for all three sets and estimate the errors from the
discrepancy between the sets.
Using the new high statistics pion form factor data [39,
1
40,41] and the amplitudes f (t) tabulated in ref. [42],
we obtain the spectral functions shown in g. 4 [38]. We
show the spectral functions weighted by 1/t2 for GE (solid
We note that representation of eq. (9) is most useful for our
purpose. The manifestly real functions J (t) = f1 (t)/F (t)
form factor which leads to inconsistencies when used together
with the updated F (t).
52 The European Physical Journal A

0.06 Im F iS Im F iV
spectral function [1/M ]
4

2
2ImGE/t
S
2
2ImGM/t
0.04
S
t t

, KK
0.02
Fig. 5. Illustration of the spectral function used in the disper-
sion analysis. The vertical dashed line separates the well-known
low-mass contributions (2, KK, and continua as well as
0 the pole) from the eective poles at higher momentum trans-
0 20 40
2 fers.
t [M ]
Fig. 4. The 2 spectral function using the new high statistics 5 Spectral functions
data for the pion form factor [39, 40, 41]. The spectral functions
weighted by 1/t2 are shown for GE (solid line) and GM (dash- As discussed above the spectral function can at present
dotted line) in units of 1/M4 . The previous results by Hohler only be obtained from unitarity arguments for the lightest
et al. [42] (without - mixing) are shown for comparison by two-particle intermediate states (2 and KK) [19,36,37].
the green lines. The continuum contribution can be obtained from the
Bonn-Julich model [46].
The remaining contributions can be parametrized by
line) and GM (dash-dotted line). The previous results by
vector meson poles. On one hand, the lower mass poles can
Hohler et al. [42] (without - mixing) are given for com-
be identied with physical vector mesons such as the and
parison by the gray/green lines. The general structure of
the . In the the case of the 3 continuum, e.g., it has been
the two evaluations is the same, but there is a dierence
shown in ChPT that the nonresonant contribution is very
in magnitude of about 10%. The dierence between the
small and the spectral function is dominated by the [25].
three data sets for the pion form factor is very small and
The higher mass poles on the other hand, are simply an
indicated by the line thickness. The dierence in the form
eective way to parametrize higher mass strength in the
factors is largest in the -peak region (cf. g. 3), but this
spectral function.
region is suppressed by the N N amplitudes f 1
(t)
For our current best t, the spectral function includes
which show a strong fall-o as t increases.
the 2, KK, and continua from unitarity and the
The spectral functions have two distinct features.
pole. In addition to that there are a number of eective
First, as already pointed out in [15], they contain the
poles at higher momentum transfers in both the isoscalar
important contribution of the -meson with its peak at
and isovector channels. The spectral function then has the
t  30M2 . Second, on the left shoulder of the , the
general structure
isovector spectral functions display a very pronounced en-
hancement close to the two-pion threshold. This is due to
the logarithmic singularity on the second Riemann sheet Im Fis (t) = Im FiK K (t) + Im Fi (t)

located at tc = 4M2 M4 /M 2 = 3.98M2 , very close to + aVi (MV2 t), i = 1, 2, (11)
the threshold. This pole comes from the projection of the V =,s1 ,...
nucleon Born graphs, or in modern language, from the
Im Fiv (t) = Im Fi2 (t)
triangle diagram. 
If one were to neglect this important unitarity correc- + aVi (MV2 t), i = 1, 2, (12)
tion, one would severely underestimate the nucleon isovec- V =v1 ,...
tor radii [17],
 which is illustrated in in g. 5. The vertical dashed line
6 dt separates the well-known low-mass contributions to the
r2 vi = Im Gvi (t), (10)
4M2 t2
where i = E, M . In fact, precisely the same eect is ob-
tained at leading one-loop accuracy in relativistic chiral
perturbation theory [44,45]. This topic was also discussed
in heavy baryon ChPT [25,27] and in a covariant calcula-
tion based on infrared regularization [26]. Thus, the most
important 2 contribution to the nucleon form factors can
be determined by using either unitarity or ChPT (in the
latter case, of course, the contribution is not included).
spectral function from the eective poles at higher mo-
mentum transfers.
In our ts, we also include the widths of the vector
mesons. The width and mass of the are taken from the
particle data tables while the masses and widths of the
eective poles are tted to the form factor data. We have
performed various ts with dierent numbers of eective
poles and including/excluding some of the continuum con-
tributions. In sect. 7, we will discuss preliminary results
of this ongoing eort.
 H.-W. Hammer: Nucleon form factors in dispersion theory
6 Constraints
The number of parameters in the t function is reduced by
enforcing various constraints. The rst set of constraints
concerns the low-t behavior of the form factors. First,
we enforce the correct normalization of the form factors,
which is given in eq. (2). Second, we constrain the neutron
radius from a low-energy neutron-atom scattering experi-
ment [47,48].
Perturbative QCD (pQCD) constrains the behavior of
the nucleon em form factors for large momentum transfer.
Brodsky and Lepage [49] nd for t ,
  Q2
(i+1) t 2
Fi (t) (t) ln , i = 1, 2,
Q20
where Q0  QCD . The anomalous dimension depends
weakly on the number of avors, = 2.148, 2.160, 2.173
for Nf = 3, 4, 5, in order.
The power behavior of the form factors at large t can
be easily understood from perturbative gluon exchange.
In order to distribute the momentum transfer from the
virtual photon to all three quarks in the nucleon, at least
two massless gluons have to be exchanged. Since each of
the gluons has a propagator 1/t, the form factor has
to fall o as 1/t2 . In the case of F2 , there is additional
suppression by 1/t since a quark spin has to be ipped.
The power behavior of the form factors leads to supercon-
vergence relations of the form
 poles in order to improve the stability.
Im Fi (t) tn dt = 0,
t0
with n = 0 for F1 and n = 0, 1 for F2 . The asymptotic
behavior of eq. (13) is obtained by choosing the residues of
the vector meson pole terms such that the leading terms
in the 1/t-expansion cancel.
The logarithmic term in eq. (13) was included in some
of our earlier analyses [20,21,23] but has little impact
on the t. The particular way this constraint was imple-
mented, however, lead to an unphysical logarithmic singu-
larity of the form factors in the time-like region. In order
to be able to include the data for the form factors at large
time-like momentum transfers, the logarithmic constraint
is not enforced in the current analysis.
The number of eective poles in eqs. (11, 12) is deter-
mined by the stability criterion discussed in detail in [50].
In short, we take the minimum number of poles necessary
to t the data. For the preliminary results discussed in the
next section, we took 4 eective isoscalar poles and 3 ef-
fective isovector poles whose residua, masses, and widths
are tted to the data. The number of free parameters is
strongly reduced by the various constraints (unitarity, nor-
malizations, superconvergence relations), so that we end
up with 19 free parameters in the preliminary t presented
in the next section. Our general strategy is to reduce the
number of parameters even further without sacricing the
quality of the t.
53
7 Fit results
We now discuss some preliminary t results that are repre-
sentative for the current status of the analysis. We present
results for a t with 4 eective isoscalar poles and 3 eec-
tive isovector poles whose residua, masses, and widths are
tted to the data.
In g. 6, we show the results for all four form factors
compared to the world data for the form factors. Our data
basis is taken from ref. [34] and in addition also includes
the new data that have appeared since 2003 (see ref. [5]).
The results for GnM , GpE , GpM are normalized to the phe-
nomenological dipole t:
 2
GD (Q ) = 1 + 2 , (15)
(13) mD
where m2D = 0.71 GeV2 . The dash-dotted line gives the
result of ref. [23], while the the solid line indicates our
present best t. The new t leads to an improved descrip-
tion of the form factor data compared with ref. [23]. In
particular, the rapid fall-o of the JLab polarization data
for GpE [51,52] is now described. The 2 per degree of
freedom is 0.84. Note that we do not obtain a pronounced
bump structure in GnE as observed in ref. [34]. We will
come back to this question in sect. 9 and discuss the mod-
ications in the spectral function required to produce this
structure.
The stability constraint requires to use the minimum
number of poles required to describe the data [50]. In the
future, we plan to further reduce the number of eective
(14)
Table 1. Nucleon radii in fm extracted from the t in g. 6.
This work Ref. [23] Recent determ.
p
rE [fm] 0.84...0.857 0.848 0.886(15) [53, 54, 55]
p
rM [fm] 0.85...0.875 0.857 0.855(35) [54, 56]
n
rE [fm] 0.12...0.10 0.12 0.115(4) [48]
n
rM [fm] 0.86...0.88 0.879 0.873(11) [57]
In Table 1, we give the nucleon radii extracted from
our t. The neutron radius is included as a soft constraint
in our t and therefore not a prediction.3 The other nu-
cleon radii are generally in good agreement with other
recent determinations using only low-momentum-transfer
data given in the table. Our result for the proton radius,
however, is somewhat small. This was already the case in
the dispersion analyses of refs. [20,23]. We speculate that
the reason for this discrepancy lies in inconsistencies in the
data sets. In this type of global analysis all four form fac-
tors are analyzed simultaneously and both data at small
and large momentum transfers enter. This can be an ad-
vantage or a disadvantage depending on the question at
hand. Another possible reason for the discrepancy is 2
physics which was neglected in the data analysis of most
older experiments [58].
3
A soft constraint is not implemented exactly but deviations
from the constraint are penalized in the 2 of the t.
54 The European Physical Journal A

0.1 1.2

0.08

GM /(nGD)
1
0.06
n
GE

n
0.04
0.8
0.02

0 0.6
0 0.5 1 1.5 2 0.01 0.1 1 10
2 2 2 2
Q [GeV ] Q [GeV ]
1.5 1.2
GM /(pGD)
1 1
GE /GD
p

0.5 0.8

0 0.6
0.01 0.1 1 10 0.01 0.1 1 10 100
2 2 2 2
Q [GeV ] Q [GeV ]
p p
Fig. 6. The nucleon electromagnetic form factors for space-like momentum transfer. The results for G n M , GE , GM are normalized
to the dipole t. The dash-dotted line gives the result of ref. [23], while the the solid line indicates our preliminary best t.

8 Time-like data
We have also performed rst ts that include data in
the time-like region. The extraction of these data is more
challenging than in the space-like region. At the nucleon-
antinucleon threshold, the electric and magnetic form fac-
tors are equal by denition: GM (4M 2 ) = GE (4M 2 ), while
one expects the magnetic form factor to dominate at large
momentum transfer. Moreover, the form factors are com-
plex in the time-like region, since several physical thresh-
olds are open. Separating |GM | and |GE | unambiguously
from the data requires a measurement of the angular dis-
tribution, which is di cult. In most experiments, it has
been assumed that either |GM | = |GE | (which should be
a good approximation close to the two-nucleon threshold)
or |GE | = 0 (which should be a good approximation for
large momentum transfers). Most recent data have been
presented using the latter hypothesis.
The time-like data were already included in the disper-
sion analyses of refs. [21,22]. The proton magnetic form
factor up to t 6 GeV2 was well described by these anal-
yses. Data at higher momentum transfers were not in-
cluded. The data for the neutron magnetic form factor are
from the pioneering FENICE experiment [11]. They have
been analyzed under both the assumption |GE | = |GM |
and |GE | = 0. The latter hypothesis is favored by the mea-
sured angular distributions [11]. Neither data set could be
described by the analysis [22].
In g. 7, we show the current status of the analysis of
the time-like data for the magnetic form factors. For the
proton magnetic form factor, data up to momentum trans-
fers t 15 GeV2 have been included [6,7,8,9,10]. Our pre-
liminary t gives a good description in the threshold re-
gion but starts to deviate signicantly around t 5 GeV 2 .
The data for t 10 GeV2 are well described. This seems
to be due to a slight inconsistency in the data around
5 GeV2 and for t 10 GeV2 . This question deserves fur-
ther attention.
The status for the neutron form factor is the same as
in the previous analysis [22]: Neither of the two data sets
from ref. [11] can be described. Even though we are not
yet in the region where perturbative QCD is applicable, it
comes as a surprise that the neutron form factor is larger
in magnitude than the proton one. Perturbative QCD pre-
dicts asymptotically equal magnitudes. In any case, there
is interesting physics in the time-like nucleon form factors
 H.-W. Hammer: Nucleon form factors in dispersion theory
p 1.2
|GM (t)|
0.4
0.2
0
5 10
2 0.0
t [GeV ]
0.8
n Fig. 8. Pion cloud contribution to the nucleon charge density.
|GM (t)|
0.6
0.4
0.2
0
3 3.5 4 4.5 5 5.5
2 tained from experimental data and unitarity as discussed
t [GeV ]
Fig. 7. Current status of our analysis of the magnetic form
factors in the time-like region compared to the world data [6,
7, 8, 9, 10, 11]. The solid line gives our preliminary best t, while
the vertical dotted line indicates the two-nucleon threshold.
and new precision experiments such as the PANDA and
PAX experiments at GSI would be very welcome.
9 Pion cloud of the nucleon
Friedrich and Walcher (FW) recently analysed the em nu-
cleon form factors and performed various phenomenolog-
ical ts [34]. Their ts showed a pronounced bump struc-
ture in GnE which they interpreted using an ansatz for
the pion cloud based on the idea that the proton can be
thought of as a virtual neutron-positively charged pion
pair. They found a very long-range contribution to the
charge distribution in the Breit frame extending out to
about 2 fm which they attributed to the pion cloud. While
naively the pion Compton wave length is of this size, these
ndings are indeed surprising if compared with the pion
cloud contribution due to the 2 continuum contribution
to the isovector spectral functions discussed in sect. 4.
As was shown by Hammer, Drechsel, and Mei ner [35],
these latter contributions to the long-range part of the
nucleon structure are much more conned in coordinate
space and agree well with earlier (but less systematic) cal-
culations based on chiral soliton models, see, e.g., [59].
In the dispersion-theoretical framework, the longest-range
part of the pion cloud contribution to the nucleon form
55
v
GM
4r (r) [1/fm]
v
0.8 GE
2
0.4
15
0.0 0.5 1.0 1.5 2.0
r [fm]
The lines show the result of Friedrich and Walcher [34], while
the bands give the result of ref. [35]. Only the long-range con-
tributions for r > 1 fm are meaningful for the comparison of
the two results.
factors is given by the 2 continuum the lowest-mass
intermediate state including only pions. Note that a one-
pion intermediate state is forbidden by parity.
The nonresonant part of the 2 continuum can be cal-
6 culated in ChPT [27] while the full continuum can be ob-
in sect. 4. The pion cloud corresponds to the nonres-
onant part of the 2 continuum excluding the . Conse-
quently, the contribution has to be subtracted from the
full 2 continuum.4 The error in this subtraction was es-
timated using three dierent methods for the separation
of the contributions [35].
The charge distribution can then be obtained from the
nonresonant part of the 2 continuum by Fourier trans-
formation. This leads to the relation:
 40M2
v 1 v er t
i (r) = dt Im Gi (t) , (16)
4 2 4M2 r
where i = E, M . The 2 contribution from t 40M2 is
small and can be neglected [35].
The result for the pion cloud contribution to the nu-
cleon charge density is shown in g. 8. The lines show
the result of FW [34], while the bands give the result of
ref. [35]. Only the long-range contributions for r > 1 fm
should be compared since the separation of the short-
range part into resonant and nonresonant contributions
is arbitrary. In comparison with ref. [34], the 2 contin-
uum contribution to the charge density is generally much
smaller at distances beyond 1 fm, e.g., by a factor of 3 for
vE (r) at r = 1.5 fm. We emphasize that this result is ob-
tained from independent physical information that deter-
mines the 2 continuum (pion form factor and N N
amplitudes, cf. sect. 4) and not from form factor ts.
As a consequence, it remains to be shown how the pro-
posed long-range pion cloud can be reconciled with what
4
Note that this separation is not unique. It is only meaning-
ful for the long-range part. The separation of the short-range
part is model- and even representation-dependent.
56 The European Physical Journal A
n 2
0.1
GE (Q )
0.05
0
0 0.2 0.4 0.6 0.8 1
2 2
Q [GeV ]
Fig. 9. The neutron electric form factor at low momentum
transfer: present t with additional low-mass strength (dashed
line) compared to the ts of Friedrich and Walcher [34] (double-
dash-dotted line). For comparison, the ts of sect. 7 (solid line)
and ref. [23] (dash-dotted line) are also shown.
is known from dispersion relations and ChPT. In order to
clarify this issue, we have performed various ts in order
to understand what structures in the spectral function are
required to reproduce the bump in GnE . We nd that the
structure can only be reproduced if additional low-mass
strength in the spectral function below t < 1 GeV2 is al-
lowed beyond the 2, KK, and continua and the
pole. In the ts of sects. 7 and 8 such strength was explic-
itly excluded.
In g. 9, we show the neutron electric form factor at
low momentum transfer. The t of FW [34] is given by
the double-dash-dotted line, while the present t with ad-
ditional low-mass strength is given by the dashed line.
For comparison, we show also the t of ref. [23] (dash-
dotted line) and the t from sect. 7 (solid line). The t
with additional low-mass strength shows a clear bump
structure around Q2 0.3 GeV2 . This structure requires
three additional low mass poles: two isoscalar poles at
Ms2 = 0.13 GeV2 , 0.54 GeV2 and one isovector pole at
Mv2 = 0.30 GeV2 . In principle, vector meson dominance
works well for t 1 GeV2 and one should be able to inter-
pret these poles as physical vector mesons. However, no
such vector mesons are known in this region. This raises
the question of whether the eective low-mass poles can
be interpreted as something else?
One possible solution would be to interpret the poles
as eective poles mimicking some continuum contribution.
It is interesting to note that the three low-mass poles hap-
pen to come out at the thresholds of the 3, 4 and 5
continua and are located in the correct isospin channel.
Maybe these higher-order pion continua are more impor-
tant than previously thought and have a threshold en-
hancement similar to the 2 continuum that is accounted
for by the eective poles?
Even though this scenario has a certain appeal, it ap-
pears unlikely given the current state of knowledge. In
ref. [25], the threshold behavior of the 3 continuum was
explicitly calculated in heavy baryon ChPT and no en-
hancement was found. Moreover, the inelasticity from four
pions in scattering and four-pion production in e+ e
annihilation at low momentum transfer are known to be
small [42,60,61].
10 Summary & outlook
Dispersion theory simultaneously describes all four nu-
cleon form factors over the whole range of momentum
transfers in both the space-like and time-like regions. It
allows for the inclusion of constraints from other physical
processes, unitarity, and ChPT and therefore is an ideal
tool to analyze the form factor data.
We have presented preliminary results for our new dis-
persion analysis that is currently carried out in Bonn. The
spectral function has been improved and contains the up-
dated 2 continuum [38], as well the KK [36,37] and
continua [46]. Our preliminary best t gives a consistent
description of the world data in the space-like region. The
understanding of the time-like form factors is more di -
cult and a future challenge for theorists and experimen-
talists alike.
As part of this ongoing theoretical program, many
things remain to be done: The stability constraint requires
to use the minimum number of poles. Our strategy for
the future is to successively reduce the number of poles
without sacrifying the quality of the t. Furthermore, the
description of the time-like data needs to be improved. In
previous experiments, the separation of GE and GM could
only be carried out under overly simplifying assumptions.
New data, such as planned for the PANDA and PAX ex-
periments at GSI, are therefore called for.
Other improvements concern the quantication of the-
oretical and systematic uncertainties in the analysis, the
inclusion of perturbative QCD corrections beyond super-
convergence (leading logarithms etc.), and the inclusion
of two-photon physics. The latter point might require to
analyze the cross section data directly. Last but not least,
the consequences of the new data for the strange vector
form factors of the nucleon need to be worked out.
This work was done in collaboration with M.B. Belushkin, D.
Drechsel, and Ulf-G. Mei ner. M.J. Ramsey-Musolf has con-
tributed in earlier stages of the project. The work was sup-
ported in part by the EU I3HP under contract number RII3-
CT-2004-506078 and the DFG through funds provided to the
SFB/TR 16 Subnuclear Structure of Matter and SFB 443
Many Body Structure of Strongly Interacting Systems . I
would like to thank Hartmuth Arenhovel, Hartmut Backe, Di-
eter Drechsel, Jorg Friedrich, Karl-Heinz Kaiser, and Thomas
Walcher for a very stimulating and enjoyable time in Mainz.
I have had many personal interactions with them through sci-
entic discussions and/or through lectures and seminars I at-
tended as a student. In particular, I want to thank my PhD
advisor Dieter Drechsel from whom I have learned much about
physics and research.
 H.-W. Hammer: Nucleon form factors in dispersion theory
References
1. H. Gao, Int. J. Mod. Phys. E 12, 1 (2003); 12, 567
(2003)(E) (arXiv:nucl-ex/0301002).
2. C.E. Hyde-Wright, K. de Jager, Annu. Rev. Nucl. Part.
Sci. 54, 217 (2004) (arXiv:nucl-ex/0507001).
3. Th. Udem et al., Phys. Rev. Lett. 79, 2646 (1997).
4. D.H. Beck, B.R. Holstein, Int. J. Mod. Phys. E 10, 1 (2001)
(arXiv:hep-ph/0102053).
5. M. Ostrick, these proceedings and references therein.
6. E835 Collaboration (M. Ambrogiani et al.), Phys. Rev. D
60, 032002 (1999).
7. BES Collaboration (M. Ablikim et al.), Phys. Lett. B 630,
14 (2005) (arXiv:hep-ex/0506059).
8. CLEO Collaboration (T.K. Pedlar et al.), Phys. Rev. Lett.
95, 261803 (2005) (arXiv:hep-ex/0510005).
9. BABAR Collaboration (B. Aubert et al.), Phys. Rev. D
73, 012005 (2006) (arXiv:hep-ex/0512023).
10. R. Baldini, E. Pasqualucci, in Chiral Dynamics: Theory
and Experiment, edited by A.M. Bernstein, B.R. Holstein,
Lect. Notes Phys., Vol. 452 (Springer, Heidelberg, 1995).
11. A. Antonelli et al., Nucl. Phys. B 517, 3 (1998).
12. H. Frohlich, W. Heitler, N. Kemmer, Proc. R. Soc. A 166,
155 (1938).
13. G.F. Chew, R. Karplus, S. Gasiorowicz, F. Zachariasen,
Phys. Rev. 110, 265 (1958).
14. P. Federbush, M.L. Goldberger, S.B. Treiman, Phys. Rev.
112, 642 (1958).
15. W.R. Frazer, J.R. Fulco, Phys. Rev. Lett. 2, 365 (1959).
16. W.R. Frazer, J.R. Fulco, Phys. Rev. 117, 1609 (1960).
17. G. Hohler, E. Pietarinen, Phys. Lett. B 53, 471 (1975).
18. G. Hohler et al., Nucl. Phys. B 114, 505 (1976).
19. G. Hohler, E. Pietarinen, Nucl. Phys. B 95, 210 (1975).
20. P. Mergell, U.-G. Mei ner, D. Drechsel, Nucl. Phys. A 596,
367 (1996) (arXiv:hep-ph/9506375).
21. H.-W. Hammer, U.-G. Mei ner, D. Drechsel, Phys. Lett.
B 385, 343 (1996) (arXiv:hep-ph/9604294).
22. H.-W. Hammer, in Proceedings of the e+ e Physics at In-
termediate Energies Conference, edited by Diego Bettoni,
eConf C010430, W08 (2001) (arXiv:hep-ph/0105337).
23. H.-W. Hammer, U.-G. Mei ner, Eur. Phys. J. A 20, 469
(2004) (arXiv:hep-ph/0312081).
24. V. Bernard, N. Kaiser, U.-G. Mei ner, Int. J. Mod. Phys.
E 4, 193 (1995) (arXiv:hep-ph/9501384).
25. V. Bernard, N. Kaiser, U.-G. Mei ner, Nucl. Phys. A 611,
429 (1996) (arXiv:hep-ph/9607428).
26. B. Kubis, U.-G. Mei ner, Nucl. Phys. A 679, 698 (2001)
(arXiv:hep-ph/0007056).
27. N. Kaiser, Phys. Rev. C 68, 025202 (2003) (arXiv:nucl-
th/0302072).
28. M.R. Schindler, J. Gegelia, S. Scherer, Eur. Phys. J. A 26,
1 (2005) (arXiv:nucl-th/0509005).
29. J.J. Sakurai, Ann. Phys. (NY) 11, 1 (1960).
30. G.J. Gounaris, J.J. Sakurai, Phys. Rev. Lett. 21, 244
(1968).
31. M. Gari, W. Krumpelmann, Z. Phys. A 322, 689 (1985).
32. E.L. Lomon, Phys. Rev. C 64, 035204 (2001) (arXiv:nucl-
th/0104039).
57
33. S. Dubnicka, A.Z. Dubnickova, P. Weisenpacher, J. Phys.
G 29, 405 (2003) (arXiv:hep-ph/0208051).
34. J. Friedrich, T. Walcher, Eur. Phys. J. A 17, 607 (2003)
(arXiv:hep-ph/0303054).
35. H.-W. Hammer, D. Drechsel, U.-G. Mei ner, Phys. Lett.
B 586, 291 (2004) (arXiv:hep-ph/0310240).
36. H.-W. Hammer, M.J. Ramsey-Musolf, Phys. Rev. C
60, 045205 (1999); 62, 049903 (2000)(E) (arXiv:hep-
ph/9812261).
37. H.-W. Hammer, M.J. Ramsey-Musolf, Phys. Rev. C
60, 045204 (1999); 62, 049902 (2000)(E) (arXiv:hep-
ph/9903367).
38. M.A. Belushkin, H.-W. Hammer, U.-G. Mei ner, Phys.
Lett. B 633, 507 (2006) (arXiv:hep-ph/0510382).
39. CMD-2 Collaboration (R.R. Akhmetshin et al.),
arXiv:hep-ex/9904027; Phys. Lett. B 527, 161 (2002)
(arXiv:hep-ex/0112031); 578, 285 (2004) (arXiv:hep-
ex/0308008).
40. KLOE Collaboration (A. Aloisio et al.), Phys. Lett. B 606,
12 (2005) (arXiv:hep-ex/0407048).
41. M.N. Achasov et al., J. Exp. Theor. Phys. 101, 1053 (2005)
(arXiv:hep-ex/0506076).
42. G. Hohler, Pion-Nucleon Scattering, Landolt-Bornstein
Vol. I/9b, edited by H. Schopper (Springer, Berlin, 1983).
43. E. Pietarinen, A calculation of N N amplitudes in
the pseudophysical region, University of Helsinki Preprint
Series in Theoretical Physics, HU-TFT-17-77, unpub-
lished.
44. J. Gasser, M.E. Sainio, A. Svarc, Nucl. Phys. B 307, 779
(1988).
45. U.-G. Mei ner, Int. J. Mod. Phys. E 1, 561 (1992).
46. U.-G. Mei ner, V. Mull, J. Speth, J.W. van Orden, Phys.
Lett. B 408, 381 (1997) (arXiv:hep-ph/9701296).
47. S. Kopecky et al., Phys. Rev. Lett. 74, 2427 (1995).
48. S. Kopecky, M. Krenn, P. Riehs, S. Steiner, J.A. Harvey,
N.W. Hill, M. Pernicka, Phys. Rev. C 56, 2229 (1997).
49. S.J. Brodsky, G.P. Lepage, Phys. Rev. D 22, 2157 (1980).
50. I. Sabba-Stefanescu, J. Math. Phys. 21, 175 (1980).
51. Jeerson Lab Hall A Collaboration (M.K. Jones et al.),
Phys. Rev. Lett. 84, 1398 (2000) (arXiv:nucl-ex/9910005).
52. Jeerson Lab Hall A Collaboration (O. Gayou et
al.), Phys. Rev. Lett. 88, 092301 (2002) (arXiv:nucl-
ex/0111010).
53. R. Rosenfelder, Phys. Lett. B 479, 381 (2000) (arXiv:nucl-
th/9912031).
54. I. Sick, Phys. Lett. B 576, 62 (2003) (arXiv:nucl-
ex/0310008).
55. K. Melnikov, T. van Ritbergen, Phys. Rev. Lett. 84, 1673
(2000) (arXiv:hep-ph/9911277).
56. I. Sick, private communication.
57. G. Kubon et al., Phys. Lett. B 524, 26 (2002) (arXiv:nucl-
ex/0107016).
58. P.A.M. Guichon, M. Vanderhaeghen, Phys. Rev. Lett. 91,
142303 (2003) (arXiv:hep-ph/0306007).
59. U.-G. Mei ner, Phys. Rep. 161, 213 (1988).
60. J. Gasser, U.-G. Mei ner, Nucl. Phys. B 357, 90 (1991).
61. G. Ecker, R. Unterdorfer, Eur. Phys. J. C 24, 535 (2002)
(arXiv:hep-ph/0203075).
Eur. Phys. J. A 28, s01, 59 70 (2006)
DOI: 10.1140/epja/i2006-09-007-4 EPJ A direct
electronic only

Chiral perturbation theory

Success and challenge
S. Scherera
Institut fur Kernphysik, Johannes Gutenberg-Universitat Mainz, J.J. Becher Weg 45, D-55099 Mainz, Germany

/
Published online: 12 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. Chiral perturbation theory is the eective eld theory of the strong interactions at low energies.
We will give a short introduction to chiral perturbation theory for mesons and will discuss, as an example,
the electromagnetic polarizabilities of the pion. These have recently been extracted from an experiment
on radiative + photoproduction from the proton (p + n) at the Mainz Microtron MAMI. Next we
will turn to the one-baryon sector of chiral perturbation theory and will address the issue of a consistent
power counting scheme. As examples of the heavy-baryon framework we will comment on the extraction of
the axial radius from pion electroproduction and will discuss the generalized polarizabilities of the proton.
Finally, we will discuss two recently proposed manifestly Lorentz-invariant renormalization schemes and
illustrate their application in a calculation of the nucleon electromagnetic form factors.

PACS. 11.10.Gh Renormalization 11.30.Rd Chiral symmetries 13.40.-f Electromagnetic processes and
properties 13.40.Gp Electromagnetic form factors 13.60.Fz Elastic and Compton scattering 13.60.Le
Meson production

1 Introduction vanishing interactions in the zero-energy limit, a deriva-

Chiral perturbation theory (ChPT) [1,2,3,4] is the eec-
tive eld theory (EFT) [5] of the strong interactions at
low energies. The central idea of the EFT approach was
formulated by Weinberg as follows [1]: . . . if one writes
down the most general possible Lagrangian, including all
terms consistent with assumed symmetry principles, and
then calculates matrix elements with this Lagrangian to
any given order of perturbation theory, the result will sim-
ply be the most general possible S-matrix consistent with
analyticity, perturbative unitarity, cluster decomposition
and the assumed symmetry principles. In the context of
the strong interactions these ideas have rst been applied
to the interactions among the Goldstone bosons of spon-
taneous symmetry breaking in quantum chromodynamics
(QCD). The eective theory is formulated in terms of the
asymptotically observed states instead of the quark and
gluon degrees of freedom of the underlying (fundamental)
theory, namely QCD. The corresponding EFT mesonic
chiral perturbation theory has been tested at the two-
loop level (see, e.g., [6,7] for a pedagogical introduction).
A successful EFT program requires both the knowledge of
the most general Lagrangian up to and including the given
order one is interested in as well as an expansion scheme
for observables. Due to the vanishing of the Goldstone bo-
son masses in the chiral limit in combination with their
a
e-mail: scherer@kph.uni-mainz.de
tive and quark-mass expansion is a natural scenario for
the corresponding EFT. At present, in the mesonic sector
the Lagrangian is known up to and including O(q 6 ), where
q denotes a small quantity such as a four-momentum or a
pion mass. The combination of dimensional regularization
with the modied minimal subtraction scheme of ChPT [2]
leads to a straightforward correspondence between the
loop expansion and the chiral expansion in terms of mo-
menta and quark masses at a xed ratio, and provides a
consistent power counting for renormalized quantities.
In the extension to the one-nucleon sector [4] an ad-
ditional scale, namely the nucleon mass, enters the de-
scription. In contrast to the Goldstone boson masses, the
nucleon mass does not vanish in the chiral limit. As a re-
sult, the straightforward correspondence between the loop
expansion and the chiral expansion of the mesonic sec-
tor, at rst sight, seems to be lost: higher-loop diagrams
can contribute to terms as low as O(q 2 ) [4]. This problem
has been eluded in the framework of the heavy-baryon
formulation of ChPT [8,9], resulting in a power counting
analogous to the mesonic sector. The basic idea consists
in expressing the relativistic nucleon eld in terms of a
velocity-dependent eld, thus dividing nucleon momenta
into a large piece close to on-shell kinematics and a soft
residual contribution. Most of the calculations in the one-
baryon sector have been performed in this framework (for
an overview see, e.g., [10]) which essentially corresponds to
60 The European Physical Journal A
a simultaneous expansion of matrix elements in 1/m and
1/(4F ). However, there is price one pays when giving up
manifest Lorentz invariance of the Lagrangian. At higher
orders in the chiral expansion, the expressions due to the
1/m corrections of the Lagrangian become increasingly
complicated [11,12]. Moreover, not all of the scattering
amplitudes, evaluated perturbatively in the heavy-baryon
framework, show the correct analytical behavior in the
low-energy region [13]. In recent years, there has been
a considerable eort in devising renormalization schemes
leading to a simple and consistent power counting for the
renormalized diagrams of a manifestly Lorentz-invariant
approach [14,15,16,17,18,19,20,21].
In the following we will highlight a few topics in chiral
perturbation theory which have been subject of experi-
mental tests at the Mainz Microtron MAMI.
2 Chiral perturbation theory for mesons
2.1 The eective Lagrangian and Weinbergs power
counting scheme
The starting point of mesonic chiral perturbation theory
is a chiral SU (Nl )L SU (Nl )R U (1)V symmetry of the
QCD Lagrangian for Nl massless (light) quarks:

Nl
1
L0QCD = (qR,l iD / qL,l ) G,a Ga . (1)
/ qR,l + qL,l iD
l=1
4 = 2Bm = M 2 , where M 2 denotes the lowest-order ex-
is a result of the nite quark masses of the u, d and s
quarks. This explicit symmetry breaking in terms of the
quark masses is treated perturbatively.
The symmetries as well as the symmetry breaking pat-
tern of QCD once the quark masses are included are
mapped onto the most general (eective) Lagrangian for
the interaction of the Goldstone bosons. The Lagrangian
is organized in the number of the (covariant) derivatives
and of the quark mass terms [1,2,3,25,26,27,28,29,30,31]
L = L2 + L4 + L6 + , (2)
1
where the lowest-order Lagrangian is given by
F2  
L2 = Tr D U (D U ) + U + U . (3)
4
Here,
   0
+
U (x) = exp i , = 20 ,
F 2
is a unimodular unitary (2 2) matrix containing the
Goldstone boson elds. In eq. (3), F denotes the pion-
decay constant in the chiral limit: F = F [1 + O(m)] =
92.4 MeV. When including the electromagnetic interac-
tion, the covariant derivative is dened as D U = U +
ieA [Q, U ], where Q = diag(2/3, 1/3) denotes the quark
charge matrix. We work in the isospin-symmetric limit
mu = md = m. The quark masses are contained in
pression for the squared pion mass and B is related to the
In eq. (1), qL,l and qR,l denote the left- and right-handed quark condensate qq 0 in the chiral limit. The next-to-
components of the light quark elds. Here, we will be con- leading-order Lagrangian contains 7 low-energy constants
cerned with the cases Nl = 2 and Nl = 3 referring to li [2]
massless u and d or u, d and s quarks, respectively. Fur-

thermore, we will neglect the terms involving the heavy- 1 L R
L4 = l5 Tr(f U fL U ) Tr(f fL + f fR )
R
quark elds. The covariant derivative D qL/R,l contains 2
the avor-independent coupling to the eight gluon gauge l6  R
potentials, and G,a are the corresponding eld strengths. +i Tr f D U (D U ) + f
L
(D U ) D U + , (4)
2
The Lagrangian of eq. (1) is invariant under separate
global SU (Nl )L/R transformations of the left- and right- where we have displayed those terms which will be relevant
handed elds. In addition, it has an overall U (1)V sym- for the discussion of Compton scattering below. In that
metry. Several empirical facts give rise to the assumption case, the eld strength is given by
that this chiral symmetry is spontaneously broken down
to its vectorial subgroup SU (Nl )V U (1)V . For example,
R
f L
= f = e( A A )Q.
the low-energy hadron spectrum seems to follow multiplic- In addition to the most general Lagrangian, one needs
ities of the irreducible representations of the group SU (Nl ) a method to assess the importance of various diagrams cal-
(isospin SU (2) or avor SU (3), respectively) rather than culated from the eective Lagrangian. Using Weinbergs
SU (Nl )L SU (Nl )R , as indicated by the absence of de- power counting scheme [1] one may analyze the behavior
generate multiplets of opposite parity. Moreover, the light- of a given diagram calculated in the framework of eq. (2)
est mesons form a pseudoscalar octet with masses that under a linear re-scaling of all external momenta, pi  tpi ,
are considerably smaller than those of the corresponding and a quadratic re-scaling of the light quark masses,
vector mesons. According to Colemans theorem [22], the m  t2 m, which, in terms of the Goldstone boson masses,
symmetry pattern of the spectrum reects the invariance corresponds to M 2  t2 M 2 . The chiral dimension D of a
of the vacuum state. Therefore, as a result of Goldstones given diagram with amplitude M(pi , m) is dened by
theorem [23,24], one would expect 6 3 = 3 or 16 8 = 8
massless Goldstone bosons for Nl = 2 and Nl = 3, respec- M(tpi , t2 m) = tD M(pi , m), (5)
tively. These Goldstone bosons have vanishing interactions
1
as their energies tend to zero. Of course, in the real world, In the following, we will give equations for the two-avor
the pseudoscalar meson multiplet is not massless which case.
 S. Scherer: Chiral perturbation theory 61

where, in n dimensions,


D = nNL 2I +
2kN2k (6)
k=1
2

 Fig. 1. One-loop contribution to the pion self-energy. The
= 2 + (n 2)NL + 2(k 1)N2k

(7) number 2 in the interaction blob refers to L2 .
k=1
2 in 4 dimensions.

Here, NL is the number of independent loop momenta, I

the number of internal pion lines, and N2k the number of
vertices originating from L2k . A diagram with chiral di-
mension D is said to be of order O(q D ). Clearly, for small
enough momenta and masses diagrams with small D, such
as D = 2 or D = 4, should dominate. Of course, the re-
scaling of eq. (5) must be viewed as a mathematical tool.
While external three-momenta can, to a certain extent, be
made arbitrarily small, the re-scaling of the quark masses
is a theoretical instrument only. Note that, for n = 4,
loop diagrams are always suppressed due to the term
2NL in eq. (6). In other words, we have a perturbative
scheme in terms of external momenta and masses which
are small compared to some scale (here 4F 1 GeV).
Figures 1 and 2 show contributions to the pion self-
energy with D = 4 1 2 1 + 2 1 = 4 and D = 4 4 2
5 + 2 2 = 10, respectively. As a specic example, let us
consider the contribution of g. 1 to the pion self-energy.
Without going into the details, the explicit result of the
one-loop contribution is given by (see, e.g., [6])
4p2 M 2
loop (p2 ) = I (M 2 , 2 , n) = O(q 4 ),
6F 2
where the dimensionally regularized integral is given by
 2  coe cients lir as
M2 M
I (M 2 , 2 , n) = R + ln + O(n 4). (8)
16 2 2 R
In eq. (8), R is dened as
2
R= [ln(4) E + 1],
n4
with n denoting the number of space-time dimensions
and E =  (1) being Eulers constant. Note that
both factors the fraction and the integral each count as
O(q 2 ) resulting in O(q 4 ) for the total expression as antic-
ipated. In other words, when calculating one-loop graphs,
using vertices from L2 of eq. (3), one generates inni-
ties (so-called ultraviolet divergences). In the framework
of dimensional regularization these divergences appear as
poles at space-time dimension n = 4, since R is innite
as n 4. The loop diagrams are renormalized by absorb-
ing the innite parts into the redenition of the elds and
the parameters of the most general Lagrangian. Since L2
of eq. (3) is not renormalizable in the traditional sense,
2 2
Fig. 2. Four-loop contribution to the pion self-energy.
every one of the innite number of interactions allowed
by symmetries, the so-called non-renormalizable theories
are actually just as renormalizable as renormalizable the-
ories. According to Weinbergs power counting of eq. (6),
one-loop graphs with vertices from L2 are of O(q 4 ). The
conclusion is that one needs to adjust (renormalize) the
parameters of L4 to cancel one-loop innities. In doing
so, one still has the freedom of choosing a suitable renor-
malization condition. For example, in the minimal sub-
traction scheme (MS) one would x the parameters of the
counterterm Lagrangian such that they would precisely
absorb the contributions proportional to 2/(n 4). In the
modied minimal subtraction scheme of ChPT (MS)  em-
ployed in [2], the seven (bare) coe cients li of the O(q 4 )
Lagrangian of (4) are expressed in terms of renormalized
li = lir + i , (10)
32 2
where the i are xed numbers.
(9)
2.2 Electromagnetic polarizabilities of the pion
In the framework of classical electrodynamics, the elec-
tric and magnetic polarizabilities and describe the
response of a system to a static, uniform, external electric
and magnetic eld in terms of induced electric and mag-
netic dipole moments. In principle, empirical information
on the pion polarizabilities can be obtained from the dif-
ferential cross section of low-energy Compton scattering
on a charged pion
  2  2
d e2 e 1 + z2
=
dlab 4M 4M 2
the innities cannot be absorbed by a renormalization of  

the coe cients F and B. However, to quote from ref. [32]:

( + )+ (1 + z)2 + ( )+ (1 z)2
. . . the cancellation of ultraviolet divergences does not re- 2
ally depend on renormalizability; as long as we include + ,
62 The European Physical Journal A

where z = qq  and  / = [1+(1z)/M ]. The forward

and backward dierential cross sections are sensitive to
( + )+ and ( )+ , respectively.
The predictions for the charged pion polarizabilities at
O(q 4 ) [33] result from an old current-algebra low-energy +
+
theorem [34]

e2 1 l6 l 5
+ = + = 2
4 (4F )2 M 6 p n

Fig. 3. The reaction p n contains Compton scattering

+

= (2.64 0.09) 104 fm3 ,
which relates Compton scattering on a charged pion,
+ + , in terms of a chiral Ward identity to radiative
charged-pion beta decay, + e+ e . The linear combi-
nation l6 l5 of scale-independent low-energy constants [2]
is xed using the most recent determination of the ratio of
the pion axial-vector form factor FA and the vector form
factor FV via the radiative pion beta decay [35]:
1 FA
= (l6 l5 ) = = 0.443 0.015.
6 FV
A two-loop analysis (O(q 6 )) of the charged-pion polariz-
abilities has been worked out in [36,37]2 :
( + )+ = (0.3 0.1) 104 fm3 , (11)
( )+ = (4.4 1.0) 104 fm3 . (12)
on a pion as a sub diagram in the t channel, where t = (pn
pp )2 .
pion-nucleus bremsstrahlung, Z Z [41], radia-
tive pion photoproduction on the nucleon, p + n [42,
43], and pion pair production in e+ e scattering, e+ e
e+ e + [44,45,46,47]. The results of the older experi-
ments are summarized in table 1.
The potential of studying the inuence of the pion po-
larizabilities on radiative pion photoproduction from the
proton was extensively studied in [48]. In terms of Feyn-
man diagrams, the reaction p + n contains real
Compton scattering on a charged pion as a pion pole dia-
gram (see g. 3). In the recent experiment on p + n
at the Mainz Microtron MAMI [43], the cross section was
obtained in the kinematic region 537 MeV < E < 817
MeV, 140 cm
 180 . The values of the pion polar-

The degeneracy + = + is lifted at the two-loop level. izabilities have been obtained from a t of the cross sec-
The corresponding corrections amount to an 11% (22%) tion calculated by dierent theoretical models to the data
change of the O(q 4 ) result for + (+ ), indicating a simi- rather than performing an extrapolation to the t-channel
lar rate of convergence as for the -scattering lengths [2, pole of the Chew-Low type [49,50]. Figure 4 shows the
38]. The eect of the new low-energy constants appear- experimental data, averaged over the full photon beam
ing at O(q 6 ) on the pion polarizability was estimated via energy interval and over the squared pion-photon center-
resonance saturation by including vector and axial-vector of-mass energy s1 from 1.5 M2 to 5 M2 as a function of
mesons. The contribution was found to be about 50% of the squared pion momentum transfer t in units of M2 .
the two-loop result. However, one has to keep in mind For such small values of s1 , the dierential cross section
that [36,37] could not yet make use of the improved anal- is expected to be insensitive to the pion polarizabilities.
ysis of radiative pion decay which, in the meantime, has Also shown are two model calculations: model 1 (solid
also been evaluated at two-loop accuracy [39,40]. Taking curve) is a simple Born approximation using the pseu-
higher orders in the quark mass expansion into account, doscalar pion-nucleon interaction including the anomalous
Bijnens and Talavera obtain (l6 l5 ) = 2.98 0.33 [39], magnetic moments of the nucleon; model 2 (dashed curve)
which would slightly modify the leading-order prediction consists of pole terms without the anomalous magnetic
to + = (2.96 0.33) 104 fm3 instead of + = moments but including contributions from the resonances
(2.7 0.4) 104 fm3 used in [36,37]. Accordingly, the dif- (1232), P11 (1440), D13 (1520) and S11 (1535). The dotted
ference ( )+ of (12) would increase to 4.9 104 fm3 curve is a t to the experimental data.
instead of 4.4 104 fm3 , whereas the sum would remain
the same as in eq. (11).
As there is no stable pion target, empirical informa- Table 1. Previous experimental data on the charged pion po-
tion about the pion polarizabilities is not easy to obtain. larizability + .
For that purpose, one has to consider reactions which 4
contain the Compton scattering amplitude as a building Reaction Experiment + [10 fm3 ]
block, such as, e.g., the Primako eect in high-energy Z Z Serpukhov [41] 6.8 1.4 1.2
p + n Lebedev Phys. Inst. [42] 20 12
2
References [36, 37] use (l6 l5 ) = 2.7 0.4 instead of + PLUTO [44] 19.1 4.8 5.7
2.64 0.72 which was obtained in ref. [2] from = 0.44 0.12. DM 1 [45] 17.2 4.6
Correspondingly, this also generates a smaller error in the DM 2 [46] 26.3 7.4
O(q 4 ) prediction + = (2.7 0.4) 10 4 fm3 instead of MARK II [47] 2.2 1.6
(2.62 0.71) 10 4 fm3 .
 d/ds1dt (nb/ )
1.2
4
1 quacy of the present V A description of the radiative
0.8
0.6
0.4
0.2
0
-10 -9 -8 -7 -6 -5 -4
Fig. 4. Dierential cross section averaged over 537 MeV <
E < 817 MeV and 1.5 M2 < s1 < 5M2 . Solid line: model 1;
dashed line: model 2; dotted line: t to experimental data.
(nb)
14
12
10
8 + p p + , 107 etas will be produced per day. The
6
4
2 decay process
0 At O(q 4 ), the amplitude is given entirely in terms of one-
550 600 650 700
Fig. 5. The cross section of the process p + n integrated
over s1 and t in the region where the contribution of the pion
polarizability is biggest and the dierence between the predic-
tions of the theoretical models under consideration does not
exceed 3 %. The dashed and dashed-dotted lines are predic-
tions of model 1 and the solid and dotted lines of model 2 for
( )+ = 0 and ( )+ = 14 10 4 fm3 , respectively.
The kinematic region where the polarizability contri-
bution is biggest is given by 5M2 < s1 < 15M2 and
12M2 < t < 2M2 . Figure 5 shows the cross section as a
function of the beam energy integrated over s1 and t in this
second region. The dashed and solid lines (dashed-dotted
and dotted lines) refer to models 1 and 2, respectively,
each with ( )+ = 0 (( )+ = 14 104 fm3 ). By
comparing the experimental data of the 12 points with
the predictions of the models, the corresponding values of
( )+ for each data point have been determined in
combination with the corresponding statistical and sys-
tematic errors. The result extracted from the combined
analysis of the 12 data points reads [43]
( )+ = (11.6 1.5stat 3.0syst 0.5mod ) 104 fm3
(13)
and has to be compared with the ChPT result of, say,
(4.9 1.0) 104 fm3 which deviates by 2 standard devi-
ations from the experimental result. On the other hand,
the application of dispersion sum rules as performed in [51]
yields ( )+ = (10.3 1.9) 104 fm3 .
Both the precision measurement of radiative pion beta
decay [35] and of radiative pion photoproduction indicate
that further theoretical and experimental work is needed.
S. Scherer: Chiral perturbation theory 63
In particular, the analysis of ref. [35] suggests an inade-
beta decay, which would also re ect itself in an inade-
quacy of the ChPT description in its present form. What
remains to be understood is why the dispersion sum rules
give such a dramatically dierent result from the ChPT
calculation where the higher-order terms have been esti-
mated from resonance saturation by including vector and
axial-vector mesons. Clearly, the model-dependent input
-3 -2 -1 0 deserves further study. In this context, a full and con-
t/
2 sistent one-loop calculation of p + n including the
Delta resonance [52] would be desirable.
For a discussion of the so-called generalized pion po-
larizabilities see [53,54,55,56].
2.3 Future perspectives at MAMI
With the setup of the Crystal Ball detector, a dedicated
physics program will be possible at MAMI. In the reaction
main physics objectives will be the investigation of neutral
decay channels.
In the framework of SU (3)L SU (3)R symmetry the
0 is closely related to 0 0 .
750 800
loop diagrams involving vertices of O(q 2 ). The prediction
E (MeV)
for the decay width was found to be two orders of mag-
nitude smaller [57] than the measured value. The pion
loops are small due to approximate G-parity invariance
whereas the kaon loops are suppressed by the large kaon
mass in the propagator. Therefore, higher-order contri-
butions must play a dominant role in 0 . Even
at O(q 6 ) dierences of a factor of two are found for
the decay rate and spectrum [57,58,59,60,61,62,63,64]
although the most recent result for the decay width of
( 0 ) = (0.45 0.12) eV agrees with the orig-
inal prediction (0.42 0.20) eV of ref. [57]. The decay
0 0 0 is a sensitive test of isospin symmetry vi-
olation with the transition amplitude being proportional
to the light quark mass dierence (mu md ) [65,66].
Moreover, the electromagnetic interaction was shown to
produce only a small contribution [67]. As a nal exam-
ple for allowed decays we refer to the rare eta decay
0 0 [68,69]. On the other hand, in the forbidden
decays such as 0 0 and 4 0 one will investigate
(P, CP ) violation which may be connected to the so-called
term in QCD.
As a nal example we would like to point at the po-
tential of investigating the + + 0 amplitude in the
p n + 0 reaction. This would allow for an alternative
test of the Wess Zumino Witten action [70,71] in terms of
the F3 amplitude (see [72] for a recent overview).
3 Chiral perturbation theory for baryons
3.1 The power counting problem
The standard eective Lagrangian relevant to the single-
nucleon sector contains, in addition to eq. (2), the most
64 The European Physical Journal A

general N Lagrangian [4,11,12],

(1) (2)
LN = LN + LN + . (14)
Due to the additional spin degree of freedom LN contains
both odd and even powers in small quantities. In order
to illustrate the issue of power counting in the baryonic Fig. 6. One-loop contribution to the nucleon self-energy. The
(1)
sector, we consider the lowest-order N Lagrangian [4], number 1 in the interaction blobs refers to LN .
expressed in terms of bare elds and parameters denoted
by subscripts 0,
  nucleon lines, vertices originating from L2k , and vertices
(k)
(1) 1 gA 0 originating from LN , respectively.
LN = 0 i m0

5 0 0 + ,
a a
2 F0 According to eq. (17), one-loop calculations in
(15) the single-nucleon sector should start contributing at
where 0 and 0 denote a doublet and a triplet of bare O(q n1 ). For example, let us consider the one-loop con-
nucleon and pion elds, respectively. After renormaliza- tribution of g. 6 to the nucleon self-energy. According to
eq. (16), the renormalized result should be of the order
tion, m, gA , and F refer to the chiral limit of the physical
nucleon mass, the axial-vector coupling constant, and the D = n 1 2 1 1 1 + 1 2 = n 1. (18)
pion-decay constant, respectively.
In sect. 2.1 we saw that, in the purely mesonic sec- We will see below that the corresponding renormalization
tor, contributions of n-loop diagrams are at least of order scheme is more complicated than in the mesonic sector.
O(q 2n+2 ), i.e., they are suppressed by q 2n in comparison An explicit calculation yields [21]
with tree-level diagrams. An important ingredient in de-
2 
riving this result was the fact that we treated the squared 3gA 0
pion mass as a small quantity of order q 2 . Such an ap- loop = (p/ + m)IN + M 2 (p/ + m)IN (p, 0)
4F02
proach is motivated by the observation that the masses 
of the Goldstone bosons must vanish in the chiral limit. (p2 m2 )p/ 2
[(p m 2
+ M 2
)I N (p, 0) + I N I ] ,
In the framework of ordinary chiral perturbation theory 2p2
M2 m which translates into a momentum expansion of
observables at xed ratio m/p2 . On the other hand, there where the relevant loop integrals are dened as
is no reason to believe that the masses of hadrons other 
dn k i
than the Goldstone bosons should vanish or become small I = 4n , (19)
(2) k M 2 + i0+
n 2
in the chiral limit. In other words, the nucleon mass enter- 
ing the pion-nucleon Lagrangian of eq. (15) should not be 4n dn k i
treated as a small quantity of, say, order O(q). Naturally IN = , (20)
(2)n k 2 m2 + i0+
the question arises how all this aects the calculation of 
loop diagrams and the setup of a consistent power count- dn k i
IN (p, 0) = 4n
ing scheme. (2) [(k p) m2 + i0+ ]
n 2

Our goal is to propose a renormalization procedure 1

generating a power counting for tree-level and loop dia- . (21)
k 2 M 2 + i0+
grams of the (relativistic) EFT for baryons which is analo-
gous to that given in sect. 2.1 for mesons. Choosing a suit-  renormalization scheme of ChPT [2,4]
Applying the MS
able renormalization condition will allow us to apply the indicated by r one obtains
following power counting: a loop integration in n dimen-


sions counts as q n , pion and fermion propagators count 2
3gAr M2
as q 2 and q 1 , respectively, vertices derived from L2k loop =
r
(p/ + m) + = O(q 2 ),
(k)
4Fr2 16 2
and LN count as q 2k and q k , respectively. Here, q gener-
ically denotes a small expansion parameter such as, e.g., where M 2 is the lowest-order expression for the squared
the pion mass. In total this yields for the power D of a 
pion mass. In other words, the MS-renormalized result
diagram in the one-nucleon sector the standard formula does not produce the desired low-energy behavior of


 eq. (18). This nding has widely been interpreted as the
D = nNL 2I IN +
2kN2k + kNkN (16) absence of a systematic power counting in the relativistic
k=1 k=1 formulation of ChPT.

 

= 1+(n2)NL + 2(k1)N2k + (k1)NkN (17)
k=1 k=1 3.2 Heavy-baryon approach
1 in 4 dimensions,
One possibility of overcoming the problem of power count-

where NL , I , IN , N2k , and NkN denote the number of ing was provided in terms of heavy-baryon chiral per-
independent loop momenta, internal pion lines, internal turbation theory (HBChPT) [8,9] resulting in a power
 S. Scherer: Chiral perturbation theory 65

counting scheme which follows eqs. (16) and (17). The 2

2 2
basic idea consists in dividing nucleon momenta into a
large piece close to on-shell kinematics and a soft resid- 1 1 1 1 1
ual contribution: p = mv + kp , v 2 = 1, v 0 1 (often
v = (1, 0, 0, 0)). The relativistic nucleon eld is expressed
Fig. 7. One-loop contributions leading to a modication of the
in terms of velocity-dependent elds, ( )
k2 dependence of E0+ .
(x) = eimvx (Nv + Hv ),
with
At threshold, the center-of-mass transition current can be
+imvx 1 +imvx 1 parameterized in terms of two s-wave amplitudes E0+ and
Nv = e (1 + v/), Hv = e (1 v/).
2 2 L0+
Using the equation of motion for Hv , one can eliminate
Hv and obtain a Lagrangian for Nv which, to lowest order, 4W 
eM |thr = i E0+ (k 2 ) + i L0+ (k 2 ) ,
reads [9] mN

LN = Nv (iv D + gA Sv u)Nv + O(1/m).

(1)
where W is the total center-of-mass energy,  = kk
and =  .
The result of the heavy-baryon reduction is a 1/m ex- The contribution from pion loops (see g. 7) has been
pansion of the Lagrangian similar to a Foldy-Wouthuysen analyzed in [75] and leads to a modication of the k 2 de-
expansion with a power counting along eqs. (16) and (17). ()
pendence of the electric dipole amplitude E0+ [at O(q 3 )]

 
3.3 Pion electroproduction near threshold and the () egA k2 1 k2 2
E0+ (k 2 ) = 1+ v + + r
axial radius 8F 4m2N 2 6 A
 2  
As an example illustrating the strength of the EFT M2 k
+ f + , (24)
approach we consider pion electroproduction (k) + 8 2 F2 M2
N (pi ) i (q) + N (pf ) near threshold (for an overview,
see ref. [73]) and the extraction of the nucleon axial radius. where v = 3.706 is the isovector anomalous magnetic mo-
To that end we introduce the Green functions ment of the nucleon and rA is the axial radius. The rst
line corresponds to the traditional expression obtained in
MA,i = N (pf )|Ai (0)|N (pi ) , the framework of the partially conserved axial-vector cur-

rent hypothesis (see, e.g., [76]). The second line generates
M
JA,i = d 4 x eiqx N (pf )|T [J (0)Ai (x)] |N (pi ) , the modication
  2   

MJP,i = d 4 x eiqx N (pf )|T [J (0)Pi (x)] |N (pi ) , M2 k k2 12
f = 1 + .
8 2 F2 M2 128F2 2
where the subscripts A, J and P refer to axial-vector
current, electromagnetic current and pseudoscalar density The reaction p(e, e + )n has been measured at MAMI
and i refers to the ith isospin component of the axial- at an invariant mass of W = 1125 MeV (corresponding to
vector current or the pseudoscalar density, respectively. a pion center of mass momentum of |q | = 112 MeV) and
The so-called Adler-Gilman relation [74] provides the chi- photon four-momentum transfers of Q2 = 0.117, 0.195 and
ral Ward identity 0.273 GeV2 [77]. Using an eective-Lagrangian model and
a dipole form as an ansatz for the axial form factor GA ,
q M
JA,i = imMJP,i + 3ij MA,j (22) an axial mass of
relating the three Green functions. In the one-photon- MA = (1.077 0.039) GeV
exchange approximation, the invariant amplitude for pion
electroproduction can be written as Mi = ie Mi , was extracted which has to be compared with the average
where  = eu u/k 2 is the polarization vector of the of neutrino scattering experiments
virtual photon and Mi the transition-current matrix ele-
ment: MA = (1.026 0.021) GeV.
Mi = N (pf ), i (q)|J (0)|N (pi ) . (23)
The relation between the Adler-Gilman relation, eq. (22), Dening MA = MA + MA , the dierence between
and pion electroproduction is established in terms of the the two results can nicely be explained in terms
Lehmann-Symanzik-Zimmermann reduction formula, of the additional k 2 dependence of eq. (24) yielding
MA = 0.056 GeV. In the meantime, the experiment
m
Mi = i lim (q 2 M2 )MJP,i has been repeated including an additional value of Q2 =
M2 F q2 M2 0.058 GeV2 [78] and is currently being analyzed.
1 Recently, there have been claims that pion electropro-
= 2 lim (q 2 M2 )(3ij MA,j q M
JA,i ). duction data at threshold cannot be interpreted in terms
M F q2 M2
66 The European Physical Journal A
of GA [79]. However, as was shown in [80], using mini-
mal coupling alone does not respect the constraints due
to chiral symmetry. In the framework of the most general
Lagrangian, this can be seen by considering the b23 term
of the O(q 3 ) Lagrangian [11],
(3) 1 0.5
Le = b23 5 [D , f ] + (25)
2(4F )2
with
f = 2( a a ) + 2i ([v , a ] [v , a ])
i -1
+ [ , v v ] + .
F
The Lagrangian of eq. (25) is of a non-minimal type and
the three terms contribute to the axial-vector matrix ele-
ment, the JA Green function and pion electroproduction
relevant to the Adler-Gilman relation. As a result it was
conrmed that threshold pion electroproduction is indeed
a tool to obtain information on the axial form factor of
the nucleon (see [80] for details).
3.4 Virtual Compton scattering and generalized
polarizabilities
As a second example, let us discuss the application of
HBChPT to the calculation of the so-called generalized
polarizabilities [81,82]. The virtual Compton scattering
(VCS) amplitude TVCS is accessible in the reaction e p
e p. Model-independent predictions, based on Lorentz
invariance, gauge invariance, crossing symmetry, and the
discrete symmetries, have been derived in ref. [83]. Up
to and including terms of second order in the momenta
|q | and |q  | of the virtual initial and real nal photons,
the amplitude is completely specied in terms of quan-
tities which can be obtained from elastic electron-proton
scattering and real Compton scattering, namely mN , ,
2
GE , GM , rE , p and p . The generalized polarizabilities
(GPs) of ref. [82] result from an analysis of the resid-
ual piece in terms of electromagnetic multipoles. A re-
striction to the lowest-order, i.e. linear terms in  leads
to only electric and magnetic dipole radiation in the -
nal state. Parity and angular-momentum selection rules,
charge-conjugation symmetry, and particle crossing gen-
erate six independent GPs [82,84,85].
The rst results for the two structure functions PLL
PT T / and PLT at Q2 = 0.33 GeV2 were obtained from
a dedicated VCS experiment at MAMI [86]. Results at
higher four-momentum transfer squared Q2 = 0.92 and
Q2 = 1.76 GeV2 have been reported in ref. [87]. Additional
data are expected from MIT/Bates for Q2 = 0.05 GeV2
aiming at an extraction of the magnetic polarizability.
Moreover, data in the resonance region have been taken
at JLab for Q2 = 1 GeV2 [88] which have been analyzed
in the framework of the dispersion relation formalism of
ref. [89,90]. Table 2 shows the experimental results of [86]
in combination with various model calculations. Clearly,
the experimental precision of [86] already allows for a crit-
ical test of the dierent models. Within ChPT and the
3
HBChPT O(p ): Electric polarization in the nucleon induced by the field Ex
1.5
1
y (fm)
0
-0.5
-1.5
-1.5 -1 -0.5 0 0.5 1 1.5
x (fm)
Fig. 8. Scaled electric polarization r 3 i1 [10 3 fm3 ] [91]. The
applied electric eld points in the x-direction.
linear sigma model, the GPs are essentially due to pionic
degrees of freedom. Due to the small pion mass the ef-
fect in the spatial distributions extends to larger distances
(see also g. 9). On the other hand, the constituent quark
model and other phenomenological models involving Gau
or dipole form factors typically show a faster decrease in
the range Q2 < 1 GeV2 .
A covariant denition of the spin-averaged dipole po-
larizabilities has been proposed in ref. [55]. It was shown
that three generalized dipole polarizabilities are needed to
reconstruct spatial distributions. For example, if the nu-
cleon is exposed to a static and uniform external electric
eld E, an electric polarization P is generated which is
related to the density of the induced electric dipole mo-
ments,
Pi (r) = 4ij (r) Ej . (26)
The tensor ij (r), i.e. the density of the full electric po-
larizability of the system, can be expressed as [55]
ij (r) = L (r)ri rj + T (r)(ij ri rj )

3ri rj ij
+ [L (r ) T (r )] r2 dr ,
r3 r
where L (r) and T (r) are Fourier transforms of the gen-
eralized longitudinal and transverse electric polarizabili-
ties L (q) and T (q), respectively. In particular, it is im-
portant to realize that both longitudinal and transverse
polarizabilities are needed to fully recover the electric po-
larization P. Figure 8 shows the induced polarization in-
side a proton as calculated in the framework of HBChPT
at O(q 3 ) [91] and clearly shows that the polarization, in
general, does not point into the direction of the applied
electric eld.
Similar considerations apply to an external magnetic
eld. Since the magnetic induction is always transverse
(i.e., B = 0), it is su cient to consider ij (r) =
(r)ij [55]. The induced magnetization M is given in
terms of the density of the magnetic polarizability as
M(r) = 4(r)B (see g. 9).
 S. Scherer: Chiral perturbation theory 67

Table 2. Experimental results and theoretical predictions for the structure functions PLL PT T / and PLT at Q2 = 0.33 GeV2
and  = 0.62. makes use of symmetry under particle crossing and charge conjugation which is not a symmetry of the
nonrelativistic quark model.

PLL PT T / [GeV 2
] PLT [GeV 2
]
Experiment [86] 23.7 2.2stat. 4.3syst. 0.6syst.norm. 5.0 0.8stat. 1.4syst. 1.1syst.norm.
Linear sigma model [92] 11.5 0.0
Eective Lagrangian model [93] 5.9 1.9
HBChPT [94] 26.0 5.3
Nonrelativistic quark model [95] 19.2|14.9 3.2| 4.5

3 In the following we will concentrate on one of

several methods that have been suggested to obtain
p
(q2) (104 fm3)

a consistent power counting in a manifestly Lorentz-

2 invariant approach [14,15,16,17,18,19,20,21], namely, the
so-called extended on-mass-shell (EOMS) renormalization
scheme [21]. The central idea of the EOMS scheme con-
1 sists of performing additional subtractions beyond the MS
scheme. Since the terms violating the power counting are
analytic in small quantities, they can be absorbed by coun-
0 terterm contributions. Let us illustrate the approach in
0 0.2 0.4 0.6 terms of the integral
2 2 
q (GeV ) H(p2 , m2 ; n) =
dn k i
,
8 (2)n [(k p)2 m2 + i0+ ][k 2 + i0+ ]
fm )
2

6 p where = (p2 m2 )/m2 = O(q) is a small quantity.

We want the (renormalized) integral to be of the order
4

4 D = n 1 2 = n 3. Applying the dimensional counting

4r (r) (10

analysis of ref. [96] (for an illustration, see the appendix

2 of ref. [97]), the result of the integration is of the form [21]

0 H F (n, ) + n3 G(n, ),
2

-2 where F and G are hypergeometric functions and are an-

0 1 2 3
r (fm)
Fig. 9. Generalized magnetic polarizability (q 2 ) and density
of magnetic polarizability (r) for the proton. Dashed lines:
contribution of pion loops; solid lines: total contribution; dot-
ted lines: VMD predictions normalized to (0) [55].
3.5 Manifestly Lorentz-invariant baryon chiral
perturbation theory
Unfortunately, when considering higher orders in the chi-
4 alytic in for any n. Hence, the part containing G for
noninteger n is proportional to a noninteger power of
and satises the power counting. On the other hand F
violates the power counting. The crucial observation is
that the part proportional to F can be obtained by rst
expanding the integrand in small quantities and then per-
forming the integration for each term [96]. This observa-
tion suggests the following procedure: expand the inte-
grand in small quantities and subtract those (integrated)
terms whose order is smaller than suggested by the power
counting. In the present case, the subtraction term reads
 
dn k i 
H subtr = 
(2) [k 2p k + i0 ][k + i0 ] p2 =m2
n 2 + 2 +

ral expansion, the expressions due to 1/m corrections

of the Lagrangian become increasingly complicated. Sec-
ondly, not all of the scattering amplitudes, evaluated per-
turbatively in the heavy-baryon framework, show the cor-
rect analytical behavior in the low-energy region. Finally,
with an increasing complexity of processes, the use of com-
puter algebra systems becomes almost mandatory. The
relevant techniques have been developed for calculations
in the Standard Model and thus refer to loop integrals of
the manifestly Lorentz-invariant type.
and the renormalized integral is written as H R = H
H subtr = O(q) as n 4. In the infrared renormalization
(IR) scheme of Becher and Leutwyler [16], one would keep
the contribution proportional to G (with subtracted diver-
gences when n approaches 4) and completely drop the F
term.
Let us conclude this section with a few remarks. With
a suitable renormalization condition one can also obtain a
consistent power counting in manifestly Lorentz-invariant
68 The European Physical Journal A

baryon chiral perturbation theory including, e.g., vector

mesons [98] or the (1232) resonance [52] as explicit de-
grees of freedom. Secondly, the infrared regularization of
Becher and Leutwyler [16] may be formulated in a form
analogous to the EOMS renormalization [99]. Finally, us-
ing a toy model we have explicitly demonstrated the appli-
cation of both infrared and extended on-mass-shell renor-
malization schemes to multiloop diagrams by considering
as an example a two-loop self-energy diagram [97]. In both
cases the renormalized diagrams satisfy a straightforward
power counting.

3.6 Applications

The EOMS scheme has been applied in several calcula-

tions such as the chiral expansion of the nucleon mass, the
pion-nucleon sigma term, and the scalar form factor [100],
the masses of the ground-state baryon octet [101] and the
nucleon electromagnetic form factors [102,103].
As an example, let us here consider the electromag-
netic form factors of the nucleon which are dened via the
matrix element of the electromagnetic current operator as
Fig. 10. The Sachs form factors of the nucleon in manifestly
N (pf ) |J (0)| N (pi ) = Lorentz-invariant chiral perturbation theory at O(q 4 ) without

 vector mesons. Full lines: results in the extended on-mass-shell
N 2 i q N 2 scheme; dashed lines: results in infrared regularization. The
u(pf ) F1 (Q ) + F (Q ) u(pi ), N = p, n,
2mN 2 experimental data are taken from ref. [105].

where q = pf pi is the momentum transfer and Q2

q 2 = t 0. Figure 10 shows the results for the electric
and magnetic Sachs form factors GE = F1 Q2 /(4m2N )F2
and GM = F1 + F2 at O(q 4 ) in the momentum transfer
region 0 GeV2 Q2 0.4 GeV2 without explicit vector-
meson degrees of freedom [102]. The O(q 4 ) results only
provide a decent description up to Q2 = 0.1 GeV2 and
do not generate su cient curvature for larger values of
Q2 . The perturbation series converges, at best, slowly and
higher-order contributions must play an important role.
Including the vector-meson degrees of freedom along
the lines of refs. [98,99] generates the additional diagrams
of g. 11. The results for the Sachs form factors including
vector-meson degrees of freedom are shown in g. 12. As
expected on phenomenological grounds [104], the quan-
titative description of the data has improved consider-
ably for Q2 0.1 GeV2 . The small dierence between the
two renormalization schemes is due to the way how the
regular higher-order terms of loop integrals are treated.
Note that on an absolute scale the dierences between
the two schemes are comparable for both GpE and GnE .
Numerically, the results are similar to those of ref. [104].
Due to the renormalization condition, the contribution of
the vector-meson loop diagrams either vanishes (infrared
renormalization scheme) or turns out to be small (EOMS). Fig. 11. Feynman diagrams involving vector mesons (double
Thus, in hindsight our approach puts the traditional phe- lines) contributing to the electromagnetic form factors up to
nomenological vector-meson dominance model on a more and including O(q 4 ).
solid theoretical basis.
 S. Scherer: Chiral perturbation theory
[ ] [ ]
/
/
[ ] [ ]
Fig. 12. The Sachs form factors of the nucleon in manifestly
Lorentz-invariant chiral perturbation theory at O(q 4 ) including
vector mesons as explicit degrees of freedom. Full lines: results
in the extended on-mass-shell scheme; dashed lines: results in
infrared regularization. The experimental data are taken from
ref. [105].
4 Summary
Chiral perturbation theory is a cornerstone of our un-
derstanding of the strong interactions at low energies.
Mesonic chiral perturbation theory has been tremendously
successful and may be considered as a full-grown and
mature area of low-energy particle physics. The appar-
ent con ict between the determination of the O(q 4 ) low-
energy constants (l6 l5 ) from radiative pion beta decay,
on the one hand, and the polarizability measurement, on
the other hand, certainly requires additional work, in par-
ticular, from the theoretical side.
The impact on baryonic chiral perturbation theory
due to the investigation of electromagnetic reactions at
MAMI such as elastic electron-nucleon scattering, (vir-
tual) Compton scattering and the electromagnetic pro-
duction of pions cannot be overestimated. The possibility
of a consistent manifestly Lorentz-invariant approach in
combination with the rigorous inclusion of (axial-) vector-
meson degrees of freedom and of the (1232) resonance
open the door to an application of ChPT in an extended
kinematic region.
I would like to thank the organizers Hartmuth Arenhovel,
Hartmut Backe, Dieter Drechsel, Jorg Friedrich, Karl-Heinz
Kaiser and Thomas Walcher of the symposium 20 Years of
69
Physics at the Mainz Microtron MAMI and express my best
wishes for the future.
References
1. S. Weinberg, Physica A 96, 327 (1979).
2. J. Gasser, H. Leutwyler, Ann. Phys. (N.Y.) 158, 142
(1984).
3. J. Gasser, H. Leutwyler, Nucl. Phys. B 250, 465 (1985).
4. J. Gasser, M.E. Sainio, A. Svarc, Nucl. Phys. B 307, 779
(1988).
5. G. Ecker, arXiv:hep-ph/0507056.
6. S. Scherer, in Advances in Nuclear Physics, edited
by J.W. Negele, E.W. Vogt, Vol. 27 (Kluwer Aca-
demic/Plenum, New York, 2003) pp. 277-538.
7. S. Scherer, M.R. Schindler, arXiv:hep-ph/0505265.
8. E. Jenkins, A.V. Manohar, Phys. Lett. B 255, 558 (1991).
9. V. Bernard, N. Kaiser, J. Kambor, U.-G. Mei ner, Nucl.
Phys. B 388, 315 (1992).
10. V. Bernard, N. Kaiser, U.-G. Mei ner, Int. J. Mod. Phys.
E 4, 193 (1995).
11. G. Ecker, M. Mojzis, Phys. Lett. B 365, 312 (1996).
12. N. Fettes, U.-G. Mei ner, M. Mojzis, S. Steininger, Ann.
Phys. (N.Y.) 283, 273 (2000); 288, 249 (2001)(E).
13. V. Bernard, N. Kaiser, U.-G. Mei ner, Nucl. Phys. A 611,
429 (1996).
14. H.B. Tang, arXiv:hep-ph/9607436.
15. P.J. Ellis, H.B. Tang, Phys. Rev. C 57, 3356 (1998).
16. T. Becher, H. Leutwyler, Eur. Phys. J. C 9, 643 (1999).
17. M.F.M. Lutz, Nucl. Phys. A 677, 241 (2000).
18. J. Gegelia, G. Japaridze, Phys. Rev. D 60, 114038 (1999).
19. J. Gegelia, G. Japaridze, X.Q. Wang, J. Phys. G 29, 2303
(2003).
20. M.F.M. Lutz, E.E. Kolomeitsev, Nucl. Phys. A 700, 193
(2002).
21. T. Fuchs, J. Gegelia, G. Japaridze, S. Scherer, Phys. Rev.
D 68, 056005 (2003).
22. S. Coleman, J. Math. Phys. 7, 787 (1966).
23. J. Goldstone, Nuovo Cimento 19, 154 (1961).
24. J. Goldstone, A. Salam, S. Weinberg, Phys. Rev. 127,
965 (1962).
25. D. Issler, SLAC-PUB-4943-REV (1990) (unpublished).
26. R. Akhoury, A. Alfakih, Ann. Phys. (N.Y.) 210, 81
(1991).
27. S. Scherer, H.W. Fearing, Phys. Rev. D 52, 6445 (1995).
28. H.W. Fearing, S. Scherer, Phys. Rev. D 53, 315 (1996).
29. J. Bijnens, G. Colangelo, G. Ecker, J. High Energy Phys.
9902, 020 (1999).
30. T. Ebertshauser, H.W. Fearing, S. Scherer, Phys. Rev. D
65, 054033 (2002).
31. J. Bijnens, L. Girlanda, P. Talavera, Eur. Phys. J. C 23,
539 (2002).
32. S. Weinberg, The Quantum Theory of Fields, Vol. 1:
Foundations (Cambridge University Press, Cambridge
1995) Chap. 12.
33. J. Bijnens, F. Cornet, Nucl. Phys. B 296, 557 (1988).
34. M.V. Terentev, Sov. J. Nucl. Phys. 16, 87 (1973) (Yad.
Fiz. 16, 162 (1972)).
35. E. Frlez et al., Phys. Rev. Lett. 93, 181804 (2004).
36. U. Burgi, Phys. Lett. B 377, 147 (1996).
37. U. Burgi, Nucl. Phys. B 479, 392 (1996).
70 The European Physical Journal A
38. J. Bijnens, G. Colangelo, G. Ecker, J. Gasser, M.E.
Sainio, Phys. Lett. B 374, 210 (1996).
39. J. Bijnens, P. Talavera, Nucl. Phys. B 489, 387 (1997).
40. C.Q. Geng, I.L. Ho, T.H. Wu, Nucl. Phys. B 684, 281
(2004).
41. Y.M. Antipov et al., Phys. Lett. B 121, 445 (1983).
42. T.A. Aibergenov et al., Czech. J. Phys. B 36, 948 (1986).
43. J. Ahrens et al., Eur. Phys. J. A 23, 113 (2005).
44. PLUTO Collaboration (C. Berger et al.), Z. Phys. C 26,
199 (1984).
45. DM1 Collaboration (A. Courau et al.), Nucl. Phys. B
271, 1 (1986).
46. DM2 Collaboration (Z. Ajaltoni et al.), in Proceedings of
the VII International Workshop on Photon-Photon Col-
lisions, Paris, 1-5 April 1986, edited by A. Courau, P.
Kessler (World Scientic, Singapore, 1986).
47. MARK II Collaboration (J. Boyer et al.), Phys. Rev. D
42, 1350 (1990).
48. D. Drechsel, L.V. Filkov, Z. Phys. A 349, 177 (1994).
49. G.F. Chew, F.E. Low, Phys. Rev. 113, 1640 (1959).
50. C. Unkmeir, PhD Thesis, Johannes Gutenberg-
Universitat, Mainz (2000).
51. L.V. Filkov, V.L. Kashevarov, Eur. Phys. J. A 5, 285
(1999).
52. C. Hacker, N. Wies, J. Gegelia, S. Scherer, Phys. Rev. C
72, 055203 (2005).
53. C. Unkmeir, S. Scherer, A.I. Lvov, D. Drechsel, Phys.
Rev. D 61, 034002 (2000).
54. T. Fuchs, B. Pasquini, C. Unkmeir, S. Scherer, Czech. J.
Phys. 52, B135 (2002).
55. A.I. Lvov, S. Scherer, B. Pasquini, C. Unkmeir, D. Drech-
sel, Phys. Rev. C 64, 015203 (2001).
56. C. Unkmeir, A. Ocherashvili, T. Fuchs, M.A. Moinester,
S. Scherer, Phys. Rev. C 65, 015206 (2002).
57. L. Ametller, J. Bijnens, A. Bramon, F. Cornet, Phys.
Lett. B 276, 185 (1992).
58. P. Ko, Phys. Lett. B 349, 555 (1995).
59. S. Bellucci, C. Bruno, Nucl. Phys. B 452, 626 (1995).
60. A.A. Belkov, A.V. Lanyov, S. Scherer, J. Phys. G 22,
1383 (1996).
61. M. Jetter, Nucl. Phys. B 459, 283 (1996).
62. J. Bijnens, A. Fayyazuddin, J. Prades, Phys. Lett. B 379,
209 (1996).
63. E. Oset, J.R. Pelaez, L. Roca, Phys. Rev. D 67, 073013
(2003).
64. S. Prakhov et al., Phys. Rev. C 72, 025201 (2005).
65. J. Gasser, H. Leutwyler, Nucl. Phys. B 250, 539 (1985).
66. H. Leutwyler, Phys. Lett. B 374, 181 (1996).
67. R. Baur, J. Kambor, D. Wyler, Nucl. Phys. B 460, 127
(1996).
68. G. Knochlein, S. Scherer, D. Drechsel, Phys. Rev. D 53,
3634 (1996).
69. B.M.K. Nefkens et al., Phys. Rev. C 72, 035212 (2005).
70. J. Wess, B. Zumino, Phys. Lett. B 37, 95 (1971).
71. E. Witten, Nucl. Phys. B 223, 422 (1983).
72. I. Giller, A. Ocherashvili, T. Ebertshauser, M.A.
Moinester, S. Scherer, Eur. Phys. J. A 25, 229 (2005).
73. D. Drechsel, L. Tiator, J. Phys. G 18, 449 (1992).
74. S.L. Adler, F.J. Gilman, Phys. Rev. 152, 1460 (1966).
75. V. Bernard, N. Kaiser, U.G. Meissner, Phys. Rev. Lett.
69, 1877 (1992).
76. S. Scherer, J.H. Koch, Nucl. Phys. A 534, 461 (1991).
77. A1 Collaboration (A. Liesenfeld et al.), Phys. Lett. B
468, 20 (1999).
78. D. Baumann, PhD Thesis, Johannes Gutenberg-
Universitat, Mainz (2004).
79. H. Haberzettl, Phys. Rev. Lett. 85, 3576 (2000).
80. T. Fuchs, S. Scherer, Phys. Rev. C 68, 055501 (2003).
81. H. Arenhovel, D. Drechsel, Nucl. Phys. A 233, 153
(1974).
82. P.A.M. Guichon, G.Q. Liu, A.W. Thomas, Nucl. Phys. A
591, 606 (1995).
83. S. Scherer, A.Y. Korchin, J.H. Koch, Phys. Rev. C 54,
904 (1996).
84. D. Drechsel, G. Knochlein, A. Metz, S. Scherer, Phys.
Rev. C 55, 424 (1997).
85. D. Drechsel, G. Knochlein, A.Y. Korchin, A. Metz, S.
Scherer, Phys. Rev. C 57, 941 (1998).
86. VCS Collaboration (J. Roche et al.), Phys. Rev. Lett. 85,
708 (2000).
87. Jeerson Lab Hall A Collaboration (G. Laveissiere et al.),
Phys. Rev. Lett. 93, 122001 (2004).
88. H. Fonvieille, Prog. Part. Nucl. Phys. 55, 198 (2005).
89. B. Pasquini, M. Gorchtein, D. Drechsel, A. Metz, M. Van-
derhaeghen, Eur. Phys. J. A 11, 185 (2001).
90. D. Drechsel, B. Pasquini, M. Vanderhaeghen, Phys. Rept.
378, 99 (2003).
91. A.I. Lvov, S. Scherer, in preparation.
92. A. Metz, D. Drechsel, Z. Phys. A 356, 351 (1996).
93. M. Vanderhaeghen, Phys. Lett. B 368, 13 (1996).
94. T.R. Hemmert, B.R. Holstein, G. Knochlein, S. Scherer,
Phys. Rev. Lett. 79, 22 (1997).
95. B. Pasquini, S. Scherer, D. Drechsel, Phys. Rev. C 63,
025205 (2001).
96. J. Gegelia, G.S. Japaridze, K.S. Turashvili, Theor. Math.
Phys. 101, 1313 (1994) (Teor. Mat. Fiz. 101, 225 (1994)).
97. M.R. Schindler, J. Gegelia, S. Scherer, Nucl. Phys. B 682,
367 (2004).
98. T. Fuchs, M.R. Schindler, J. Gegelia, S. Scherer, Phys.
Lett. B 575, 11 (2003).
99. M.R. Schindler, J. Gegelia, S. Scherer, Phys. Lett. B 586,
258 (2004).
100. T. Fuchs, J. Gegelia, S. Scherer, Eur. Phys. J. A 19, 35
(2004).
101. B.C. Lehnhart, J. Gegelia, S. Scherer, J. Phys. G 31, 89
(2005).
102. T. Fuchs, J. Gegelia, S. Scherer, J. Phys. G 30, 1407
(2004).
103. M.R. Schindler, J. Gegelia, S. Scherer, Eur. Phys. J. A
26, 1 (2005).
104. B. Kubis, U.-G. Mei ner, Nucl. Phys. A 679, 698 (2001).
105. J. Friedrich, Th. Walcher, Eur. Phys. J. A 17, 607 (2003).
Eur. Phys. J. A 28, s01, 71 80 (2006)
DOI: 10.1140/epja/i2006-09-008-3 EPJ A direct
electronic only

Two-photon physics
M. Vanderhaeghena
Physics Department, The College of William & Mary, Williamsburg, VA 23187, USA and Theory Group, Jeerson Lab,
12000 Jeerson Ave, Newport News, VA 23606, USA

/
Published online: 15 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. It is reviewed how Compton scattering sum rules relate low-energy nucleon structure quantities
to the nucleon excitation spectrum. In particular, the GDH sum rule and recently proposed extensions of it
will be discussed. These extensions are sometimes more calculationally robust, which may be an advantage
when estimating the chiral extrapolations of lattice QCD results, such as for anomalous magnetic moments.
Subsequently, new developments in our description of the nucleon excitation spectrum will be discussed,
in particular a recently developed chiral eective eld theory framework for the (1232)-resonance region.
Within this framework, we discuss results on N and masses, the N transition and the magnetic
dipole moment.

PACS. 25.20.Dc Photon absorption and scattering 12.39.Fe Chiral Lagrangians 13.40.Gp Electromag-
netic form factors 13.40.Em Electric and magnetic moments

1 Introduction chiral behavior of the nucleon magnetic moment allows to

Sum rules for Compton scattering o a nucleon oer a
unique tool to relate low energy nucleon structure quan-
tities to the nucleon excitation spectrum [1]. E.g., the
Gerasimov, Drell, Hearn (GDH) sum rule (SR) [2] relates a
systems anomalous magnetic moment to a weighted inte-
gral over a combination of doubly polarized photoabsorp-
tion cross sections. Impressive experimental programs to
measure these photoabsorption cross-sections for the nu-
cleon have recently been carried out at ELSA and MAMI
(for a review see ref. [3]). Such measurements provide an
empirical test of the GDH SR, and can be used to gen-
erate phenomenological estimates of electromagnetic po-
larizabilities via related SRs. The GDH SR is particularly
interesting because both its left- and right-hand-sides can
be reliably determined, thus providing a useful verica-
tion of the fundamental principles (such as unitarity and
analyticity) which go into its derivation. At the present
time, it has been established that the proton sum rule is
satised within the experimental precision, while the case
is still out for the neutron.
After a lightning review of the GDH and related sum
rules in sect. 2, I discuss a recently proposed linearized ver-
sion of the GDH sum rule [4,5]. When applying this new
sum rule to the nucleon in the context of chiral perturba-
tion theory, it allows for an elementary calculation (to one
loop) of quantities such as magnetic moments and polariz-
abilities to all orders in the heavy-baryon expansion. The
a
e-mail: marcvdh@jlab.org
make a link with lattice QCD calculations.
Subsequently, the nucleon excitation spectrum is dis-
cussed in sect. 3. Many Compton scattering sum rules,
such as the GDH sum rule, are dominated by the (1232)
resonance. I discuss a recently proposed relativistic chi-
ral eective eld theory as a new systematic framework
to both extract resonance properties from the experiment
and to perform a chiral extrapolation of lattice QCD re-
sults for those resonance properties.
2 Sum rules in Compton scattering
2.1 Derivation of forward Compton scattering sum
rules
The forward-scattering amplitude describing the elastic
scattering of a photon on a target with spin s (real Comp-
ton scattering) is characterized by 2s + 1 scalar functions
which depend on a single kinematic variable, e.g., the pho-
ton energy . In the low-energy limit each of these func-
tions corresponds to an electromagnetic moment charge,
magnetic dipole, electric quadrupole, etc. of the target.
In the case of a spin-1/2 target, such as the nucleon, the
forward Compton amplitude is generally written as
T () =  f () + i (  ) g() , (1)
where ,  is the polarization vector of the incident and
scattered photon, respectively, while are the Pauli ma-
trices representing the dependence on the target spin. The
72 The European Physical Journal A
crossing symmetry of the Compton amplitude of eq. (1)
means invariance under  , , which obviously
leads to f () being an even and g() being an odd function
of the energy : f () = f (), g() = g(). The two
scalar functions f (), g() admit the following low-energy
expansion:
e2
f () = + (E + M ) 2 + O( 4 ), (2)
4M
e2 2
g() = + 0 3 + O( 5 ), (3)
8M 2
and hence, in the low-energy limit, are given in terms of
the targets charge e, mass M , and anomalous magnetic
moment (a.m.m.) . The next-to-leading order terms are
given in terms of the nucleon electric (E ), magnetic (M ),
and forward spin (0 ) polarizabilities.
In order to derive sum rules (SRs) for these quanti-
ties one assumes the scattering amplitude is an analytic
function of everywhere but the real axis, which allows
writing the real parts of the functions f () and g() as
a dispersion integral involving their corresponding imagi-
nary parts. The latter, on the other hand, can be related to
combinations of doubly polarized photoabsorption cross-
sections via the optical theorem,
and, in principle, one could continue in order to isolate
higher-order moments [8].
Recently, the helicity dierence which enters the
integrands of eqs. (9) and (10) has been measured. The
rst measurement was carried out at MAMI (Mainz) for
photon energies in the range 200 MeV < < 800 MeV [9,
10], and was extended at ELSA (Bonn) [11] for up to
3 GeV. This dierence, shown in g. 1, uctuates much
more strongly than the total cross section T . The thresh-
old region is dominated by S-wave pion production, and
therefore mostly contributes to the cross section 1/2 . In
the region of the (1232) with spin J = 3/2, both helicity
cross sections contribute, but since the transition is essen-
tially M 1, we nd 3/2 /1/2 3. As seen from g. 1, 3/2
also dominates the proton photoabsorption cross section
in the second and third resonance regions.
600
3/2-1/2 (b)
500
400



Im f () = 1/2 () + 3/2 () , (4) 300
8


Im g() = 1/2 () 3/2 () , (5) 200
8
where is the doubly-polarized total cross-section of the
100
photoabsorption processes, with specifying the total he-
licity of the initial system. Averaging over the polarization
of initial particles gives the total unpolarized cross-section, 0
T = 21 (1/2 + 3/2 ).
After these steps one arrives at the results (see, e.g., [1] -100
for more details): 0 200 400 600 800 1000 1200 1400 1600 1800
 (MeV)
2 T (  )
f () = f (0) + 2 2
d  , (6)
2 0 2 i Fig. 1. The helicity dierence 3/2 () 1/2 () for the pro-

(  ) ton. The calculations include the contribution of N interme-
g() = 2  d  , (7) diate states (dashed curve) [12], N intermediate state (dotted
4 0 2 2 i
curve) [13], and the N intermediate states (dashed-dotted
with 3/2 1/2 , and where the sum rule for curve) [14]. The total sum of these contributions is shown by
the unpolarized forward amplitude f () has been once- the full curves. The MAMI data are from ref. [9, 10] and the
subtracted to guarantee convergence. These relations can ELSA data from ref. [11].
then be expanded in energy to obtain the SRs for the dif-
ferent static properties introduced in eqs. (2), (3). In this
way we obtain the Baldin SR [6,7]:
2.2 Linearized GDH sum rule

1 T ()
E + M = d, (8) Recently, it was shown that by taking derivatives of the
2 2 0 2 GDH sum rule with respect to the a.m.m. one can ob-
the GDH SR: tain a new set of sum-rule like relations with intriguing

properties [4,5].
e2 2 1 () To derive such sum rules. one begins by introducing a
= d, (9)
2M 2 0 classical (or trial ) value of the particles a.m.m., 0 .
At the Lagrangian level this amounts to the introduction
a SR for the forward spin polarizability: of a Pauli term for the spin-1/2 eld :

1 () i0
0 = 2 d, (10) LPauli = F , (11)
4 0 3 4M

Proton magnetic moment Neutron magnetic moment
4 0
IR
3
 1
2
SR
2 
1 by means of electromagnetic probes became possible with
HB
3
0 0.2 0.4 0.6 0.8 1 0 0.2
m 2 GeV2  m 2 GeV2 
Fig. 2. Chiral behavior of proton and neutron magnetic mo-
ments (in nucleon magnetons) to one loop compared with lat-
tice data (solid circles). SR (dotted lines): one-loop relativis-
tic result based on eq. (12), IR (blue long-dashed lines):
infrared-regularized relativistic result, HB (green dashed
lines): leading non-analytic term in the heavy-baryon expan-
sion. Red solid lines: single-parameter t based on the SR re-
sult, see refs. [4, 5]. The open diamonds represent the experi-
mental values at the physical pion mass.
where F is the electromagnetic eld tensor and =
(i/2)[ , ] is the usual Dirac tensor operator. At the end
of the calculation, 0 is set to zero, but in the evaluation of
the absorption cross sections the total value of the a.m.m.
is = 0 + , with denoting the loop contribution. It
was shown in ref. [4,5] that this yields the SR :
 C2/M 1 are at the level of a few percent.
4 2 em d
2
=  ()|0 =0 , (12)
M 0
where  () is the derivative of an absorption cross sec-
tion w.r.t. the trial a.m.m. value 0 . The striking feature
of this sum rule is the linear relation between the a.m.m.
and the (derivative of the) photoabsorption cross section,
in contrast to the GDH SR where appears quadratically.
Although the cross-section quantity  () is not an ob-
servable, it is very clear how it can be determined within
a specic theory. Thus, for example, the rst derivative of
the tree-level cross-section with respect to 0 , at 0 = 0,
in QED was worked out in ref. [4], yielding Schwingers
one-loop result. It is noteworthy that this result is repro-
duced by computing only a (derivative of the) tree-level
Compton scattering cross-section and then performing an
integration over energy. This is denitely much simpler
than obtaining the Schwinger result from the GDH SR di-
rectly [15], which requires an input at the one-loop level.
The SR of eq. (12) can furthermore be applied to study
the magnetic moment and polarizabilities of the nucleon in
a relativistic chiral EFT framework [4,5]. In particular it
allows to study the chiral extrapolation of these quantities,
as shown in g. 2 for the magnetic moments. One sees that
the SR calculation, strictly satisfying analyticity, is better
suited for the chiral extrapolation of lattice QCD results
than the usual heavy-baryon expansions or the infrared-
regularized relativistic theory.
3 Nucleon excitation spectrum
The sum rules for Compton scattering o the nucleon are
dominated by its rst excited state the (1232) reso-
M. Vanderhaeghen: Two-photon physics 73
nance, as is apparent from g. 1. Through the sum rules,
the therefore plays a preponderant role in our under-
standing of low-energy nucleon structure. This justies a
dedicated eort to study this resonance.
High-precision measurements of the N -to- transition
the advent of the new generation of electron scattering
0.4 0.6 0.8 1 facilities, such as BATES, MAMI, and JLab, many mea-
surements being completed in recent years [16,17,18,19].
The electromagnetic nucleon-to- (or, in short N )
transition is predominantly of the magnetic dipole (M 1)
type. In a simple quark-model picture, this M 1 transition
is described by a spin ip of a quark in the s-wave state.
Any d-wave admixture in the nucleon or the wave-
functions allows for the electric (E2) and Coulomb (C2)
quadrupole transitions. Therefore by measuring these one
is able to assess the presence of the d-wave components
and hence quantify to which extent the nucleon or the
wave-function deviates from the spherical shape, i.e., to
which extent they are deformed [20]. The N tran-
sition, on the other hand, was accurately measured in
the pion photo- and electro-production reactions in the
-resonance energy region. The E2 and C2 transitions
were found to be relatively small at moderate momentum-
transfers (Q2 ), the ratios REM = E2/M 1 and RSM =
Traditionally, the resonance parameters are extracted
by using unitary isobar models [21,22,23,24,25,12,26],
which in essence are unitarized tree-level calculations
based on phenomenological Lagrangians. However, at low
Q2 the N -transition shows great sensitivity to the
pion cloud , which until recently could only be compre-
hensively studied within dynamical models [27,28,29,30,
31,32], which unlike the isobar models include quan-
tum eects due to pion loops.
With the advent of the chiral eective eld the-
ory (EFT) of QCD [33,34] and its extensions to the
(1232) region [35,36,37,38,39,40], it has become pos-
sible to study the nucleon and -resonance properties
in a profoundly dierent way. Recently, rst relativistic
EFT studies were performed of the N -transition in
pion electroproduction [41,42] and of the (1232) mag-
netic dipole moment (MDM) in the radiative pion pho-
toproduction [43]. The advantages over the previous dy-
namical approaches are apparent: EFT is a low-energy
eective eld theory of QCD and as such it provides a
rm theoretical foundation, with all the relevant symme-
tries and scales of QCD built in consistently.
The EFT of the strong interaction is indispensable,
at least at present, in relating the low-energy observables
(e.g., hadron masses, magnetic moments, form factors) to
ab initio QCD calculations on the lattice. On the other
hand, EFT can and should be used in extracting vari-
ous hadronic properties from the experiment. The EFT
fullls both of these roles in a gratifying fashion.
The following sections review recent progress in the
EFT in the -resonance region that has been obtained
for the N and masses [44], the N transition [41,42],
and the MDM [43].
74 The European Physical Journal A

3.1 Chiral eective eld theory in the (1232) region in this case constraints do not hold for su ciently strong
electromagnetic elds, see, e.g., [50]. In extracting the
Starting from the eective Lagrangian of chiral pertur- MDM, it is therefore assumed that the electromagnetic
bation theory (PT) with pion and nucleon elds [45], eld is weak, compared to the mass scale.
the is included explicitly in the so-called -expansion The inclusion of the -resonance introduces another
scheme [39]. In the following, the Lagrangian L(i) is or- light scale besides the pion mass in the theory, the
ganized such that superscript i stands for the power of resonance excitation energy: M MN 0.3 GeV.
electromagnetic coupling e plus the number of derivatives This energy scale is still relatively light in comparison
of pion and photon elds. Writing here only the terms in- to the chiral symmetry breaking scale SB 1 GeV.
volving the spin-3/2 isospin-3/2 eld of the -isobar Therefore, = /SB can be treated as a small param-
gives:1 eter. The question is, how to compare this parameter with
the small parameter of chiral perturbation theory (PT),
(1)
LN = (i D M )  = m /SB .
In most of the literature (see, e.g., refs. [35,36,37,38,
ihA  
+ N Ta ( ) D a + H.c. 40]) they are assumed to be of comparable size, .
2f M This, however, leads to a somewhat unsatisfactory result
HA because obviously the -contributions are overestimated
T a ( ) a , (13)
2M f at lower energies and underestimated at the resonance
(2) ie( 1) energies. To estimate the -resonance contributions cor-
LN = F rectly, and depending on the energy region, one needs to
2M
3iegM   count and  dierently.
+ N T3 F + H.c. A relation  = 2 was suggested and explored in [39],
2MN (MN + M ) and is referred to as the -expansion. The second power is
ehA   indeed the closest integer power for the relation of these
N Ta A a + H.c. , (14) parameters in the real world. In refs. [44,41,42,43] this re-
2f M
(3) 3e lation was used for power-counting purposes only, and was
LN = N T 3 5 [gE ( ) not imposed in the actual evaluations of diagrams. Each
2MN (MN + M )
 diagram is simply characterized by an overall -counting
igC index n, which tells us that its contribution begins at
+ ( ) F + H.c., (15)
M O( n ).
Because of the distinction of m and the counting
where MN and M are, respectively, the nucleon and - of a given diagram depends on whether the characteristic
isobar masses, N and a (a = 1, 2, 3) stand for the nu- momentum p is in the low-energy region (p m ) or in
cleon and pion elds, D is the covariant derivative en- the resonance region (p ). In the low-energy region the
suring the electromagnetic gauge-invariance, F and F index of a graph with L loops, N pion propagators, NN
are the electromagnetic eld strength and its dual, Ta are nucleon propagators, N -propagators, and Vi vertices
the isospin 1/2 to 3/2 transition matrices, and T a are of dimension i is
the generators in the isospin 3/2 representation of SU (2),  

satisfying T a T a = 5/3. The coupling constants are given n=2 iVi + 4L NN 2N N 2nPT N ,
by : f = 92.4 MeV, hA  2.85 is obtained from the - i
resonance width, = 0.115 GeV, and for HA the large- (16)
Nc relation HA = (9/5)gA is adopted, with gA  1.267 where nPT is the index in PT with no s [45]. In the
the nucleon axial-coupling constant. resonance region, one distinguishes the one--reducible
Note that the electric and the Coulomb N couplings (OR) graphs [39]. Such graphs contain propagators
(gE and gC , respectively) are of one order higher than the which go as 1/(p ), and hence for p they are
magnetic (gM ) one, because of the 5 which involves the large and all need to be included. This gives an incen-
small components of the fermion elds and thus intro- tive, within the power-counting scheme, to resum con-
duces an extra power of the 3-momentum. The MDM tributions. Their resummation amounts to dressing the
is dened here in units of [e/2M ]. Higher electromagnetic propagators so that they behave as 1/(p ). The
moments are omitted, because they do not contribute at self-energy begins at order p3 and thus a dressed OR
the orders that we consider. propagator counts as 1/ 3 . If the number of such propa-
(1)
Note that L contains the free Lagrangian, which is gators in a graph is NOR , the power-counting index of
formulated in [46] such that the number of spin degrees of this graph in the resonance region is given by
freedom of the relativistic spin-3/2 eld is constrained to
the physical number: 2s + 1 = 4. The N to transition n = nPT N 2NOR , (17)
couplings in eqs. (13,14,15) are consistent with these con-
straints [47,48,49]. The coupling is more subtle since where N is the total number of -propagators.
A word on the renormalization program, as it is an
1
Here we introduce totally antisymmetric products of - indivisible part of power counting in a relativistic theory.
matrices: = 12 [ , ], = 12 { , } = i 5 . Indeed, without some kind of renormalization the loop
 M. Vanderhaeghen: Two-photon physics
graphs diverge as N , where is an ultraviolet cuto,
and N is a positive power proportional to the power-
counting index of the graph. Also, contributions of heavy
scales, such as baryon masses, may appear as M N . The
renormalization of the loop graphs can and should be per-
formed so as to absorb these large contributions into the
available low-energy constants, thus bringing the result in
accordance with power counting [51].
To give an example, consider the one-N -loop con-
tribution to the nucleon mass. For the N N vertex,
the power counting tells us that this contribution be-
gins at O(m3 ). An explicit calculation, however, will show
(e.g., [45]) that the loop produces O(m0 ) and O(m2 )
terms, both of which are (innitely) large. This is not a
violation of power counting, because there are two low-
energy constants: the nucleon mass in the chiral limit,
M (0) , and c1N , which enter at order O(m0 ) and O(m2 ),
respectively, and renormalize away the large contributions
coming from the loop. The renormalized relativistic result,
up to and including O(m3 ), can be written as [44]:
(0)
MN = MN 4 c1N m2 (18)
   5/2
2
3 gA m2 m
2
m3 4 1 2 arccos
(8f ) 4MN 2MN
 3
17m m
+
16MN 2MN
  2  4  
m m m m
+ 30 10 + ln ,
8MN MN MN MN
and one can easily verify that the loop contribution begins
at O(m3 ) in agreement with power counting.
Likewise, the mass has also been calculated in rela-
tivistic EFT see ref. [44] for details.
The m dependence of the nucleon and -resonance
masses are compared with lattice results in g. 3. One of
the two parameters in eq. (18) is constrained by the physi-
cal nucleon mass value at m = 0.139 GeV, while the other
parameter is t to the lattice data shown in the gure.
This yields : MN = 0.883 GeV and c1N = 0.87 GeV1 .
(0)
As is seen from the gure, with this two-parameter form
for MN , a good description of lattice results is obtained up
to m2  0.5 GeV2 . Analogously to the nucleon case, one
low-energy constant for the is xed from the physical
value of the mass, while the second parameter is t to
(0)
the lattice data shown in g. 3, yielding : M = 1.20 GeV
and c1 = 0.40 GeV1 . As well as for the nucleon, this
two-parameter form for M yields a fairly good descrip-
tion of the lattice results up to m2  0.5 GeV2 .
3.2 N transition
The N transition is usually studied through the pion
electroproduction process. The pion electroproduction
amplitude to NLO in the expansion, in the resonance
region, is given by graphs in g. 4(a) and (b), where the
shaded blobs in graph (a) include corrections depicted in
75
1.8
M (GeV)
1.7
1.6
1.5
1.4
1.3
1.2
1.1
1 N
0.9
0.8
0 0.1 0.2 0.3 0.4 0.5
2
m (GeV ) 2
Fig. 3. Pion-mass dependence of the nucleon and (1232)
masses. The curves are two-parameter expressions for the N
loop contributions to MN and M as calculated in ref. [44]
(see text for the values of the low-energy constants). The red
squares are lattice results from the MILC Collaboration [52].
The stars represent the physical mass values.
(a) (b)
(c) (d) (e) (f)
Fig. 4. Diagrams for the eN eN reaction at NLO in the
-expansion. Double lines represent the propagators.
g. 4(c f). The hadronic part of graph (a) begins at O( 0 )
which here is the leading order. The Born graphs (b) con-
tribute at O(). The one-loop vertex corrections of g. 4(e)
and (f) to the N -transition form factors have been eval-
uated in two independent ways in refs. [41,42], to which
we refer for details. At NLO there are also vertex cor-
rections of the type (e) and (f) with nucleon propagators
in the loop replaced by the -propagators. However, after
the appropriate renormalizations and Q2  SB , these
graphs start to contribute at next-next-to-leading order.
The vector-meson diagram, g. 4(d), contributes to NLO
for Q2 SB . It was included eectively in refs. [41,42]
by giving the gM -term a dipole Q2 -dependence, in anal-
ogy to how it is usually done for the nucleon isovector
form factor.
The resonant pion photoproduction multipoles are
used to determine the two low-energy constants: gM and
gE , the strength of the M 1 and E2 N transitions.
In g. 5, we show the result of the EFT calculations
(3/2)
for the pion photoproduction resonant multipoles M1+
(3/2)
and E1+ , around the resonance position, as function of
the total c.m. energy W of the N system. These two
multipoles are well established by the MAID [12] and
SAID [53] partial-wave solutions which allow us to t
76 The European Physical Journal A

Re M1+3/2 (10-3/m)
30 4
40

T + L (b/sr)
20
2

LT (b/sr)
10
0 0
20
-10
-2
-20
-4
Im M1+3/2 (10-3/m)

0
40
10 6
30

LT (b/sr)
TT (b/sr)
0 4
20
-10 2
10

0 -20 0
1.18 1.2 1.22 1.24 1.26 1.28

W (GeV) -30 -2
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
cos cos
1
Re E1+3/2 (10-3/m)

0.5 Fig. 6. EFT NLO results for the dependence of the p

0 0 p cross sections at W = 1.232 GeV and Q2 = 0.127 GeV2 .
-0.5 The theoretical error bands are described in the text. Data
-1
points are from BATES experiments [18, 54].
-1.5
-2
-2.5
Im E1+3/2 (10-3/m)

0.5
0
-0.5
-1
-1.5
-2
-2.5
1.18 1.2 1.22 1.24
(3/2)
Fig. 5. Multipole amplitudes M1+ (top panels) and E1+
(bottom panels) for pion photoproduction as function of the in-
variant mass W of the N system. Dashed curves: contribu-
tion without the N -vertex corrections, (i.e., g. 4(a) with-
out g. 4(e, f)). Dotted curves: adding the Born contributions,
g. 4(b), to the dashed curves. Solid curves: complete NLO
calculation, includes all graphs from g. 4. In all curves the
low-energy parameters are chosen as : gM = 2.9, gE = 1.0.
The data point are from the SAID analysis (FA04K) [53] (red
circles), and from the MAID 2003 analysis [12] (blue squares).
the two low-energy constants of the chiral Lagrangian of
eqs. (14,15) as : gM = 2.9, gE = 1.0. As is seen from
g. 5, with these values the NLO results (solid lines) give
a good description of the energy dependence of the reso-
nant multipoles in a window of 100 MeV around the -
resonance position. Also, these values yield REM = 2.3
%, in a nice agreement with experiment [16].
The dashed curves in g. 5 show the contribution of
the -resonant diagram of g. 4(a) without the NLO ver-
tex corrections g. 4(e, f). For the M1+ multipole this is
the LO contribution. For the E1+ multipole the LO con-
tribution is absent (the gE coupling is of one order higher
than gM ). Hence, the dashed curve represents a partial
NLO contribution to E1+ therein. Upon adding the non-
resonant Born graphs, g. 4(b), to the dashed curves one
obtains the dotted curves in g. 5. These non-resonant
contributions are purely real at this order and do not af-
fect the imaginary part of the multipoles. One sees that
the resulting calculation is awed because the real parts
of the resonant multipoles now fail to cross zero at the res-
onance position and hence unitarity, in the sense of Wat-
sons theorem [55], is violated. The complete NLO calcu-
1.26 1.28
lation, shown by the solid curves in g. 5, includes in addi-
W (GeV) tion the vertex corrections, g. 4(e, f), which restore uni-
tarity exactly. Watsons theorem is satised exactly by the
(3/2)
NLO, up to-one-loop amplitude given the graphs in g. 4.
Figure 6 shows the NLO results for dierent vir-
tual photon absorption cross sections (for denitions,
see ref. [42]) at the resonance position, and for Q2 
0.127 GeV2 , where recent precision data are available. Be-
sides the low-energy constants gM and gE , which were
xed from the resonant multipoles in g. 5, the only other
low-energy constant from eq. (15) entering the NLO elec-
troproduction calculation is gC . The main sensitivity on
gC enters in LT . A best description of the LT data in
g. 6 is obtained by choosing gC = 2.36.
The theoretical uncertainty due to the neglect of
higher-order eects was estimated in ref. [42]. We know
that they must be suppressed by at least one power of
(= /SB ) as compared to the NLO and two powers of
as compared to the LO contributions. These error esti-
mates are shown by the bands in g. 6. One sees that the
NLO EFT calculation, within its accuracy, is consistent
with the experimental data for these observables.
Figure 7 shows the Q2 dependence of the ratios REM
and RSM . Having xed the low energy constants gM , gE
and gC , this Q2 dependence follows as a prediction. The
theoretical uncertainty here (shown by the error bands)
was also estimated in ref. [42] over the range of Q2 from 0
to 0.2 GeV2 . One sees that the NLO calculations are con-
sistent with the experimental data for both of the ratios.
 M. Vanderhaeghen: Two-photon physics 77

Figure 8 shows the m -dependence of the ratios REM

0 and RSM and compares them to lattice QCD calculations.
The recent state-of-the-art lattice calculations of REM
-2 and RSM [58] use a linear, in the quark mass (mq m2 ),
REM (%)

extrapolation to the physical point, thus assuming that the

-4 non-analytic mq -dependencies are negligible. The thus ob-
tained value for RSM at the physical m value displays a
large discrepancy with the experimental result, as seen
-6
in g. 8. The relativistic EFT calculation, on the other
hand, shows that the non-analytic dependencies are not
-2
negligible. While at larger values of m , where the is
-4 stable, the ratios display a smooth m dependence, at
-6
m = there is an inection point, and for m the
RSM (%)

non-analytic eects are crucial.

-8 One also notices from g. 8 that there is only little
-10 dierence between the EFT calculations with the m -
dependence of MN and M accounted for, and an earlier
-12
calculation [41], where the ratios were evaluated neglecting
-14
0 0.1 0.2
the m -dependence of the masses.
0.05 0.15 0.25
2 2 Figure 8 also shows a theoretical uncertainty of the
Q (GeV )
ratios REM and RSM taken over the range of m2 from 0
Fig. 7. Q2 dependence of the NLO results (solid curves) for to 0.15 GeV2 . The m dependence obtained from EFT
REM (upper panel) and RSM (lower panel) [41, 42]. The blue clearly shows that the lattice results for RSM may in fact
dashed curves represent a phenomenological estimate of N2 LO be consistent with experiment.
eects by including Q2 -dependence in gE according to a dipole
behavior, see ref. [42]. The blue circles are data points from
MAMI for REM [16], and RSM [56, 57]. The green squares are
data points from BATES [18]. 3.3 (1232) magnetic dipole moment

Although the (1232)-isobar is the most distinguished

4 and well-studied nucleon resonance, such a fundamen-
tal property as its magnetic dipole moment (MDM) has
2 thusfar escaped a precise determination. The problem is
REM (%)

generic to any unstable particle whose lifetime is too short

0
for its MDM to be measurable in the usual way through
-2 spin precession experiments. A measurement of the MDM
of such an unstable particle can apparently be done only
-4 indirectly, in a three-step process, where the particle is
rst produced, then emits a low-energy photon which
4 plays the role of an external magnetic eld, and nally
2 decays. In this way the MDM of ++ is accessed in the
0 reaction + p + p [59,60] while the MDM of + can
RSM (%)

be determined using the radiative pion photoproduction

-2
(p 0 p  ) [61].
-4 A rst experiment devoted to the MDM of + was
-6 completed in 2002 [62]. The value extracted in this exper-
-8 iment, + = 2.7+1.0 1.3 (stat.) 1.5(syst.) 3(theor.) (nu-
-10 clear magnetons), is based on theoretical input from the
0 0.05 0.1
2
0.15
2
0.2 0.25 phenomenological model [63,64] of the p 0 p  reac-
m (GeV ) tion. To improve upon the precision of this measurement,
Fig. 8. m dependence of the NLO results at Q2 = 0.1 GeV2 a dedicated series of experiments has recently been car-
for REM (upper panel) and RSM (lower panel). The blue cir- ried out by the Crystal Ball Collaboration at MAMI [65].
cle is a data point from MAMI [56], the green squares are These experiments achieve about two orders of magnitude
data points from BATES [18]. The three lled black diamonds better statistics than the pioneering experiment [62]. The
at larger m are lattice calculations [58], whereas the open aim of the investigation within EFT was to complement
diamond near m
0 represents their extrapolation assum- these high-precision measurements with an accurate and
ing linear dependence in m2 . Solid curves: NLO result when model-independent analysis of the p 0 p  reaction.
accounting for the m dependence in MN and M ; Dashed The optimal sensitivity of the p 0 p  reaction
curves: NLO result of ref. [41], where the m -dependence of to the MDM term is achieved when the incident photon
MN and M was not accounted for. energy is in the vicinity of , while the outgoing photon
78 The European Physical Journal A

For + , the curve is in disagreement with the trend of the

recent lattice data, which possibly is due to the quench-
ing in the lattice calculations. The dotted line in g. 10
(a) (b) (c) shows the result [4] for the magnetic moment for the pro-
ton. One sees that + and p , while having very distinct
behavior for small m , are approximately equal for larger
values of m .
(d) (e) (f) We next discuss the EFT results for the p 0 p 
observables. The NLO calculation of this process in the -
Fig. 9. Diagrams for the p p reaction at NLO in the
0
expansion corresponds with the diagrams of g. 9. This
-expansion, considered in this work. Double lines represent
the propagators.
calculation completely xes the imaginary part of the
vertex. It leaves as only free parameter, which en-
(2)
ters as a low energy constant in LN . Thus the real part of
+ is to be extracted from the p 0 p  observables,
6 some of which are shown in g. 11 for an incoming photon
 energy Elab = 400 MeV as function of the emitted photon
Re  energy E c.m. . In the soft-photon limit (E c.m. 0), the
5
p 0 p  reaction is completely determined from the
4 bremsstrahlung process from the initial and nal protons.
The deviations of the p 0 p  observables, away from
3 the soft-photon limit, will then allow to study the sensi-
Re  tivity to + . It is therefore very useful to introduce the
ratio [64]:
2
1 d
R E , (19)
1 dE
Im 
where d/dE is the p 0 p  cross section integrated
0 0.2 0.4 0.6 0.8 1 over the pion and photon angles, and is the angular in-
m 2 GeV2  tegrated cross section for the p 0 p process weighted
with the bremsstrahlung factor, as detailed in [64]. This
Fig. 10. Pion mass dependence of the real (solid curves) and ratio R has the property that in the soft-photon limit, the
imaginary (dashed curves) parts of ++ and + MDMs (in low energy theorem predicts that R 1. From g. 11 one
nuclear magnetons). Dotted curve is the result for the proton then sees that the EFT calculation obeys this theorem.
magnetic moment from ref. [4]. The experimental data point This is a consequence of gauge-invariance which is main-
for ++ is from PDG analysis [66]. Lattice data are from [67] tained exactly throughout the calculation, also away from
for ++ and from [68] for + .
the soft-photon limit.
The EFT result for R shows clear deviations from
unity at higher outgoing photon energies, in good agree-
energy is of order of m . In this case the p 0 p  am- ment with the rst data for this process [62]. The sensi-
plitude to next-to-leading order (NLO) in the -expansion tivity of the EFT calculation to the is a very promis-
is given by the diagrams of g. 9(a - c), where the shaded ing setting for the dedicated second-generation experiment
blobs, in addition to vertices from eqs. (13,14,15), contain which has recently been completed by the Crystal Ball
the one-loop corrections shown in g. 9(d - f). The contri- Coll. at MAMI [65]. It improves upon the statistics of the
butions to of diagrams (e) and (f ) in g. 9 have been rst experiment (g. 11) by at least two orders of magni-
calculated in ref. [43], to which we refer for technical de- tude and will allow for a reliable extraction of + using
tails. The evaluation of these loop diagrams also allows to the EFT calculation presented here.
quantify the m dependence of which can be used to Besides the cross section, the p 0 p  asymme-
compare with lattice QCD results. As all lattice data for tries for linearly and circularly polarized incident photons
at present and in the foreseeable future are for larger have also been measured in the recent dedicated exper-
than the physical values of m , their comparison with ex- iment [65]. They are also shown in g. 11. The photon
periment requires the knowledge of the m -dependence asymmetry for linearly polarized photons, , at E = 0
for this quantity. Figure 10 shows the pion mass depen- exactly reduces to the p 0 p asymmetry. It is seen
dence of real and imaginary parts of the + and ++ from g. 11 that the EFT calculation is in good agree-
MDMs, according to our one-loop calculation. Each of the ment with the experimental value. At higher outgoing
two solid curves has a free parameter, the counterterm photon energies, the photon asymmetry as predicted by
(2)
from LN , adjusted to agree with the lattice data at larger the NLO EFT calculation remains nearly constant and
values of m . As can be seen from g. 10, the MDM is very weakly dependent on . It is an ideal observable
develops an imaginary part when m < = M M , for a consistency check of the EFT calculation and to
whereas the real part has a pronounced cusp at m = . test that the diagrams of g. 9 indeed dominate the
 M. Vanderhaeghen: Two-photon physics 79

4 Conclusions
2 Elab = 400 MeV

1.5 It was discussed here how Compton scattering sum rules

R

relate low-energy nucleon structure quantities to the nu-

1 cleon excitation spectrum, with special emphasis on the
0.5
GDH sum rule. I demonstrated the utility of taking deriva-
tives of the GDH sum rule, in order to convert it to forms
0 which are sometimes more calculationally robust. In par-
0.8 E lab = 400 MeV
ticular it was shown how it allows to estimate the chiral
c.m. = 90o
0.6 extrapolations of lattice QCD results for anomalous mag-
netic moments of nucleons.

0.4 Subsequently, new developments in our description of

Re = 5
Re = 3 the nucleon excitation spectrum were discussed. In partic-
0.2 Re = 1
Re = 3, Im = 0
ular I reviewed recent work on a EFT framework for the
0 (1232)-resonance region. This framework plays a dual
0.2 0o < c.m. < 90o role, in that it allows for an extraction of resonance pa-
0o < c.m. < 180o rameters from observables and predicts their m depen-
dence. In this way it may provide a crucial connection of
circ

0.1
present lattice QCD results obtained at unphysical values
of m to the experiment. This was demonstrated here ex-
0 E lab = 400 MeV plicitely for the N and masses, the N transition and
c.m. = 90o the magnetic dipole moment. As the next-generation
-0.1
0 20 40 60 80 100 120 140 160 lattice calculations of these quantities are on the way [69],
E c.m. (MeV)
such a EFT framework will, hopefully, complement these
eorts.
Fig. 11. The outgoing photon energy dependence of the
p 0 p observables for dierent values of + (in units
e/2M ). Top panel: the ratio of p 0 p to p 0 p cross- I am grateful to my colleagues in Mainz for the unique culture
sections eq. (19). Data points are from [62]. Middle panel: the of cross-fertilization between experiment and theory. On the
linear-polarization photon asymmetry of the p 0 p cross- subject of two-photon physics, I like to thank in particular
sections dierential w.r.t. the outgoing photon energy and pion Dieter Drechsel and Barbara Pasquini, for the many collabo-
c.m. angle. The data point at E = 0 corresponds with the rations. I also like to acknowledge Vladimir Pascalutsa for very
p 0 p photon asymmetry from [16]. Lower panel: the fruitful recent collaborations on the EFT in the -resonance
circular-polarization photon asymmetry (as dened in [64]), region. This work is supported in part by DOE grant no. DE-
where the outgoing photon angles have been integrated over FG02-04ER41302 and contract DE-AC05-84ER-40150 under
the indicated range. which SURA operates the Jeerson Laboratory.

process. Mechanisms involving -photoproduction Born
terms followed by N rescattering have been considered
in model calculations [63,64]. In the -counting they start
contributing at next-next-to-leading order and therefore
will provide the main source of corrections to the present
NLO results.
The asymmetry for circularly polarized photons, circ ,
(which is exactly zero for a two-body process due to re-
ection symmetry w.r.t. the reaction plane) has been pro-
posed [64] as a unique observable to enhance the sensi-
tivity to . Indeed, in the soft-photon limit, where the
p 0 p  process reduces to a two-body process, circ
is exactly zero. Therefore, its value at higher outgoing pho-
ton energies is directly proportional to . One sees from
g. 11 (lower panel) that our EFT calculation supports
this observation, and shows sizeably dierent asymmetries
for dierent values of . A combined t of all three ob-
servables shown in g. 11 will therefore allow for a very
stringent test of the EFT calculation, which can then be
used to extract the + MDM.
References
1. D. Drechsel, B. Pasquini, M. Vanderhaeghen, Phys. Rep.
378, 99 (2003).
2. S.B. Gerasimov, Sov. J. Nucl. Phys. 2, 430 (1966) (Yad.
Fiz. 2, 598 (1966)); S.D. Drell, A.C. Hearn, Phys. Rev.
Lett. 16, 908 (1966).
3. D. Drechsel, L. Tiator, Annu. Rev. Nucl. Part. Sci. 54, 69
(2004).
4. V. Pascalutsa, B.R. Holstein, M. Vanderhaeghen, Phys.
Lett. B 600, 239 (2004).
5. B.R. Holstein, V. Pascalutsa, M. Vanderhaeghen, Phys.
Rev. D 72, 094014 (2005).
6. A.M. Baldin, Nucl. Phys. 18, 310 (1960).
7. L.I. Lapidus, Sov. Phys. JETP 16, 964 (1963).
8. B.R. Holstein, D. Drechsel, B. Pasquini, M. Vander-
haeghen, Phys. Rev. C 61, 034316 (2000).
9. GDH and A2 Collaborations (J. Ahrens et al.), Phys. Rev.
Lett. 84, 5950 (2000).
10. GDH and A2 Collaborations (J. Ahrens et al.), Phys. Rev.
Lett. 87, 022003 (2001).
11. GDH Collaboration (H. Dutz et al.), Phys. Rev. Lett. 91,
192001 (2003).
80 The European Physical Journal A
12. D. Drechsel, O. Hanstein, S.S. Kamalov, L. Tiator, Nucl.
Phys. A 645, 145 (1999).
13. D. Drechsel, S.S. Kamalov, L. Tiator, Phys. Rev. D 63,
114010 (2001).
14. H. Holvoet, PhD Thesis, University Gent (2001).
15. D.A. Dicus, R. Vega, Phys. Lett. B 501, 44 (2001).
16. R. Beck et al., Phys. Rev. Lett. 78, 606 (1997); Phys. Rev.
C 61, 035204 (2000).
17. G. Blanpied et al., Phys. Rev. Lett. 79, 4337 (1997).
18. C. Mertz et al., Phys. Rev. Lett. 86, 2963 (2001); N.F.
Sparveris et al., Phys. Rev. Lett. 94, 022003 (2005).
19. V.V. Frolov et al., Phys. Rev. Lett. 82, 45 (1999); K. Joo
et al., Phys. Rev. Lett. 88, 122001 (2002).
20. S.L. Glashow, Physica A 96, 27 (1979); N. Isgur, G. Karl,
R. Koniuk, Phys. Rev. D 25, 2394 (1982); S. Capstick,
G. Karl, Phys. Rev. D 41, 2767 (1990); S. Capstick, B.D.
Kiester, Phys. Rev. D 51, 3598 (1995); G.A. Miller, Phys.
Rev. C 68, 022201(R) (2003); A.M. Bernstein, Eur. Phys.
J. A 17, 349 (2003).
21. M.G. Olsson, E.T. Osypowski, Nucl. Phys. B 87, 399
(1974); Phys. Rev. D 17, 174 (1978); R.M. Davidson,
N.C. Mukhopadhyay, R.S. Wittman, Phys. Rev. D 43, 71
(1991).
22. H. Garcilazo, E. Moya de Guerra, Nucl. Phys. A 562, 521
(1993); C. Fernandez-Ramirez, E. Moya de Guerra, J.M.
Udias, arXiv:nucl-th/0509020.
23. M. Vanderhaeghen, K. Heyde, J. Ryckebusch, M. Waro-
quier, Nucl. Phys. A 595, 219 (1995).
24. V. Pascalutsa, O. Scholten, Nucl. Phys. A 591, 658 (1995);
O. Scholten, A.Y. Korchin, V. Pascalutsa, D. Van Neck,
Phys. Lett. B 384, 13 (1996); A.Y. Korchin, O. Scholten,
R.G.E. Timmermans, Phys. Lett. B 438, 1 (1998).
25. T. Feuster, U. Mosel, Phys. Rev. C 59, 460 (1999); G.
Penner, U. Mosel, Phys. Rev. C 66, 055211 (2002); H.
Lenske, V. Shklyar, U. Mosel, arXiv:nucl-th/0512044.
26. I.G. Aznauryan, Phys. Rev. C 68, 065204 (2003).
27. S. Nozawa, B. Blankleider, T.-S.H. Lee, Nucl. Phys. A 513,
459 (1990); T.-S.H. Lee, B.C. Pearce, Nucl. Phys. A 530,
532 (1991); T. Sato, T.-S.H. Lee, Phys. Rev. C 54, 2660
(1996); 63, 055201 (2001).
28. Y. Surya, F. Gross, Phys. Rev. C 53, 2422 (1996); 47, 703
(1993).
29. S.S. Kamalov, S.N. Yang, Phys. Rev. Lett. 83, 4494 (1999);
S.S. Kamalov, G.Y. Chen, S.N. Yang, D. Drechsel, L. Tia-
tor, Phys. Lett. B 522, 27 (2001); L. Tiator, D. Drechsel,
S.S. Kamalov, S.N. Yang, Eur. Phys. J. A 17, 357 (2003).
30. M.G. Fuda, H. Alharbi, Phys. Rev. C 68, 064002 (2003).
31. V. Pascalutsa, J.A. Tjon, Phys. Lett. B 435, 245 (1998);
Phys. Rev. C 61, 054003 (2000); 70, 035209 (2004).
32. G. Caia, V. Pascalutsa, J.A. Tjon, L.E. Wright, Phys. Rev.
C 70, 032201(R) (2004); G. Caia, L.E. Wright, V. Pasca-
lutsa, Phys. Rev. C 72, 035203 (2005).
33. S. Weinberg, Physica A 96, 327 (1979).
34. J. Gasser, H. Leutwyler, Ann. Phys. (N.Y.) 158, 142
(1984).
35. E. Jenkins, A.V. Manohar, Phys. Lett. B 255, 558 (1991);
259, 353 (1991).
36. T. Hemmert, B.R. Holstein, J. Kambor, Phys. Lett. B 395,
89 (1997); G. Gellas et al., Phys. Rev. D 60, 054022 (1999).
37. H.-B. Tang, P. Ellis, Phys. Lett. B 387, 9 (1996).
38. N. Fettes, U.G. Meissner, Nucl. Phys. A 679, 629 (2001).
39. V. Pascalutsa, D.R. Phillips, Phys. Rev. C 67, 055202
(2003); 68, 055205 (2003).
40. C. Hacker, N. Wies, J. Gegelia, S. Scherer, Phys. Rev. C
72, 055203 (2005).
41. V. Pascalutsa, M. Vanderhaeghen, Phys. Rev. Lett. 95,
232001 (2005).
42. V. Pascalutsa, M. Vanderhaeghen, arXiv:hep-ph/0512244.
43. V. Pascalutsa, M. Vanderhaeghen, Phys. Rev. Lett. 94,
102003 (2005).
44. V. Pascalutsa, M. Vanderhaeghen, arXiv:hep-ph/0511261.
45. J. Gasser, M.E. Sainio, A. Svarc, Nucl. Phys. B 307, 779
(1988).
46. W. Rarita, J.S. Schwinger, Phys. Rev. 60, 61 (1941).
47. V. Pascalutsa, Phys. Rev. D 58, 096002 (1998).
48. V. Pascalutsa, R.G.E. Timmermans, Phys. Rev. C 60,
042201(R) (1999).
49. V. Pascalutsa, Phys. Lett. B 503, 85 (2001).
50. S. Deser, V. Pascalutsa, A. Waldron, Phys. Rev. D 62,
105031 (2000).
51. J. Gegelia, G. Japaridze, Phys. Rev. D 60, 114038 (1999);
J. Gegelia, G. Japaridze, X.Q. Wang, J. Phys. G 29, 2303
(2003).
52. C.W. Bernard et al., Phys. Rev. D 64, 054506 (2001).
53. R.A. Arndt, W.J. Briscoe, I.I. Strakovsky, R.L. Workman,
Phys. Rev. C 66, 055213 (2002).
54. C. Kunz et al., Phys. Lett. B 564, 21 (2003).
55. K.M. Watson, Phys. Rev. 95, 228 (1954).
56. T. Pospischil et al., Phys. Rev. Lett. 86, 2959 (2001).
57. D. Elsner et al., arXiv:nucl-ex/0507014.
58. C. Alexandrou et al., Phys. Rev. Lett. 94, 021601 (2005).
59. B.M.K. Nefkens et al., Phys. Rev. D 18, 3911 (1978).
60. A. Bosshard et al., Phys. Rev. D 44, 1962 (1991).
61. D. Drechsel, M. Vanderhaeghen, M.M. Giannini, E. San-
topinto, Phys. Lett. B 484, 236 (2000).
62. M. Kotulla et al., Phys. Rev. Lett. 89, 272001 (2002).
63. D. Drechsel, M. Vanderhaeghen, Phys. Rev. C 64, 065202
(2001).
64. W.T. Chiang, M. Vanderhaeghen, S.N. Yang, D. Drechsel,
Phys. Rev. C 71, 015204 (2005).
65. R. Beck, B. Nefkens, spokespersons Crystal Ball at MAMI
experiment.
66. Particle Data Group (K. Hagiwara et al.), Phys. Rev. D
66, 010001 (2002).
67. D.B. Leinweber, T. Draper, R.M. Woloshyn, Phys. Rev.
D 46, 3067 (1992); I.C. Cloet, D.B. Leinweber, A.W.
Thomas, Phys. Lett. B 563, 157 (2003).
68. F.X. Lee, R. Kelly, L. Zhou, W. Wilcox, arXiv:hep-
lat/0410037.
69. C. Alexandrou et al., arXiv:hep-lat/0509140.
Eur. Phys. J. A 28, s01, 81 90 (2006)
DOI: 10.1140/epja/i2006-09-009-2 EPJ A direct
electronic only

Electromagnetic form factors of the nucleon

Experiments at MAMI
M. Ostricka
Physikalisches Institut, Universitat Bonn, 55115 Bonn, Germany

/
Published online: 12 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. Elastic form factors are of fundamental importance for the understanding of microscopic spatial
structures. In case of the proton and the neutron, charge and magnetic form factors can be studied in
elastic electron scattering. Techniques to accelerate polarised continuous electron beams, the availability of
polarised targets as well as modern concepts and instrumentation for coincidence experiments and recoil
polarimetry had an enormous impact on these measurements. The developments and experiments at the
Mainz Microtron MAMI will be discussed in a general context.
PACS. 13.40.Gp Electromagnetic form factors 13.85.Dz Elastic scattering 13.88.+e Polarization in
interactions and scattering 25.30.Bf Elastic electron scattering

1 Introduction protons and neutrons are modied by charge (GE,p , GE,n )

Elastic and inelastic scattering experiments at dierent
energy scales and with dierent projectiles provide essen-
tial insight into microscopic structures in terms of exci-
tation spectra or spatial and momentum distributions of
constituents.
Form factors measured in elastic scattering are in par-
ticular determined by the ability of a system to absorb a
momentum without excitation and, therefore, re ects the
wave function of constituents in the ground state. Nonrel-
ativistically, elastic form factors are momentum represen-
tations of spatial densities like mass or electroweak char-
ge densities. In case of spherical symmetry, an expansion
close to the static limit of zero momentum transfer is given
by the integral quantity, e.g. the total charge Z, and the
corresponding mean square radius:
q2  2 
F (|
q |) = Z r + O(q 4 ). (1)
6 in single arm experiments at low duty cycle machines. It
The rst evidence for such nite size eects of atomic nu-
clei was found by Lyman, Hanson and Scott in electron
scattering experiments at a 20 MeV betatron [1]. Their
pioneering work together with the rst electron-proton-
scattering experiments by Hofstadter [2] mark the begin-
ning of a fruitful era of using electromagnetic probes to
analyse nuclear and subnuclear hadronic structures. The
electromagnetic coupling is weak enough to allow a per-
turbative treatment and strong enough for precise mea-
surements even at higher values of momentum transfer.
Compared to pointlike fermions, the contributions of
charge and total magnetic moment to the vector current of
a
e-mail: ostrick@kph.uni-mainz.de
and magnetic form factors (GM,p , GM,n ) which depend on
the square of the 4 momentum transfer Q2 .
The unpolarised cross section for electron scattering
o nucleons can be expressed in leading order as an inco-
herent sum of the response to longitudinal and transverse
polarisation components of the exchanged virtual photon.
In case of elastic scattering these responses are equal to
G2E (Q2 ) and G2M (Q2 ), respectively:
d M  2 2
= GE (Q ) + G2M (Q2 ) . (2)
d (1 + )
Here = Q2 /4M 2 ,  = [1 + 2(1 + ) tan2 2e ]1 describes
the photon polarisation and M is the Mott cross section
for a pointlike particle. In principle, both form factors can
thus be determined by studying the  dependence of the
cross section at xed values of Q2 . This technique, known
as Rosenbluth or LT separation, has been intensively used
requires measurements at dierent scattering angles and
beam energies as well as the knowledge of absolute lumi-
nosities. This Rosenbluth separation technique is intrinsi-
cally limited if one of the two summands in the unpolarised
cross section, G2E or G2M , is small compared to the other
one. In this case, only the dominating form factor can be
extracted reliably.
For the proton, since two decades the cross section
measurements of Simon et al. at the Mainz 300 MeV Linac
provide the most precise data for GE,p at low Q2 allow-
ing an accurate determination of the proton charge ra-
dius [3]. However, at higher values of Q2 the transverse
part, T G2M , is getting more and more dominant
and a separation of GE,p suers from statistical as well as
82 The European Physical Journal A

from systematic uncertainties. In single-arm experiments

at high momentum transfer up to Q2 = 30 GeV2 only the
proton magnetic form factor, GM,p , has been determined
precisely at the SLAC accelerator [4].
Measurements of neutron form factors, GM,n and
GE,n , are even more di cult due to the lack of a free e
neutron target and the smallness of GE,n (see sect. 4.3).
The results of these unpolarised, single-arm electron
scattering experiments have usually been summarised in
n/p sp
ec
the past in terms of the approximate scaling relation tro
m
et
GE,p GM,p /p GM,n /n (3) er
A
and the approximate dipole form
 2
1
GM,p p GD = (4)
1 + Q2 /m2
with the phenomenological parameter m2 = 0.71 GeV2 .
Indispensable prerequisites for more detailed analyses
and interpretations are precise measurements of all four
nucleon form factors over a larger range in momentum
transfer. To overcome the intrinsic limitations discussed hadrondetector
above, techniques are required to accelerate polarised con- Fig. 1. Setup to measure Rd at MAMI.
tinuous electron beams together with polarised targets
and an experimental instrumentation for coincidence ex-
periments and recoil polarimetry. These technological and for scattering from neutrons and protons in quasi-free
conceptual developments allow to exploit the full poten- kinematics. In plane-wave-impulse-approximation spec-
tial of electromagnetic probes and are characteristic for tral functions and the elementary electron nucleon scat-
the modern era of electron scattering. tering factorise so that the dependence on nuclear wave
These developments and form factor measurements at functions cancels in the ratio. Higher-order corrections like
the Mainz Microtron MAMI will be discussed in the next nal state interactions (FSI) or meson exchange currents
sections. A recent review on form factors and their mea- (MEC) are small ( 2%) and calculable [9,10].
surements in general can be found in [5]. Measurements of the ratio Rd thus provided a signif-
icant break-through in the knowledge of GM,n at low Q2
and have been pioneered at NIKHEF [11], ELSA [12] and
2 Coincidence experiments to measure GM,n MAMI [13,14].
Lacking a free neutron target Ehrenberg and Hofstadter The setup at MAMI is shown in g. 1. The electrons
for the rst time used inclusive electron scattering o light are detected in a magnetic spectrometer with a solid angle
nuclei in quasi free kinematics to measure the neutron of 28 msr, a momentum acceptance of 20% and a resolu-
magnetic form factor [6]. tion p/p 104 (spectrometer A [15]). In coincidence
In general, this procedure requires a separation of the with the electron, the scattered nucleons are identied as
longitudinal RL G2E,p + G2E,n and the transverse cross proton or neutron in a well shielded scintillator array. As
d(e, e n) and d(e, e p) yields are measured simultaneously,
section RT G2M,p + G2M,n and a subsequent subtrac-
the ratio Rd is independent of uctuations in luminosity
tion of the proton contribution. In addition to this com-
and acceptance of the electron detector. The main exper-
plex procedure the inuence of nuclear binding introduces
imental di culty is the absolute calibration and monitor-
model dependences even in case of the deuteron. At mo-
ing of the neutron detection e ciency which enters di-
mentum transfers above 1 2 (GeV/c)2 the eect of GM,n
rectly in the ratio Rd . A calibration using the kinemati-
is large enough to employ this technique reliably [7].
cally complete p(n, p)n reaction to tag neutrons requires
The subtraction of the substantial proton contribution
measurements under dierent experimental conditions at
can be avoided by measuring the scattered neutron in co-
a facility providing intense neutron beams (e.g. [16]). Con-
incidence with the electron [8]. In order to achieve ac-
siderable care has to be taken to monitor eective detec-
ceptable signal-to-noise ratios in such coincidence experi-
tion thresholds and to ensure portability of the measured
ments, electron beams with a high duty factor are essen-
e ciencies [13,14]. In contrast, Bruins et al. [12] used the
tial. The di culty is then shifted to the experimental task
p(, + n) to calibrate their neutron detector in situ. How-
to calibrate and monitor the neutron detection e ciency.
ever, reactions from electroproduction p(e, + n)e , where
In addition, the sensitivity to nuclear structure can be
the exact kinematical correlation is lost in the 3 body -
signicantly reduced by measuring the cross section ratio
nal state, may lead to an underestimation of the detection
(d(e, e n)) e ciency. This has been suggested [17,18] as origin of the
Rd = (5) 10% discrepancy in the extracted values for GM,n which
(d(e, e p))
 M. Ostrick: Electromagnetic form factors of the nucleon 83

y
reaction plane

x pe pn

z
q R
pe pp
d(e,en), Bates 93
Rn, NIKHEF 94
Rn, ELSA 95 nq
Rn, MAMI 98, 02
3
He(e,e), Bates 94
3
He(e,e), Jlab 00
electron scattering plane
3
He(e,e), Jlab 03 Fig. 3. Reference frame and kinematics of the d(e, e n) reac-
tion.
2 2
Q /(GeV/c)
where the contribution of GE,p to the unpolarised cross
Fig. 2. The neutron magnetic form factor in units of n GD section is kinematically suppressed. The increased sensi-
as function of Q2 measured in coincidence and polarisation tivity of double polarisation observables to GE,n and GE,p
experiments [8, 11, 12, 13, 14, 19, 20, 21]. at high Q2 has already been pointed out more than 40
years ago [23,24,25].
For the ideal case of scattering longitudinally polarised


electron o free nucleons, N (
are summarised in g. 2 in units of the empirical dipole
e , e N ), the components of
expression (eq. (3)).
the recoil polarisation are given by
An alternative method to determine GM,n is provided
by inclusive scattering of polarised electrons from po- 
2 (1 ) GE GM
larised 3 He in quasi-elastic kinematics [19,20,21]. Results Px = Pe , (6)
obtained with this technique at Bates and Jeerson Lab  G2E + G2M
are included in g. 2 as well as absolute d(e, e n) cross Py = 0 , (7)
section measurements from Bates [8].
1  GM
2 2
Recently, new measurements of Rd at Q2 values up to Pz = P e . (8)
5 (GeV/c)2 have been completed at Jeerson Lab [22].  G2E + G2M
The large solid angle covered by the CLAS spectrome-
They are equivalent to cross section asymmetries with re-
ter allows to perform the e ciency determination simul-
spect to the beam helicity for the dierent nucleon spin
taneously with the Rd measurement by tagging neutrons
orientations in the scattering from polarised targets:
in the p(e, e + )n reaction where both, scattered electron
and + are detected. Preliminary results show, that GM,n Ax = Px ; Ay = 0; Az = P z . (9)
follows the dipole approximation up to Q2 = 5 (GeV/c)2
within 10%. The x and z direction are dened by the electron scatter-
ing plane with z given by the momentum transfer q (see
g. 3).
3 Double-polarisation observables In Px and Ax both form factors enter linearly which
increases the sensitivity compared to the unpolarised cross
Experiments using polarised electron beams in combina- section, if G2E  G2M .
tion with polarised nucleons either in the initial or nal In case of the neutron, the free e-n scattering has again
state oer possibilities to measure interferences between to be approximated by the quasi-free scattering o light
longitudinal and transverse amplitudes which do not ap- nuclei (2 H, 3 He) and one has to pay attention to nuclear
pear in the unpolarised cross section. This is particularly binding and rescattering eects. In leading order, spectral
interesting in cases where one part is completely dom- functions cancel in the polarisation and asymmetry com-
inating and unpolarised cross section measurements are ponents being ratios of cross sections. However, higher
not su cient to separate additional small amplitudes. Fur- order eects like FSI and MEC as well as inuences of
thermore, most polarisation observables are insensitive to Fermi-motion on the projections of polarisation compo-
absolute luminosities and other experimental calibration nents have to be taken into account.
factors. Polarised 3 He can be used as an eective polarised neu-
In electron-neutron scattering for example, the small- tron target because in its ground state the two protons are
ness of the electric form factor GE,n compared to the dominantly in the s-state with the spins coupled to zero.
dominant magnetic form factor makes a reliable Rosen- Thus the spin of the 3 He is predominantly carried by the
bluth separation impossible. As mentioned above, the sit- neutron. Additional d-wave components, meson exchange
uation is similar for protons at high momentum transfer currents and nal state interactions have recently been
84 The European Physical Journal A

transverse polarisation nq = 0 and has to be taken into account in the determi-

nation of GE,n (see section 4.3).
0.2 o
R=180
4 Double-polarisation experiments
0
The realisation of double-polarisation experiments des-
o
2
Q =0.32 GeV /c
2 2 R =0 cribed in the previous section demands the technically
sophisticated combination of continuous, high intensity,
0.2 polarised electron beams with polarised targets or recoil
0 5 10 15
polarimetry. At MAMI such experiments have been per-
formed since the beginning of the 1990s.

0.2 o
R=180 4.1 Polarised electrons at MAMI

In 1992 a spin-polarised electron beam was accelerated

0 through MAMI for the rst time. The electron source was
o based on photoemission of GaAsP illuminated by circu-
2
Q =0.12 GeV
2
/c
2 R =0
0.2
0 5 10 15
nq in deg
Fig. 4. Dependence of transverse neutron polarisation in the
d(
e ,e
n ) reaction on the neutron kinematics for two dierent
values of Q2 . nq is the angle between the neutron recoil mo-
mentum and momentum transfer. The dashed curves indicate
the in uence of the Fermi-motion, the solid curves are results
of a calculation by Arenhovel [27, 28] including FSI and further
higher-order contributions.
analysed at low Q2 within full three body calculations [26].
The dominant correction, that has to be applied in analy-

ses of 3 H e(
e , e n) experiments, originates from scattering
o protons followed by a charge-exchange reaction simu-
lating quasifree n(e, e n) events (see sect. 4.3).
In the d(
e , e
n ) reaction, the neutron recoil momen-
tum p n in general deviates from the direction of momen-
tum transfer q due to Fermi motion as indicated by the
angles nq and R in g. 3. At nite angles nq the trans-
verse polarisation Pt of the recoiling neutron gets admix-
tures from the Pz component. This is demonstrated in
g. 4, where Pt is plotted as a function of the angle nq
for the two extreme situations R = 0 and R = 180 .
The dashed curves indicate the admixture of Pz due to
this purely kinematical eect which averages out if the
detector acceptance is completely symmetric in R .
In studying the details of the d( e , e
n ) reaction,
Arenhovel et al. have shown that meson exchange and
isobar currents have a negligible eect in quasifree kine-
matics as does the choice of the N N -potential so that
there is essentially no dependence on the deuteron wave
function [27,28]. However, at momentum transfers below
Q2 = 0.25 (GeV/c)2 a strongly rising inuence of nal
state interactions, especially of charge exchange reactions,
is found (solid curves in g. 4). This leads to a shift in
the observed polarisation component Px even in the case
larly polarised laser light [29]. The helicity sign of the
laser and consequently of the electron beam was ipped
at a rate of 1 Hz by reversing the high voltage of a Pockels
cell. In order to have longitudinally polarised electrons at
the experiment the spin precession in the magnetic elds
of accelerator and transfer beamlines has to be compen-
sated. As there are no depolarising resonances in a mi-
crotron, this can be done at low energies before acceler-
ator injection or by slightly tuning the beam energy [30,
31]. In the beginning, a polarisation of about 30-35% at
beam currents of 5-10 A has been achieved. The use of
strained layer GaAs cathodes increased the polarisation
signicantly [32]. The lower quantum e ciency could be
compensated by increasing the laser power and the trans-
fer e ciency into the accelerator.
The beam polarisation is measured and monitored
using the spin dependence of Mott , Moeller and
Compton scattering.
Today, high intensity (up to 80 A), highly polarised
(Pe 80%) beams with sophisticated monitoring systems
are available which allow to measure even tiny parity-
violating asymmetries [33,34].
4.2 Polarisation transfer to protons
In one of the early double polarisation experiments at
MAMI the spin transfer to protons in the p( e , e
p)

and d( e , e p ) reactions has been analysed for the rst
time [35]. The transverse polarisation component Px
(eq. (7)) is accessible experimentally by measuring an
asymmetry in the azimuthal angular distribution ( ) of
protons scattered in a carbon analyser:
A( ) = aT sin  = Pe ApC Px sin  . (10)
The analysing power ApC of inclusive proton-carbon scat-
tering is known and has frequently been used for spin
analyses at proton facilities. With the number of events
N ( ) for both helicity states of the electron beam, this
 M. Ostrick: Electromagnetic form factors of the nucleon 85

Px in percent
1.2 Pospischil et al., MAMI(2001)
Milbrath et al., Bates (1999)
Jones et al., JLab (2000)
40 Simon et al., Mainz (1980)

G Ep / (G Mp / p)
1.1
30

1
20

10
eff
0.9
Px
0
0 2 4 6 8 10 12 0.35 0.4 0.45 0.5
c.m.
np Q 2 (GeV 2/c 2)

Fig. 5. Transverse proton polarisation Px measured in Fig. 6. The ratio p GE,p /GM,p from polarisation transfer
p(
e ,e p ) (square) and d( e ,e

p ) (circles) at Q2 = measurements at low Q2 [36, 37, 38]. The dotted line corre-
0.3 (GeV/c)2 [35]. In case of the deuteron target the observed sponds to the exact scaling behaviour of eq. (4), the solid line
dependence on the angle np of the proton-neutron relative is a t to Rosenbluth separated data [3].
momentum is shown. The line is a calculation of Arenhovel et
al. [27, 28].

azimuthal asymmetry A( ) can be determined through

the ratio

1 A( ) N + ( ) N ( + )
= , (11)
1 + A( ) N ( ) N + ( + )

which is insensitive to detector e ciencies and luminosity

uctuations.
The detector system was completely non magnetic con-
sisting out of a segmented lead-glass calorimeter for the
electrons and a scintillator hodoscope including the car-
bon analyser for the proton detection and spin analysis.
In the kinematics chosen for the experiment the depen-
dence of Px on GE,p is weak and inuences of binding and
rescattering eects can be tested. The measured polarisa-
tion transfer Px at Q2 = 0.3 (GeV/c)2 is shown in g. 5.
No signicant dierence in the spin transfer between free
protons and protons bound in deuterons was observed at
Q2 = 0.3 (GeV/c)2 in quasifree kinematics. In case of the
deuteron target the results are in agreement with calcula-
tions of Arenhovel et al. [27,28].
The full power of measuring the polarisation transfer
to protons has been demonstrated in experiments using
polarimeters consisting out of tracking detectors in front
and behind a carbon analyser sitting in the focal plane of
a magnetic spectrometer. If the spin precession along the
proton path is accurately taken into account, in principle
all three polarisation components of the recoiling proton
are accessible.
In case of elastic scattering the ratio Px /Pz is directly
proportional to the ratio of the charge and magnetic form
factors:
Px 2 GE
=  . (12)
Pz (1 + ) GM
Q 2 /(GeV/c) 2
Fig. 7. The ratio p GE,p /GM,p from polarisation trans-
fer measurements (triangles, [38, 39]) compared to results ex-
tracted from Rosenbluth separation [40, 41].
Experimental calibration factors as the absolute value of
the beam polarisation or the eective analysing power of
the polarimeter cancel in this ratio.
The proton charge form factor has been measured us-
ing this technique at Bates [36] and MAMI [37] at low Q2
(g. 6) as well as in Hall A at Jeerson Lab [38,39] (g. 7).
The measurements at JLab covered a Q2 range from 0.5
to 5 (GeV/c)2 and enormously inuenced our knowledge
about GE,p . The observed linear decrease of the ratio
p GE,p /GM,p at Q2 > 1 (GeV/c)2 contradicts the pre-
viously assumed approximate scaling behaviour (eq. (4))
and is in clear disagreement with the results obtained by
LT separations of unpolarised cross sections.
An experimental origin of this discrepancy has re-
cently been excluded by a new dedicated Rosenbluth
extraction which is in agreement with the earlier re-
sults [41]. The eect of GE,p in the cross section is so
small, that unknown  dependent corrections to the
86 The European Physical Journal A

one-photon exchange approximation may have a similar

size and could disturb the Rosenbluth extraction. In
general, two-photon exchanges are suppressed by the
electromagnetic coupling em 1/137 and they are Electron Detector
partly taken into account in radiative corrections [42]. 16x16 array of
However, there exist contributions which depend on the leadglassdetectors
hadronic structure and on intermediate excited states of
the nucleon. Calculations of these corrections are model _~ 100 msr
dependent and they have been neglected in the past. A
recent discussion of two-photon eects and the hadronic
target
physics involved can be found in [43].
e 1m
4.3 Measurements of GE,n /GM,n
concrete shielding
The vanishing charge of the neutron makes any small elec-
tric interaction enormously di cult to detect. Attempts
to measure rst moments of a charge distribution orig-
inate in the idea of Fermi and Marshal to study the
scattering of thermal neutrons o atomic electrons [44]. Magnet
These experiments have been rened and a negative mean
square charge radius close to the so called Foldy term
3/2Mn2 = 0.126 fm2 has been established [45,46]. How-
ever, a precise determination is still suering from system-
atic uncertainties due to the dominating nuclear scattering
amplitude [47,48].
At Q2 > 0 the smallness of G2E,n compared to G2M,n
makes a reliable Rosenbluth separation impossible. Finite
values for GE,n have been extracted from the deuteron

structure function A(Q2 ), measured in elastic electron-
deuteron scattering. A(Q2 ) provides sensitivity to GE,n
through the mixed term GE,p GE,n in the square of the plasticscintillators
isoscalar form factor (GE,n + GE,p )2 . However, the neces-
sary unfolding of the deuteron wave function introduces
substantial model dependences ([49,50] and g. 10).


The rst 3 H e( e , e n)pp and D(
e , e

n )p experiments
at MAMI have been performed with one common large
NeutronDetector
solid angle detector system (g. 8) [51,52,53,54]. The scat- and Polarimeter
tered electrons were detected in a segmented lead-glass
calorimeter with an energy resolution E/E 25% su -
cient to suppress inelastic events from -production. Only
the electron angles entered the reconstruction of the 3-
10
A in %

body nal state, which became kinematically complete

through the measurement of the neutron time of ight and
hit position in arrays of plastic scintillators well shielded
by concrete and lead. 5
In the d(e , e

n ) reaction, the polarisation of the re-
coiling neutron perpendicular to its momentum can be
analysed using the detection process itself. n-p-scattering
as well as inelastic processes, e.g., 12 C(n, n p)11 B, which 0
contribute to a neutron detection in a plastic scintillator, 0 50 100 150
provide reasonable analysing power Ae (n , Tn ). The re-
sulting asymmetry in the azimuthal angular distribution,
in degree
N ( ), of the detected neutrons can be observed through
the hit distribution in a second scintillator wall (g. 8)
and analysed via eq. (11). Fig. 8. Detector-setup for the rst double polarisaton exper-
In front of the analyser a dipole magnet has been in- iments at MAMI and a typical asymmetry in the azimuthal
stalled which allows to avoid an external calibration of the angular distribution of neutrons detected in both scintillator
eective analysing power through a controlled precession walls calculated via eq. (11).
 M. Ostrick: Electromagnetic form factors of the nucleon 87

a T () in % 0.12
GE,n ,
D(e,e n), Bates
A eff= 20% 0.1
,
D(e,e n), MAMI, A3
3 ,
He(e,e n), MAMI, A1
3 ,
5 0.08 He(e,e n), MAMI, A3
,
D(e,e n), NIKHEF

A eff= 8% 0.06
0
0.04

5
0 0.02

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
10
100 50 0 50 100 2 2
Q / (GeV/c)
in degree
Fig. 10. First GE,n results from double-polarisation observ-
Fig. 9. Measured azimuthal asymmetries a () for various ables. The arrows indicate the inuence of few-body eects
precession angles with two dierent cuts on the analysing mainly due to nal state interactions. The shaded area repre-
reaction leading to dierent analysing powers but leaving 0 sents the model dependence of GE,n values extracted from elas-
unchanged. tic D(e, e ) experiments [50] The dashed line is the parametri-
sation of Galster et al. [49].

of the neutron spin. This technique has become standard

in modern neutron polarimeters.
After precession by the angle
 lators serve in this case as neutron detector and time of
n
= B(l)dl (13)
2 n c L
the transverse neutron polarisation as well as the resulting
azimuthal asymmetry a become a superposition of x and
z components:
a () = Pe Ae (Px cos Pz sin ) (14) d(e , e
= a0 sin( 0 ). 0.3 (GeV/c) has been resolved in full 3-body calcula-
One immediately nds that the angle 0 of the zero cross-
ing a (0 ) = 0 is directly related to the ratio Px /Pz
(eq. (12)) and depends neither on the analysing power
of the polarimeter nor on the polarisation of the electron
beam,
Pe Ae Px
tan 0 = . (15)
Pe Ae Pz
Measured asymmetries for various precession angles are
shown in g. 9. Kinematic cuts on the analysing reaction
change the amplitude, i.e. the eective analysing power,
but not the zero crossing angle 0 .
To extract values for GE,n binding eects mainly due
to nal state interactions have been taken into account
according to calculations of H. Arenhovel [27,28]. At low
momentum transfer (Q2 = 0.12 (GeV/c)2 ) a correction
of almost 100% is required which drops rapidly to 8% at
Q2 = 0.35 (GeV/c)2 (see g. 4, [54]).

In case of the 3 H e(e , e n) reaction the 3 He gas is
polarised by optical pumping a metastable excited state
which then transfers the polarisation to the ground state.
In the rst experiments, after optical pumping, the gas
was compressed to 1bar in the target cell [51,52]. Today
typical target polarisations of about 50% at pressures up
to 5 bar are achieved [55]. For dierent orientations of the
target spin cross section asymmetries with respect to the
beam helicity are measured. The arrays of plastic scintil-
ight spectrometer only, not as polarimeter.
In 1998 no full 3 body calculations were avail-

able and the rst 3 H e( e , e n) experiments have been
analysed under the assumption of quasifree scattering
from a neutron with no higher order eects taken into
account [51,52]. The initial 50% discrepancy between

n ) and 3 H e(e , e n) experiments around Q2 =
2

tions of the 3 H e( e , e n) reaction including nal state
interactions [26]. At higher Q2 , binding and rescatter-
ing eects are expected to decrease signicantly as for
deuterium [55]. A detailed discussion of polarised 3 He
targets and their use in experiments at MAMI can be
found in [56].
Figure 10 summarises results for GE,n obtained in the
rst double-polarisation experiments. The arrows indicate
the necessary corrections due to nal state interactions.

In the meantime, d ( e , e
e , e n), d(
n ) and 3 H e(
e,
e n) reactions have been measured several times with re-
ned techniques. At MAMI the use of magnetic spec-
trometers to detect the scattered electron improved back-
ground suppression and kinematical reconstruction con-
siderably [57,55,58]. In particular, the direction of the mo-
mentum transfer vector q , which denes the relevant co-
ordinate system for the spin analysis, can be reconstructed
precisely. At Jeerson Lab the Q2 range has been extended
up to 1.5 (GeV/c)2 [59,60]. Below Q2 = 1 (GeV/c)2 ,
new preliminary data obtained with the Blast-detector at
Bates have recently been shown [61].
Furthermore, GE,n has been extracted from an analy-
sis of the deuteron quadrupole form factor FQ obtained
from recent tensor polarisation measurements in elas-
tic electron deuteron scattering [62]. Compared to the
88 The European Physical Journal A

G E,n r 2 (r) fm
Bates
0.015
3
MAMI He(e,e n) MAMI
0.1 D(e,en)
JLAB
0.01
NIKHEF D(e,e)D
0.08 D(e,en)
JLAB
0.005
0.06

0
0.04

0.005
0.02

0 0 1 2 3 4 5
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Q 2 / (GeV/c) 2 r/fm
Fig. 11. Present status of GE,n measurements compared to Fig. 12. Neutron charge distribution obtained from Fourier
the parametrisation of Galster et al. ([49], dashed line) and transforms of GE,n by Friedrich and Walcher (solid line [63])
Friedrich, Walcher ([63], solid line). The data are from experi- and the Galster parametrisation (dashed line).
ments using polarised Deuterium [64, 65, 59], polarised 3 He [55,
26, 58], recoil polarisation [66, 54, 60], and from an analysis of
the elastic deuteron quadrupole form factor [62]. imately 0.2 (GeV/c)2 as well as a corresponding bump in
GE,n are clearly revealed [63].
Nonrelativistically, if the Compton wavelength of a
above-mentioned older analyses of the elastic deuteron system
 is negligible compared to its size C = h/M c 
structure functions A [50], the model uncertainties are re-
duced by the direct use of FQ .
Taking all these novel approaches together, a consis-
tent picture of the charge form factor of the neutron is
starting to arise (see g. 11). Results obtained with dif-
ferent targets and in dierent reactions are in fair agree-
ment with each other even though at low Q2 substantial
corrections due to rescattering are unavoidable. The data
r2 , form factors can be measured over a su ciently
large range in momentum transfer in order to calculate
spatial densities by a Fourier transform without relativis-
tic eects becoming important. This is the case for atomic
nuclei and detailed information about nuclear charge dis-
tributions has been obtained from electron scattering [67].
Presently, also the extraction of mass or neutron densities
are discussed [68,69]. 
roughly follow a phenomenological parametrisation given
by Galster et al. already in 1971 [49]. However, the accu-
racy is reaching a level at which deviations from such a
simple, smooth behavior start to become signicant [63].
5 Interpretation
With modern experimental techniques, for the rst time
all elastic nucleon form factors, including the neutron
charge form factor, have been measured precisely over a
nite range in momentum transfer.
Both magnetic form factors, GM,p and GM,n , fol-
low the dipole approximation within 10% up to Q2 =
5 (GeV/c)2 . The scaling relation (eq. (4)) is violated
considerably for the proton electric form factor. The al-
most linear decrease of the ratio GE,p /GM,p at Q2 >
1 (GeV/c)2 , as revealed by spin transfer measurements
at JLab, indicates that the charge density of the proton is
signicantly softer than its magnetisation density.
Friedrich and Walcher emphasised local deviations
from a smooth shape in the Q2 dependence of all four
form factors. By tting the available data with an ansatz
given by the sum of a Gaussian and two dipoles dis-
cribing the smooth part, local minima in GE,p , GM,p and
GM,n around Q2 = 0.25 (GeV/c)2 with a width of approx-
Although for nucleons C 0.25 r2 , a similar in-
terpretation of GE (Q2 ) and GM (Q2 ) as momentum repre-
sentations of spatial charge and magnetisation densities is
still possible in the Breit frame of vanishing energy trans-
fer. Figure 12 shows the corresponding charge distribution
of the neutron as calculated by Friedrich and Walcher from
their ts.
In coordinate space, the structures observed around
Q2 = 0.25 (GeV/c)2 inuence the long distance tail
(r 1.5 2 fm) of charge and magnetisation densities and
may be interpreted as resulting from a pion cloud sur-
rounding a bare nucleon. Within this picture, the data for
all four form factors can be described by an intuitive phe-
nomenological ansatz consisting of dipole functions for the
constituent quarks together with a p-shell harmonic oscil-
lator behaviour of the pion cloud [63].
Another method to analyse and interpret form fac-
tors, which does not directly refer to a particular model
for nucleon structure, is based on dispersion relations in
Q2 . They provide a mathematical framework to connect
experimental date in spacelike (Q2 > 0) as well as in
timelike (Q2 < 0) regions with spectral functions describ-
ing the spectrum of virtual intermediate states, through
which a photon can couple to a nucleon [70]. Already
early form factor data have been analysed systematically
in this framework and it has been established that the
spectral functions can be approximated by poles due to
 M. Ostrick: Electromagnetic form factors of the nucleon
the existence of vector mesons and their coupling to nu-
cleons. The prediction of the (770) meson was an early
success of this approach [71].
Rened analyses demonstrated the importance of non
resonant multi-pion intermediate states [72,73,74]. Two-
and tree-pion systems are the lightest possible intermedi-
ate states. They provide a link to pion-nucleon scattering
and to model-independent predictions from chiral pertur-
bation theory. The analysis method based on dispersion
relations as well as the inuence of recent data on the
spectral functions is discussed by H.W. Hammer [75].
6 Conclusions
In facilities like the Mainz Microtron MAMI, high inten-
sity, polarised, continuous electron beams in the energy
range relevant to study phenomena at hadronic scales
are available and can be combined with polarised targets
and sophisticated detector systems for coincidence exper-
iments and polarimetry.
Electromagnetic form factors are signicant observ-
ables, directly related to the spatial structure of the nu-
cleon. For the rst time, all four nucleon form factors have
been measured with a precision su cient to identify local
structures in the Q2 dependence at a few percent level.
In the near future, measurements of GE,p /GM,p and
GE,n will be extended to higher values of Q2 at Jeerson

Lab. Below Q2 = 2 (GeV/c)2 , new 3 H e( e , e n) as well as
 28. H. Arenhovel et al., Phys. Rev. C 52, 1232 (1995).
absolute p(e, e ) cross section measurements are planned
at MAMI [76,77].
Besides electromagnetic form factors, a deeper under-
standing of the elastic nucleon response also includes the
weak vector and axial-vector currents. The nucleon axial
form factor at low Q2 has recently been measured in pion
electroproduction at MAMI [78] and present-day exper-
iments in parity-violating electron scattering provide ac-
cess to the two weak vector form factors, which will allow
a avour decomposition of the charge and magnetisation
distributions in the nucleon [34,69].
Also beyond elastic scattering, the experimental tech-
niques discussed above help to fully exploit the properties
of electromagnetic probes for studies of the much poorer
known structure and dynamics of resonances in inelastic
processes [79,80].
I would like to thank the organisers of the symposium 20 Years
of Physics at the Mainz Microtron MAMI, Hartmut Arenhovel,
Hartmut Backe, Dieter Drechsel, Jorg Friedrich, Karl-Heinz
Kaiser and Thomas Walcher and express all the best wishes
for the future.
References
1. E.M. Lyman, A.O. Hanson, M.B. Scott, Phys. Rev. 84,
626 (1951).
2. R. Hofstadter, R.W. McAllister, Phys. Rev. 98, 217 (1955).
3. G.G. Simon et al., Nucl. Phys. A 333, 381 (1980).
4. L. Andivahis et al., Phys. Rev. D 50, 5491 (1994).
89
5. C.E. Hyde-Wright, K. de Jager, Annu. Rev. Nucl. Part.
Sci. 54, 217 (2004).
6. H.F. Ehrenberg, R. Hofstadter, Phys. Rev. 110, 544
(1958).
7. A. Lung et al., Phys. Rev. Lett. 70, 718 (1993).
8. P. Markowitz et al., Phys. Rev. C 48, R5 (1993).
9. W. Fabian, H. Arenhovel, Nucl. Phys. A 314, 253 (1979).
10. M. Schwamb, these proceedings.
11. H. Anklin et al., Phys. Lett. B 336, 313 (1994).
12. E.E.W. Bruins et al., Phys. Rev. Lett. 75, 21 (1995).
13. H. Anklin et al., Phys. Lett. B 428, 248 (1998).
14. G. Kubon et al., Phys. Lett. B 524, 26 (2002).
15. K.I. Blomqvist et al., Nucl. Instrum. Methods A 403, 263
(1998).
16. J. Arnold et al., Nucl. Instrum. Methods A 386, 211
(1997).
17. J. Jourdan, I. Sick, J. Zhao, Phys. Rev. Lett. 79, 5186
(1997).
18. E.E.W. Bruins et al., Phys. Rev. Lett. 79, 5187 (1997).
19. H. Gao et al., Phys. Rev. C 50, R546 (1994).
20. W. Xu et al., Phys. Rev. Lett. 85, 2900 (2000).
21. W. Xu et al., Phys. Rev. C 67, 012201 (2003).
22. W.K. Brooks, J.D. Lachniet, Nucl. Phys. A 755, 261
(2005).
23. A.I. Akhiezer et al., Sov. Phys. JETP 6, 588 (1958).
24. N. Dombey, Rev. Mod. Phys. 41, 236 (1969).
25. R.G. Arnold, C.E. Carlson, F. Gross, Phys. Rev. C 23, 363
(1981).
26. J. Golak et al., Phys. Rev. C 63, 034006 (2001).
27. H. Arenhovel et al., Z. Phys. A 331, 123 (1988).
29. K. Aulenbacher et al., Nucl. Instrum. Methods A 391, 498
(1997).
30. K.H. Steens et al., Nucl. Instrum. Methods A 325, 378
(1993).
31. V. Tioukine et al., contribution to the 8th European Par-
ticle Accelerator Conference (EPAC 2002), Paris, France,
3-7 June 2002.
32. P. Drescher et al., Nucl. Instrum. Methods A 381, 169
(1996).
33. A. Jankowiak, these proceedings.
34. F. Maas, these proceedings.
35. D. Eyl et al., Z. Phys. A 352, 211 (1995).
36. B.D. Milbrath et al., Phys. Rev. Lett. 80, 452 (1998).
37. T. Pospischil et al., Eur. Phys. J. A 12, 125 (2001).
38. M.K. Jones et al., Phys. Rev. Lett. 84, 1398 (2000).
39. O. Gayou et al., Phys. Rev. Lett. 88, 092301 (2002).
40. J. Arrington, Phys. Rev. C 69, 022201 (2004).
41. I.A. Qattan et al., Phys. Rev. Lett. 94, 142301 (2005).
42. L.W. Mo, Y.S. Tsai, Rev. Mod. Phys. 41, 205 (1969).
43. M. Vanderhaeghen, these proceedings.
44. E. Fermi, L. Marshal, Phys. Rev. 72, 1139 (1947).
45. S. Kopecki et al., Phys. Rev. C 56, 2229 (1997).
46. Yu.A. Alexandrov et al., Phys. Part. Nucl. 30, 29 (1999).
47. H. Leeb, C. Teichtmeister, Phys. Rev. C 48, 1719 (1993).
48. Yu.A. Alexandrov, Phys. Rev. C 49, 2297 (1994).
49. S. Galster et al., Nucl. Phys. B 32, 221 (1971).
50. S. Platchkov et al., Nucl. Phys. A 510, 740 (1990).
51. M. Meyerho et al., Phys. Lett. B 327, 201 (1994).
52. J. Becker et al., Eur. Phys. J. A 6, 329 (1999).
53. M. Ostrick et al., Phys. Rev. Lett. 83, 276 (1999).
54. C. Herberg et al., Eur. Phys. J. A 5, 131 (1999).
55. D. Rohe et al., Phys. Rev. Lett. 83, 4257 (1999).
90 The European Physical Journal A

56. D. Rohe, these proceedings. 70. G. Hohler et al., Nucl. Phys. B 114, 505 (1976).
57. D.I. Glazier et al., Eur. Phys. J. A 24, 101 (2005). 71. W.R. Frazer, J.R. Fulco, Phys. Rev. Lett. 2, 365 (1959).
58. J. Bermuth et al., Phys. Lett. B 564, 199 (2003). 72. P. Mergell, U.G. Meissner, D. Drechsel, Nucl. Phys. A 596,
59. G. Warren et al., Phys. Rev. Lett. 92, 042301 (2004). 367 (1996).
60. R. Madey et al., Phys. Rev. Lett. 91, 122002 (2003). 73. H.W. Hammer, U.G. Meissner, D. Drechsel, Phys. Lett. B
61. R. Alarcon et al., contribution to the 16th International 385, 343 (1996).
Spin Physics Symposium (SPIN 2004), Trieste, Italy, 10- 74. H.W. Hammer, D. Drechsel, U.G. Meissner, Phys. Lett. B
16 Oct 2004. 586, 291 (2004).
62. R. Schiavilla, I. Sick, Phys. Rev. C 64, 041002 (2001). 75. H.W. Hammer, these proceedings
63. J. Friedrich, Th. Walcher, Eur. Phys. J. A 17, 607 (2003). 76. M.O. Distler (contact person) et al., Experiment MAMI
64. I. Passchier et al., Phys. Rev. Lett. 82, 4988 (1999). A1-2/2005.
65. H. Zhu et al., Phys. Rev. Lett. 87, 081801 (2001). 77. M.O. Distler, W. Heil, D. Rohe (contact persons) et al.,
66. T. Eden et al., Phys. Rev. C 50, 1749 (1994). Experiment MAMI A1-1/2005.
67. J. Friedrich, N. Voegler, Nucl. Phys. A 373, 219 (1982). 78. A. Liesenfeld et al., Phys. Lett. B 468, 20 (1999).
68. B. Krusche, Eur. Phys. J. A 26, 7 (2005). 79. R. Beck, these proceedings.
69. S. Kowalski, these proceedings. 80. H. Schmieden, these proceedings.
Eur. Phys. J. A 28, s01, 91 100 (2006)
DOI: 10.1140/epja/i2006-09-010-9 EPJ A direct
electronic only

Photo- and electro-excitation of the -resonance at MAMI

H. Schmiedena
Physikalisches Institut, Universitat Bonn, Germany

/
Published online: 15 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. Over the last decade accurate experiments at MAMI played an essential role to improve our
understanding of the nucleon to (1232) transition. Originally to a large extent motivated through intra
quark hyperne interactions anticipated in QCD-inspired quark models they showed that pionic degrees
of freedom are essential. The meson cloud is mainly responsible for the observed quadrupole excitation
strength and aects the magnetic dipole transition strength as well.

PACS. 13.60.Le Meson production 13.40.-f Electromagnetic processes and properties 14.20.Gk Baryon
resonances with S = 0

1 Introduction
At very high energy and momentum transfers in deep in-
elastic lepton scattering, proton and neutron reveal their
substructure of pointlike, almost massless spin-1/2 con-
stituents, the quarks, and of gluons as the exchange bosons
mediating the color force between them. In this regime
of asymptotic freedom Quantum Chromodynamics is well
established as the basic underlying theory of strong inter-
action. However, at momentum transfers corresponding to
the nucleons size, the nonlinear strong couplings prohibit
the solution of the QCD eld equations using perturbation
theory. Hence, basic properties such as mass, size and exci-
Fig. 1. Photoexcitation of the N transition. Top: M1
tation spectrum are only qualitatively [1] understood and
photon generating a spin-ip transition; bottom: E2 or C2 pho-
remain still a domain of models [2]. To prove that QCD ton introducing angular momentum L = 2.
provides also the correct theory at the connment scale is
the challenge of Lattice calculations, currently stepping
beyond quenched approximations [3] by realistic light
quark vacuum polarisation and chiral quark actions [4,5]. -nucleon channel, perfectly tting the present nominal
Analogously to atomic spectroscopy at the threshold to MAMI energy and thus most suited for a detailed study.
the era of quantum mechanics, baryon spectroscopy serves From the viewpoint of the inner quark dynamics the
today as a tool to improve our understanding of the inner nucleon to (1232) transition is very interesting. In the
dynamics of the nucleon. Generally, the high level density SU (6) symmetric constituent quark model it corresponds
of excited states in combination with their short lifetimes, to a pure spin-ip of one of the quarks, yielding the spin
and thus large natural widths, provides an annoying ex- 3/2 of the (1232). Electromagnetic excitation requires
perimental obstacle. However, there are two states which thus an M1 magnetic dipole photon as schematically de-
can be experimentally almost exclusively prepared, out- picted in g. 1 (top). Parity and angular momentum con-
standing in cross section and well separated in mass from servation would alternatively allow the absorption of an
their neighbours. The N (1535)S11 negative-parity partner L = 2 photon, coupling together with the nucleons spin
of the nucleon which selectively couples to the -nucleon 1/2 to J = 3/2, cf. g. 1 (bottom). However, this re-
nal state. Its mass puts it at the very edge of the energy quires L = 2 quadrupole components in one or both of
range accessible with MAMI B. Contrary, the decuplet the nucleons and deltas quark wave functions, in anal-
ground state (1232)P33 couples almost entirely into the ogy to the deuteron in nuclear physics. As in the latter
case, the quadrupole components can be associated with
a
e-mail: schmieden@physik.uni-bonn.de a spherical deformation of the system. They originate from
92 The European Physical Journal A

tensor parts in the interaction of the constituents which,

in QCD-motivated models, are attributed to the color hy- t=nxl
perne interaction among the quarks [6,7]. y
e x 0
While most of our knowledge about the nucleon excita-
tion spectrum stems from pion-nucleon scattering experi- z
ments, those are insensitive to the quadrupole strength in q n=qxp p
e l
the nucleon to (1232) transition. Due to the positive par- p
ity of both nucleon and , the angular momentum L = 1 scattering plane (lab)
magnetic transition is generated in -N scattering, but
not the L = 2 one. The latter requires photons which
provide positive parity with both L = 1 and L = 2. reaction plane (cm)

For photon energies up to 2000 MeV, g. 2 shows Fig. 3. Kinematics of pion electroproduction. The vectors of
the total absorption cross section of real photons on the the reaction plane are in the rest frame of the .
proton. The (1232) resonance exhibits almost isolated
around E = 340 MeV, whereas the second and third
resonance regions are composed of numerous overlapping In its rest frame, the recoiling hadronic system decays
resonances which, in the total absorption cross section, back to back into 0 and proton. and p denote the
cannot be separated. Due to their short life time, the di- pion and proton angles with respect to q in the hadronic
rect detection of the resonances is precluded. Since the cm frame. The reaction plane, which is given by l = pcm p ,
(1232) almost exclusively decays into pion and nucleon, n = q p , and t = n l, is tilted against the electron
cm
pion photo- and electro-production o the nucleon pro- scattering plane by the angle .
vide the major experimental tools for the investigation of In one photon exchange approximation the vefold dif-
its properties. ferential cross section
d5 d2 v
cm
= (1)
2 Kinematics and cross section of pion photo- dEe de d dcm
and electro-production factorizes into the virtual photon ux,
The kinematics of pion electroproduction are shown in E  k 1
= , (2)
g. 3 at the example of the e+p e+p+ 0 reaction. The 2 2 E Q2 1 
electron is scattered by the laboratory angle e and the
exchanged virtual photon transfers the dierence between and d2 v /dcm , the virtual photon cm cross section.
incoming and scattered electron energies and momenta, denotes the ne structure constant, k = (W 2 m2p )/2mp
= E E  and q = k k , respectively. Q2 = q q > 0 the real photon equivalent laboratory energy for the exci-
represents the invariant mass of the virtual photon. The tation of the target with mass mp to the cm energy W ,
electron scattering plane is spanned by the unit vectors and  = [1+(2|q|2 /Q2 ) tan2 2e ]1 the photon polarisation
z = q = q/|q|, y = k k , and x = y z. parameter.
Without target or recoil polarisation, the virtual pho-
ton cross section is given by [12]
0.6 d2 v
= [RT + L RL + c+ RLT cos +
dcm
0.5 RT T cos 2 + Pe c RLT  sin ]. (3)

0.4 The ratio = |pcm |/kcm is determined by the pion cm

momentum pcm and kcm = (mp /W )k . The structure
/mb

0.3 functions, Ri , parameterize the response of the hadronic

tot

system to the various polarisation states of the photon

0.2
eld, which are described by the transverse and longitu-
2
dinal polarisation,  and L = Q2 , also contained within
 cm

0.1 the factors c = 2L (1 ). The degree of longitudinal

electron polarisation is denoted by Pe .
0.0 In lowest order, the tree graphs depicted in g. 4 con-
200 400 600 800 1000 1200 1400 1600 1800 2000
E /MeV tribute to the cross section in the (1232) resonance re-
gion. For 0 production the pion pole and the contact term
Fig. 2. World data of the real photon total absorption cross vanish, because the photon cannot couple to the charge-
section as a function of the photon energy [8]. The open and less pion. Thus, usually the 0 channel is chosen to tag
full circles represent the data from MAMI [9] and NINA [10]. the (1232) intermediate state. Nevertheless, the unam-
The parameterization after [11] is shown as the solid line. biguous tagging is hampered by the inevitable background
 H. Schmieden: Photo- and electro-excitation of the -resonance at MAMI 93

Table 1. Photon and pion multipoles up to L = 2.

-N system -N system
parity
L -multipole J l -multipole
0 C0 1/2 1 S1 +
Fig. 4. Lowest-order graphs of single-pion electroproduction.
1/2 0 E0+ /S0+
E1/C1
3/2 2 E2 /S2
from non-resonant production (cf. g. 4). Therefore, to
1 1/2 1 M1
extract the small quadrupole admixture, a full multipole
M1 +
analysis would be desirable in analogy to -N scattering. 3/2 1 M1+
However, in pion production experiments with real and,
3/2 1 E1+ /S1+
in particular, virtual photons this is much more di cult E2/C2 +
to achieve, because more invariant amplitudes need to be 5/2 3 E3 /S3
independently determined. 2 3/2 2 M2
M2
5/2 2 M2+
2.1 Multipole decomposition

The response functions R of eq. (3) can be decomposed as electric/coulombic (E/C) or magnetic (M ). The pion
into pion multipoles of the nal state. In S and P wave multipoles AIl are characterized through their magnetic,
approximation this yields [12,13]: electric or scalar (longitudinal) nature, A = M, E, S (L),
the isospin, I, and the pion-nucleon relative angular mo-
RL = 2 [ |S0+ |2 + 4|S1+ |2 + |S1 |2 mentum, l . The coupling of l with the nucleon spin to

4e{S1+ S1 } + 2 cos e{S0+ (4S1+ + S1 )} the total angular momentum, J, is indicated by . Omis-
sion of the isospin index as in eqs. (4-8) indicates ampli-
+12 cos2 (|S1+ |2 + e{S1+ S1 }) ], (4)
tudes in the p 0 charge channel throughout this paper.
1
RT = |E0+ |2 + |2M1+ + M1 |2 In general, the unambiguous tagging of a resonance re-
2 quires the multipoles to be determined in the appropriate
1 isospin channel. However, at the resonance position of the
+ |3E1+ M1+ + M1 |2
2 (1232) the p 0 amplitudes provide a very good approx-

+2 cos e{E0+ (3E1+ + M1+ M1 )} imation to the separated isospin components. Both meth-

ods, simultaneous measurement of 0 and + production
+ cos2 |3E1+ + M1+ M1 |2 and sole 0 production, have been exploited.
1
|2M1+ + M1 |2
2  3 Experimental methods and results
1
|3E1+ M1+ + M1 |2 } , (5)
2 The sensitivity to the N (1232) quadrupole ampli-

3 1 tudes is directly illustrated by the S and P wave multi-

RT T = 3 sin2 |E1+ |2 |M1+ |2 pole decomposition of the cross section in eqs. (4 8). All
2 2
 structure functions contain E1+ or S1+ amplitudes, either
quadratically or as bilinear combinations.
e{E1+ (M1+ M1 ) + M1+ M1 } , (6)
The smallness of the quadrupole amplitudes makes a

RLT = sin e{ S0+ (3E1+ M1+ + M1 ) determination of their square against the leading |M1+ |2
very di cult, e.g., a Rosenbluth separation of RL and RT
(2S1+ S1 )E0+ for the extraction of |S1+ |2 . It is much more promising

+6 cos (S1+ (E1+ M1+ + M1 ) to exploit interferences with the dominating M1+ multi-

+S1 E1+ ) }, (7) pole. Thus, the experiments based on unpolarised electron
beams extracted the quadrupole over dipole ratios
RLT  = sin m{ S0+ (3E1+ M1+ + M1 )

(2S1+ S1 )E0+ 0 e{E1+ M1+ }
EMR = (9)
+6 cos (S1+ (E1+ M1+ + M1 ) |M1+ |2

+S1 E1+ ) }. (8)
and
0 e{S1+ M1+ }
= cm /|qcm | is a current conservation factor related to CMR = (10)
|M1+ |2
the denition of the amplitudes. The photon and pion mul-
tipoles up to order L = 2 are summarized in table 1. Ac- from measurements of the RT T and RLT type structure
cording to their parity the photon multipoles are classied functions.
94 The European Physical Journal A

3.1 Real photon experiments which provide a very high sensitivity to the EMR. In or-
der to exploit those experimentally, the cos 2 azimuthal
With real photon beams, i.e. Q2 = 0, the longitudinal modulation of the cross section needs to be determined.
parts of the cross section vanish. Of eqs. (4 8) thus only A detector with cylinder symmetry is ideally suited for
RT and RT T contribute and the cross section is often writ- this purpose. The rst experiments of the A2 collabora-
ten in the form [12] tion at MAMI consequently used the DAPHNE setup to

2  detect the recoil protons from the + p 0 + p reaction,
d2 d schematically depicted in g. 5 [14,15].
= [1 P cos 2], (11)
dcm dcm 0
Results of such measurements are shown in g. 6. Ex-
d2 ploiting the cleanly measured azimuthal modulation (left
where [ d cm ]0denotes the unpolarised cross section and
part of gure) the photon beam asymmetries (right part)
P is the degree of linear polarisation of the photon can be determined as a function of polar angle over the
beam. The photon-beam asymmetry represents the ra- entire energy region of the (1232) resonance. In conjunc-
tio RT T /RT and thus enters the cross section with the tion with the simultaneously measured dierential cross
cos 2 modulation characteristic for the directional sensi- sections it is possible to perform a detailed mulipole anal-
tivity against the plane of linear polarisation. This plane ysis. Results are depicted in g. 7 [15,16]. Furthermore,
is intrinsically xed by kinematic constraints in electron including the results of the reaction + p + + n with
scattering, usually the detection of the nal-state electron linearly polarised photon beam enables an isospin decom-
under a certain angle. Contrary, real photon beams from position. At the resonance position it provides EMR re-
electron bremsstrahlung are a homogeneous superposition sults in agreement with the 0 channel alone. As nal re-
of all polarisation directions. The net polarisation conse- sult [15] the electric quadrupole to magnetic dipole ratio
quently vanishes.
To obtain linear polarised photon beams it is necessary
to x the electron scattering plane in the bremsstrahlung
process. This can be achieved by o-axis tagging or, as
exploited at MAMI, through coherent bremsstrahlung o
a diamond crystal. Under certain kinematic conditions de-
pending on the crystal alignment relative to the electron
beam the crystal lattice, similar to the Mossbauer eect,
coherently takes the bremsstrahlung recoil instead of indi-
vidual nuclei. This generates intensity peaks in the other-
wise  1/E distribution of the bremsstrahlung spectrum.
Within the energy region of the coherent peak the crys-
tal orientation determines a particular electron scattering
plane. Hence, the photon beam is linearly polarised and
the degree of linear polarisation is related to the intensity
excess over the incoherent spectrum. Fig. 6. Left: Measured relative cross section of the reaction
According to eq. (11), linear polarised beams with + p 0 + p using linearly polarised photon beam as a
P = 0 enable the contribution of the beam asymmetry function of the azimuthal angle at xed polar angle cm =
to the observed cross section. Its size is determined by 90 . Right: Polar angle dependence of the extracted photon

interferences of the type |M1+ |2 + 6 e{E1+ M1+ } + beam asymmetry.

Fig. 7. Multipole ts based on measurements of cross section

Fig. 5. The cylinder-symmetric DAPHNE detector (Figure and photon asymmetry using linearly polarised real photon
courtesy of MAMI-A2 collaboration). beams [16].
 H. Schmieden: Photo- and electro-excitation of the -resonance at MAMI 95

e{E1+ M1+ }
= (2.5 0.1stat 0.2syst ) % (12)
|M1+ |2
was obtained.
The beam asymmetry had also been extracted using
the large acceptance TAPS photon detector, covering the
full polar angular range for the 0 decay photons [17].
Despite the coplanar detector arrangement, could be
determined through the deliberate rotation of the photon
polarisation plane.

3.2 Virtual photon experiments

Due to the nonzero mass of the exchange photon, in elec-

troproduction also longitudinal pieces contribute, in total
four structure functions with unpolarised electron beam
and a fth one with longitudinally polarised beam. If ei-
ther a polarised target is provided in the initial state or
the nal state proton polarisation is determined, then 13
more structure functions enter the cross section, i.e. a to-
tal of 18. Not all of those are independent.
In addition to the EMR now also the CMR (cf.
eq. (10)) is accessible. As in the photoproduction case, suf-
cient sensitivity is only obtained through interferences of
the S1+ amplitude with the dominating M1+ . Fig. 8. Three-Spectrometer-Setup of the MAMI-A1 collabora-
tion. Spectrometer B (left) is lifted out of plane by 10 degrees
(see text). Picture courtesy of Markus Weis.
3.2.1 Unpolarised electroproduction
The sensitivity to S1+ can be seen from the decomposition
According to the discussion of sect. (3.1), in electron scat- of eq. (14) into the leading S and P partial waves. At the
tering a linearly polarised photon eld is already obtained (1232)-resonance position the asymmetry
without the need to polarise the electrons. Thus, the EMR

can be determined similar to real photon experiments with {(S0+ + 6S1+ 0 )M1+ }
cos cm
0 )  f ( 0 )
LT (cm cm
a large acceptance detector [18,19]. The kinematic focus- (15)
|M1+ |2
sing at large Q2 enables full centre-of-mass coverage us-
ing a relatively small acceptance in the laboratory, e.g. of is approximated. Thus measurements of LT in the for-
magnetic spectrometers [20]. At smaller Q2 the required ward (1 ) and backward cm-hemisphere (2 = 1 )
angular range must be covered by subsequent settings allow the extraction of S1+ /M1+ and, simultaneously, of
of the spectrometers. This technique has been used for S0+ /M1+ , i.e. the main contribution of inherent non
0
the MAMI experiments in order to determine the CMR resonant background:
from the longitudinal-transverse interference part of the

cross section. Since RLT comes along with a cos az- {S1+ M1+ }
= f1 (1,2 ) [LT (1 ) LT (2 )] + C1 , (16)
imuthal modulation, cf. eq. (3), it is su cient to measure |M1+ |2
the recoil protons of the p(e, e p) 0 reaction within the
{S0+ M1+ }
electron scattering plane, i.e. within the laboratory oor = f0 (1,2 ) [LT (1 ) + LT (2 )] + C0 . (17)
plane. At each 0 centre-of-mass polar angle cm |M1+ |2
0 , the co-

sine term exhibits through a cross section asymmetry in

measurements left ( = 0 ) and right( = 180 ) of the
direction of three-momentum transfer, q:
dv ( = 0 ) dv ( = 180 )
LT (cm
0 ) :=
dv ( = 0 ) + dv ( = 180 )
According to eq. (3), it is related to the partial responses
The functions f denote kinematic factors, C0 and C1 con-
tributions of multipoles beyond the simple approximation.
Using the 3-Spectrometer-Setup of the MAMI-A1 col-
laboration shown in g. 8, the asymmetries
. (13)
0 = 160 ) = (12.18 0.27stat 0.82sys ) %,
LT (cm
LT (0 = 20 ) = (11.68 2.36stat 2.36sys ) %
cm

via
have been measured [21]. They are shown in g. 9 along
c+ RLT with dierent calculations. The prediction of Sato and
LT (cm
0 ) = . (14) Lee [22] describes the asymmetries quite well, to a lesser
RT + L RL + RT T
96 The European Physical Journal A
Fig. 9. Left-right asymmetries LT measured in 0 production
with unpolarised electron beam compared to model predictions
from MAID2003 [23] (dotted), DMT2001 [24, 25] (dashed),
Sato/Lee [22] (dashed dotted). The full curve represents a
MAID2003 re-t [21]. The inner and outer errors are purely
statistical and quadratically summed statistical and systemat-
ical, respectively.
extent the calculation within the dynamical model of
Kamalov and Yang [25]. Also the standard MAID2003
parametrisation [23] provided only moderate agreement,
in particular at backward pion angle. This was resolved
through a re-t including also polarisation data as de-
scribed in the following subsect. 3.2.2.
Using the MAID re-t, the multipole ratios of eqs. (16)
and (17) were determined to
{S1+ M1+ }
= (5.45 0.42) %, (18)
|M1+ |2
{S0+ M1+ }
= (2.56 2.25) %. (19)
|M1+ |2
The S0+ result is compared to older measurements in
0
g. 10. It practically excludes that a large negative CMR
found by ref. [26] at Q2 = 0.12 (GeV/c)2 can be ex-
plained by the particular contribution of a large negative
S0+ /M1+  10 % [27] in forward pion kinematics.
3.2.2 Polarised electron beams: 5th structure function
RLT  , the 5th structure function of eq. (3), measures the
imaginary part of the same interference which RLT and
LT contain the real part of. This is obvious from com-
parison of the S- and P -wave decompositions in eqs. (7)
and (8). Experimentally, RLT  can be extracted through
the cross section asymmetry with regard to the ip of
beam helicity between 1:
d + d
LT  = (20)
d + d
c RLT  sin
= . (21)
RT + L RL + c+ RLT cos + RT T cos 2
Fig. 10. Result for the S0+ strength from 0 electroproduc-
tion at low Q2 (black cross). The open sympols represent older
measurements from DESY [28] and NINA [29].
Fig. 11. Schematics of the Mller polarimeter of the MAMI-
A1 collaboration.
The measurement requires longitudinally polarised elec-
tron beam and, since RLT  enters the cross section with
sin (cf. eq. (3)), out-of-plane detection of the recoil pro-
tons or the scattered electrons of the p(e, e p) 0 reaction.
Intense and highly polarised electron beams are rou-
tinely available at MAMI [30]. The degree of longitu-
dinal polarisation at the beam axis in the spectrome-
ter hall is accurately measured by a Mller polarimeter,
where the beam electrons are scattered o polarised elec-
trons. Those are provided in an iron foil magnetised to
saturation within the 4 Tesla eld of a superconducting
solenoid. Symmetric scattering kinematics is selected by
momentum-specic detection of the outgoing electron pair
behind a magnetic dipole eld. The setup is schematically
shown in g. 11. A cm angular acceptance of /  20 %
around cm = 90 makes the inuence of the Fermi mo-
tion of the bound electrons, the so-called Levchuk ef-
fect [31], negligible [32].
Out-of-plane detection capability is provided by spec-
trometer B, which can be tilted up to 10 [33] as visible in
g. 8. Using this setup the beam-helicity asymmetry was
determined at the energy of the (1232) resonance and
Q2 = 0.2 (GeV/c)2 [34]. The result is shown in g. 12.
The calculations within the dynamical models of Sato
and Lee [22] as well as of DMT2001 [25], and the
MAID2000 parametrisation [23] all were found to disagree
 H. Schmieden: Photo- and electro-excitation of the -resonance at MAMI 97

Fig. 13. Recoil polarisation components in parallel kinemat-

ics of the reaction p(e, e p) 0 . For visual clarity only, the 0
is drawn slightly sideways. The x, y, and z components are
dened relative to the electron scattering plane.

tion [37]:
Fig. 12. Beam-helicity asymmetry LT  measured in 0 pro-
duction with longitudinally polarised electron beam. The dot- 
0 Px = Pe 2L (1 )RLT t
, (22)
ted curve represents the original MAID2003 calculation [23]. 
The dashed-dotted and dashed curves are the results of the 0 Py = 2L (1 + )RLT ,
n
(23)
dynamical models of Sato-Lee [22] and Kamalov-Yang [25], re- 
spectively. The full curve is the MAID re-t of ref. [21]. Errors 0 Pz = Pe 1 2 RTl T  . (24)
are purely statistical.
According to g. 13 the axes are dened relative to the
electron scattering plane and the virtual photon direction.
This is the natural choice, since in parallel kinematics the
with this measurement. However, rescaled by a factor 0.75 recoil polarisation is completely determined by the angular
MAID described the asymmetry very well. The later re- momentum transfer from the photon eld. Decomposition
t of real and imaginary parts of the S1+ and S0+ 0 - up to S and P partial waves,
amplitudes in MAID2003 on basis of the MAMI LT and
LT  data provided a very satisfactory description of the 0 Px = Pe c e{(4S1+ + S1 S0+ )
polarisation data, cf. the full curve in g. 12, and of the (M1+ M1 E0+ + 3E1+ )}, (25)
unpolarised measurement discussed in the preceding sec-
0 Py = c+ m{(4S1+ + S1 S0+ )
tion as well [21]. This underlines the absolute need to un-
derstand the physical background amplitudes before the (M1+ M1 E0+ + 3E1+ )}, (26)

small quadrupole contributions can be reliably extracted.
Ideally, a complete experiment with respect to a multipole
decompostion is required.
3.2.3 Double Polarisation Experiments: Recoil proton
polarimetry
A complete experiment in the above sense would require
to measure more than 11 independent observables over the
energy range of the (1232) and the full angular range.
Even with application of Watsons theorem [35,36], which
below 2 0 threshold relates real and imaginary parts of
the amplitudes, this presently is out of reach in pion elec-
troproduction. In order to get as close as possible and thus
minimise the model dependence in the extraction of the
quadrupole amplitudes, it is mandatory to measure double
polarisation observables.
In parallel kinematics of the p(e, e p) 0 reaction (cf.
g. 13) each of the three cartesian components of the
proton polarisation, in addition to the unpolarised cross
section 0 , only depends on one specic structure func-
0 Pz = Pe c0 |M1+ |2 + |M1 |2 + 9|E1+ |2 + |E0+ |2
+ e{6E1+
(M1+ M1 ) 2M1+
M1
2E0+ (M1+ M1 + 3E1+ )} , (27)
reveals the high sensitivity to the quadrupole amplitudes.
In particular, since 0 is dominated by |M1+ |2 , Px pro-
0
videsa rather direct measure of the CMR . The factor
c0 = 1  entering Pz is related to the circular polari-
2
sation of the photon eld.
Except the small contribution of RL to 0 , in parallel
kinematics Pz is entirely determined by kinematic con-
stants, i.e. the transfer of circular photon polarisation to
the protons, since RTl T  = RT . The transverse components
Px and Py measure real and imaginary parts of the same
interferences, similar to LT and LT  but in dierent com-
binations and with dierent weights of amplitudes. Thus,
0
the model dependence in the extraction of the CMR is
dierent from unpolarised measurements. Morever, also
the composition of systematic errors is completely dier-
ent due to the orthogonal experimental technique.
The rst N double-polarisation experiment
ever has been performed at the 3 Spectrometer setup of
MAMI [38]. Central to the experiment is the measurement
of proton polarisation behind the focal plane of one of the
spectrometers [39,40]. This is based on inclusive proton
98 The European Physical Journal A

carbon scattering. Due to the strong spin-orbit coupling,

the transverse polarisation components, Pn and Pt , gen-
erate an azimuthal modulation of the unpolarized cross
section, 0C :

C = 0C (s , T ) [1 + AC (s , T ) (Pt sin Pn cos )] .

(28)
The relative strength is determined by the known [41,42]
analyzing power, AC .
The two polarisation components accessible at the fo-
cal plane must be traced back to the target through the
spectrometer magnets. This complication on the one hand
provides, on the other hand, the opportunity to determine
the otherwise inaccessible longitudinal polarisation com- 0

ponent. A separation of Px , Py and Pz is achieved since Fig. 14. CMR from recoil polarisation (open diamond) [38]
x and z components are odd under beam-helicity rever- in comparison to unpolarised measurements [18, 19, 26, 28,
29, 44, 45] at low Q2 . The curves show model calcula-
sal, while Py is even (cf. eqs. (22) (24)). At the energy
tions MAID2003 [23] (solid), DMT2001 [25] (dashed) and
of the (1232) resonance and a momentum transfer of
Sato/Lee [22] (dashed dotted).
Q2 = 0.121 (GeV/c)2 the three polarisations were simul-
taneously measured to
1
momentum conservation [46]:
Px /Pe = (11.4 1.3stat 1.4syst ) %, (29) 1
2x + 2y = z (1 z ). (33)
Py = (43.1 1.3stat 2.2syst ) %, (30) L
Pz /Pe = (56.2 1.5stat 2.6syst ) %. (31) The reduced polarisations are dened by
t
1 RLT 
x = Px = , (34)
0
Pe c R T + L R L
Based on the Px result, the CMR n
1 RLT
y = Py = , (35)
c+ R T + L R L

{S1+ M1+ } 1 RTl T 
= (6.4 0.7stat 0.8syst ) % (32) z = Pz = . (36)
|M1+ |2 Pe c0 R T + L R L
The consistency relation seems hardly fullled by the
was extracted using the MAID parametrisation [38]. measured recoil polarisations. The experimental result for
Within the errors this agrees with the alternative method eq. (33) is
of extraction using Px /Pz , which experimentally provides
3.9 0.4stat 0.4syst = 7.9 0.7stat 1.2syst . (37)
the advantage that the magnitudes of both analysing
power and beam polarisation drop out. Contrary, the ra- Despite the non-linear error propagation on the r.h.s. of
tio is aected by a larger uncertainty in spin precession of eq. (33), the probability for such a nding is only a few
the z-component compared to Px . To rst order the latter percent. While MAID2000 fullls the consistency rela-
would not precess at all in a homogeneous vertical-bend tion, there is a discrepancy with the Py measurement.
dipole eld. The MAID-ret mentioned above slightly improves this
0 situation as shown in g. 15, where the MAMI recoil po-
The result for the CMR is depicted in g. 14 along
larisation data [38] are compared to the MAID versions
with unpolarised measurements. Keeping in mind that
0 2000, 2003 and the re-t. Nevertheless, the role of non-
the large negative CMR of ref. [26] (full triangle tip resonant background, which particularly shows up in Py ,
down) is practically excluded by the S0+ results discussed still remains unresolved. Another measurement of Py [48]
in sect. 3.2.1, very convincing agreement is observed be- is in agreement with the MAMI data but has a large er-
tween the various measurements. A recent very high statis- ror. Unfortunately, ref. [43] gives no explicit values of the
tics measurement of the angular distributions of recoil measured polarisations.
polarisation [43] found, just outside the range of g. 14,
0 1
CMR = (6.610.18) % at Q2 = 1 GeV/c2 . This agree- Despite the occurence of L in eq. (33), the polarisation
ment extends also to the positive S0+ /M1+ ratio deter- relation is nevertheless frame independent. The longitudinal
mined from LT . polarisation parameter is also contained in the denition of
the factors c of the reduced polarisations and thus could be
The three polarisation components measured in paral- eliminated. A similar relation holds in elastic electron nucleon
lel kinematics are model independently related by angular scattering, where y vanishes [47].
 H. Schmieden: Photo- and electro-excitation of the -resonance at MAMI
5
0 -25
(%)
(%)
-5
Pxsp / Pe
Pysp
-10
-15
0 0.1 0.2 0.3 0.4
Q2 (GeV2/c2)
Fig. 15. Components of recoil polarisation measured at the energy of the (1232)-resonance in the reaction p(e, e p) 0 [38].
Curves show the MAID versions 2000 (full), 2003 (dashed), and the re-t (dashed-dotted) on basis of LT and LT  [21] discussed
in the text. Note the suppressed zero middle and right.
In principle, from the reduced polarisations it is possi-
ble to determine the ratio of longitudinal to transverse re-
sponse, RL /RT , without the need of a classical Rosenbluth
separation. According to ref. [46] this can be achieved in
three dierent ways, using either
only the longitudinal component, z (cf. also [49]),
the quadratic sum of the transverse components,
2x + 2y , or
all three reduced polarisations.
However, as a re ection of the almost violated consistency
relation the results obtained from the MAMI measure-
ments vary signicantly. The smallest value, RL /RT =
(4.7 0.4stat 0.6syst ) %, is extracted from the quadratic
sum 2x + 2y of the transverse reduced polarisations and
the largest one, RL /RT = (12.2 +1.7 +2.9
1.6stat 2.7syst ) %, from
z alone. This presently prohibits a reliable extraction
of RL /RT but stresses the importance of a simultaneous
measurement of all polarisation components with further
improved accuracy.
4 Interpretation
The MAMI experiments towards the quadrupole strength
in the N (1232) transition yield very consistent re-
sults. The rst order physical background contributions
like mS0+ are now much better under control. Remain-
ing uncertainties and inconsistencies seem related to the
contribution of higher partial waves especially in the imag-
inary parts of interferences as in LT  and Py . Though
important for our understanding of the non-resonant pro-
cesses, they are very much suppressed in the discussed ob-
servables with high sensitivity to EMR and CMR. At least
in the vicinity of the photon point, a reliable extraction of
the EMR and CMR is possible with a remaining relative
model uncertainty of the order 10 %. Thus it is evident
that the experimental results are an order of magnitude
larger than expected from the quark model calculations.
99
-20 70
-30 65
(%)
-35
Pzsp / Pe
60
-40
-45
55
-50
-55 50
0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
Q2 (GeV2/c2) Q2 (GeV2/c2)
Obviously, pionic degrees of freedom which in a mi-
croscopic picture would be responsible for e.g. the mS0+
contribution (g. 10) play an important role. This is
made particularly transparent within the dynamical mod-
els [24,22]. Such calculations can be split into the so-called
bare and dressed parts, which represent the nucleon/delta
core, and the pion cloud, respectively. This is also at-
tempted within dispersion relation approaches [50]. While,
in general, such a separation suers from an unitary am-
biguity [51], the models yield consistent results. E.g., the
full model of ref. [24] describes the data while the bare cal-
culation yields a very small and positive CMR. Moreover,
3/2
the full calculation yields an M1+ amplitude in agreement
with the experimental data. In contrast, the bare calcu-
3/2
lation gives only about two thirds of M1+ and a CMR
which is an order of magnitude too small. This is simi-
lar to quark model calculations without pion degrees of
freedom [52,53], which also underestimate the quadrupole
strength by a factor of ten and get only about 60 % of the
experimental M1+ .
Despite the yet unsatisfactory statistics of dynamic lat-
tice calculations [54], those are in qualitative agreement
with the experimental results and hence further support
the important role of the pion cloud.
As outlined in the introduction the quadrupole transi-
tion strength can be interpreted in terms of a deformation
of the baryons involved. Buchmann and Henley nd oppo-
site deformations of equal strength for nucleon and [55].
Also the quadrupole moments for the nucleon core and
the pion cloud are of opposite sign. However, the core ap-
pears almost spherical and the deformation due to the pion
cloud is an order of magnitude stronger, very similar to
the observation within the dynamical model. Buchmann
and Henley quote total quadrupole moments in the range
of Qnucl = (0.113 . . . 0.5) fm2 for nucleon (upper sign)
and (lower sign). It is interesting to note that, relative
to the size of the objects, the corresponding deformation
Qnucl
2 > 0.043
rN
of the nucleon
100 The European Physical Journal A
compared to
Qdeut
= 0.07 of the deuteron
rd2
would be of similar magnitude.
5 Summary and outlook
The MAMI experiments on pion photo- and electropro-
duction provided cornerstones for our understanding of
the N (1232) transition. At the photon point and
at low Q2 they unambiguously demonstrated that the
quadrupole strength is an order of magnitude larger than
expected in QCD-motivated constituent quark models.
Often a precursor for similar experiments at other lab-
oratories, polarisation techniques played a key role. They
have already been extended to polarised target measure-
ments and to energies slightly above the (1232) [56].
In view of the higher resonances which become accessi-
ble with the increased energy of MAMI C, their potential
seems not at all exhausted yet. Of particular interest will
be the determination of Q2 -slopes of transition amplitudes
to Roper, N (1440)P11 , and N (1535)S11 resonances. Since
related to the spatial extention of the systems, new in-
sights into the debated hybrid or molecular structure of
such states can be expected. Extension of the experimen-
tal techniques to associated strangeness production seems
relatively straightforward, provided that Kaons can be un-
ambiguously identied and su ciently high beam energy
is available.
In particular through polarisation data of unprece-
dented accuracy MAMI has the potential to continuously
challenge our understanding of the many body structure
of the nucleon. With all the virtues of the electromagnetic
probe it can provide experimental key observables also for
the comparison to the progressivly improving Lattice cal-
culations.
It is a special pleasure for me to thank the retirees Hartmuth
Arenhovel, Hartmut Backe, Dieter Drechsel, Jorg Friedrich,
Karl-Heinz Kaiser and Thomas Walcher who, through their
invaluable individual contributions over the last two decades,
made it possible to turn MAMI into the big success it has
become. I also greatfully acknowledge the support of D. Elsner
in preparing talk and written manuscript of this overview.
References
1. F. Wilczek, hep-ph/0201222, MIT CTP 3236.
2. A. Thomas, W. Weise, The Structure of the Nucleon
(Wiley VCH 2001).
3. S. Aoki et al., Phys. Rev. Lett. 84, 238 (2000).
4. Q. Mason et al., Phys. Rev. Lett. 95, 052002 (2005).
5. J.W. Negele, hep-lat/0509101 and references therein.
6. A. de Rujula, H. Georgi, S.L. Glashow, Phys. Rev. D 12,
147 (1975).
7. S.L. Glashow, Physica A 96, 27 (1979).
8. Compilation by J. Ahrens, Mainz (2000).
9. M. MacCormick et al., Phys. Rev. C 53, 41 (1996).
10. T.A. Armstrong et al., Phys. Rev. D 5, 1640 (1972).
11. D. Babusci et al., Phys. Rev. C 57, 291 (1998).
12. D. Drechsel, L. Tiator, J. Phys. G: Nucl. Part. Phys. 18,
449 (1992).
13. G. Knochlein, D. Drechsel, L. Tiator, Z. Phys. A 352, 327
(1995).
14. R. Beck et al., Phys. Rev. Lett. 78, 606 (1997); 79, 4515
(1997) (E).
15. R. Beck et al., Phys. Rev. C 61, 035204 (2000).
16. O. Hanstein, D. Drechsel, L. Tiator, Nucl. Phys. A 632,
561 (1998).
17. R. Leukel, doctoral thesis, Mainz (2001).
18. K. Joo et al., Phys. Rev. Lett. 88, 122001 (2002).
19. R.W. Gothe, Prog. Part. Nucl. Phys. 44, 185 (2000).
20. V.V. Frolov et al., Phys. Rev. Lett. 82, 45 (1999).
21. D. Elsner et al., Eur. Phys. J. A 27, 91 (2006).
22. T. Sato, T.-S.H. Lee, Phys. Rev. C 63, 055201 (2001).
23. D. Drechsel, O. Hanstein, S.S. Kamalov, L. Tiator, Nucl.
Phys. A 645, 145 (1999); http://www.kph.uni-mainz.de/
MAID/.
24. S.S. Kamalov, S.N. Yang, Phys. Rev. Lett. 83, 4494 (1999).
25. S.S. Kamalov, S.N. Yang, D. Drechsel, L. Tiator, Phys.
Rev. C 64, 032201 (2001).
26. F. Kalleicher et al., Z. Phys. A 359, 201 (1997).
27. H. Schmieden, Proceedings of NSTAR2001, edited by D.
Drechsel, L. Tiator (World Scientic, 2001) p. 27.
28. J.C. Alder et al., Nucl. Phys. B 46, 573 (1972).
29. R. Siddle et al., Nucl. Phys. B 35, 93 (1971).
30. K. Aulenbacher, these proceedings.
31. L.G. Levchuk, Nucl. Instrum. Methods A 345, 496 (1994).
32. P. Bartsch, diploma thesis KPH 11/96, Mainz (1996).
33. K.I. Blomqvist et al., Nucl. Instrum. Methods A 403, 263
(1998).
34. P. Bartsch et al., Phys. Rev. Lett. 88, 142001 (2002).
35. K.M. Watson, Phys. Rev. 95, 228 (1954).
36. E. Fermi, Suppl. Nuovo Cimento 2, 17 (1955).
37. H. Schmieden, Eur. Phys. J. A 1, 427 (1998).
38. Th. Pospischil et al., Phys. Rev. Lett. 86, 2959 (2001).
39. Th. Pospischil et al., Nucl. Instrum. Methods A 483, 713
(2002).
40. Th. Pospischil et al., Nucl. Instrum. Methods A 483, 726
(2002).
41. E. Aprile-Giboni et al., Nucl. Instrum. Methods 215, 147
(1983).
42. M.W. McNaughton et al., Nucl. Instrum. Methods 241,
435 (1985).
43. J.J. Kelly et al., Phys. Rev. Lett. 95, 102001 (2005); nucl-
ex/0509004.
44. K. Batzner et al., Nucl. Phys. B 76, 1 (1974).
45. N.F. Sparveris et al., Phys. Rev. Lett. 94, 022003 (2005).
46. H. Schmieden, L. Tiator, Eur. Phys. J. A 8, 15 (2000).
47. Th. Pospischil et al., Eur. Phys. J. A 12, 125 (2001).
48. G.A. Warren et al., Phys. Rev. C 58, 3722 (1998).
49. J.J. Kelly, Phys. Rev. C 60, 054611 (1999).
50. I.G. Aznauryan, S.G. Stepanyan, Phys. Rev. D 59, 054009
(1999).
51. P. Wilhelm et al., Phys. Rev. C 54, 1423 (1996).
52. N. Isgur, G. Karl, R. Koniuk, Phys. Rev. D 25, 2394
(1982).
53. S.S. Gershtein, G.V. Dzhikiya, Sov. J. Nucl. Phys. 34, 870
(1981).
54. C. Alexandrou et al., Phys. Rev. Lett. 94, 021601 (2005);
hep-lat/0509140.
55. A.J. Buchmann, E.M. Henley, Phys. Rev. C 63, 015202
(2000).
56. A. Thomas, these proceedings.
Eur. Phys. J. A 28, s01, 101 106 (2006)
DOI: 10.1140/epja/i2006-09-011-8 EPJ A direct
electronic only

Parity violation in electron scattering

S. Kowalskia
Laboratory for Nuclear Science and Department of Physics, Massachusetts Institute of Technology, Cambridge, MA, 02139, USA

/
Published online: 26 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. Parity-violating electron scattering has been a very useful tool for probing the structure of
neutral currents and providing detailed information on electroweak form factors. A pioneering SLAC mea-
surement in the mid-70s provided an important early test of the Standard Model. Modern electron acceler-
ators provide high-intensity (> 100 A), CW beams with polarizations as high as 85%. Experiments such
as SAMPLE, A4, HAPPEX and G0 have exploited these capabilities and obtained new information on
electroweak strange form factors in the Q2 range of 0.1 1.0 (GeV/c)2 . That activity continues. Other ex-
periments are designed to provide stringent tests of the Standard Model. E-158 at SLAC recently measured
the weak charge of the electron. Qweak is a challenging new experiment at JLAB which is designed to mea-
sure the weak charge of the proton. This will probe for physics beyond the Standard Model corresponding
to energy scales of more than 5 TeV.

PACS. 12.15.-y Electroweak interactions 14.20.Dh Protons and neutrons 21.10.Gv Mass and neutron
distributions

1 Introduction These were designed as tests of the Standard Model. Dur-

The electromagnetic probe is one of our most important
tools for probing nucleon and nuclear structure and dy-
namics. Modern accelerators use electron scattering to
probe distances of less than 1 fm where signatures of quark
degrees of freedom are expected to be observable. A new
generation of electron accelerators, 0.5 6 GeV, with CW
capability and intense polarized beams have become op-
erational over the past decade. This has enhanced our ca-
pability for doing coincidence experiments and other mea-
surements by more than two orders of magnitude. During
this time, longitudinally polarized electrons have emerged
as a very important new tool to study nucleon and nu-
clear structure. They have provided precise new informa-
tion on nuclear electromagnetic form factors and other
nuclear structure functions.
An important class of experiments using polarized
electrons has been the study of parity violation. Parity-
violating electron scattering involves the scattering of lon-
gitudinally polarized electrons from an unpolarized target.
A change in counting rate resulting from a reversal of the
beam helicity is a signal of a parity non-conserving ef-
fect. At high energies the SLAC parity violation exper-
iment provided a crucial understanding of electro-weak
processes and a precise measurement of the weak mixing
angle, sin2 W . At low energies the rst parity violation
experiments were carried out at MIT-Bates and Mainz.
a
e-mail: sbk@mit.edu
ing the past decade there has been an extensive program
of such parity violating experiments at MIT-Bates, JLAB
and Mainz. The goal of these parity-violating experiments
currently is three-fold:
study the strangeness content of the nucleon,
measure the neutron density distribution in a heavy
nucleus
and
sensitive tests of the Standard Model.
In this paper, we will discuss and report on the re-
cent progress of parity violation experiments in these three
broad physics areas. Results and future prospects of the
dierent experiments will be presented. The Mainz A4
experiment will not be discussed since it is the subject of
another contribution to these proceedings.
2 Physics program
2.1 Nuclear structure: strange quarks
The proton is made up of three valence quarks, uud, and
a sea of gluons and qq pairs all of which contribute to
its electromagnetic properties at short distance scales. It
was realized [1] that a measurement of the parity-violating
asymmetry arising from the interference between the elec-
tromagnetic and neutral current amplitudes would allow
us to extract the contributions of strange quarks to the
102 The European Physical Journal A

ground state charge and magnetization distributions (e.g.,

magnetic moment) of the nucleon.
There have been many theoretical estimates of strange
quark contributions to nucleon properties. These include
both phenomenological models and lattice-gauge calcula-
tions. Separation of strange quark contributions to nu-
cleon currents was developed by Kaplan and Manohar [2].
The parity-violating asymmetry [3] for scattering longitu-
dinally polarized electrons from a proton can be written as

GF
AP V =
4 2

p Z 

GE GE + GpM GZ p
M (1 4sin W )GM GA
2 e
. (1)
(GpE )2 + (GpM )2

The three terms in AP V arise as a result of the in-

terference between the electromagnetic and weak interac-
tions. The terms contain bilinear products of electromag- Fig. 1. A schematic view of one module of the SAMPLE ex-
netic and weak form factors. GpE,M are the form factors perimental apparatus. Ten mirror-phototube pairs are placed
associated with the distribution of the protons charge and asymmetrically about the beam axis.
magnetism. The weak form factors GZ e
E,M and GA contain
s
contributions from strange quarks, GE,M .
The kinematic factors depend upon specics of an ex-
periment. They are chosen to enhance the relative sensi-
tivity of individual terms to AP V . The rst two terms are
important at forward angles and the last two at backward
angles:

= Q2 /4Mp2 , (2)
 1
= 1 + 2(1 + )tan2 (e /2) (3)

and 
 = (1 2 ) (1 + ) . (4)
1. SAMPLE
SAMPLE [4] at MIT-Bates was the rst experiment to use
parity violation as a probe for strange quarks in the pro-
ton. A longitudinally polarized electron beam of 200 MeV
was incident on a liquid hydrogen or deuterium target.
The polarized electrons were produced using a bulk GaAs
crystal resulting in an average polarization of 36%. The
linac produces a pulsed beam of 25 s duration at a rep-
etition rate of 600 Hz. The beam current was 40 A. The
helicity of the beam is changed randomly pulse-by-pulse
and in addition a half-wave plate can be inserted to change Fig. 2. Uncertainty bands of GsM vs. GeA at Q2 = 0.1 (GeV/c)2
the overall sign of the helicity. for the SAMPLE experiment in both hydrogen and deuterium.
The scattered electrons were detected in a large solid Also shown is the uncertainty band of the theoretical expecta-
angle (1.5 sr) air Cerenkov detector spanning angles be- tion for GeA .
tween 130 and 170 . At backward angles SAMPLE is
mostly sensitive to GsM and GeA . Figure 1 shows one of
the 10 detector modules which are placed A symmetri- was critical to the success of the experiment. Several feed-
cally about the beam axis. The Cherenkov light is focused back systems were used to minimize such eects. These
by an ellipsoidal mirror unto a phototube. The integrated included energy, beam position, angle, and intensity. All
light is proportional to the scattered electron rate of about parity experiments implement similar feedback controls.
108 s1 in a beam pulse. Experiments were carried out at 200 MeV on both hy-
A shutter located in front of the phototube could be drogen and deuterium targets. In addition a measurement
closed providing a measurement of the background origi- at 125 MeV was also made on a deuterium target. The two
nating from neutrons and charged particles. Tight control targets in principle allow a separation of GsM and GeA . The
of helicity-correlated eects on the properties of the beam results for both hydrogen and deuterium are summarized
 S. Kowalski: Parity violation in electron scattering 103

in g. 2. Using a theoretical prediction for GeA (T = 1)

one can extract a value for GsM :

GsM (Q2 = 0.1 (GeV/c)2 ) = 0.37 20 0.26 0.07. (5)

The experiment favors a small positive value for the

magnetic moment contribution, s , which is at variance
with most theoretical predictions.
2. HAPPEX
HAPPEX at JLAB was the second experiment designed
to look for strange quarks using parity violating elec-
tron scattering. The rst measurement was carried out at
Q2 = 0.48 (GeV/c)2 . A 3.2 GeV beam of polarized elec-
trons was incident on a liquid hydrogen target. The scat-
tered electrons were detected in the pair of high resolution
spectrometers in Hall-A at a scattering angle of 12.5 .
Beam currents up to 100 A were used. The Cherenkov
light from a lead-lucite sandwich was integrated over the
duration of the helicity window. Beam helicity was re- Fig. 3. The four AP V measurements at Q2 = 0.1 (GeV/c)2
versed at 30 Hz. Beam polarization was 38% in the rst are shown, with shaded bands representing 1-sigma combined
run and improved to 70% in the second run. It was mea- statistical and systematic uncertainty. Also shown is the com-
sured with both Moeller and Compton polarimeters. Tight bined 95% C.L. ellipse from all four measurements.
control of helicity correlated systematics was similar to
those used in the SAMPLE experiment.
The experiment is sensitive to both GsE and GsM . They
extracted [5] a linear combination of strange form factors,

GsE + 0.392GsM = 0.014 0.020 0.010. (6)

In 2004, a second HAPPEX measurement was carried
out at Q2 = 0.10 (GeV/c)2 . Measurements were made
on both 1 H and 4 He targets. Helium is a special target
since the nuclear spin I = 0. It has sensitivity only to GsE .
Combining both the hydrogen and helium results allows a
direct extraction of GsE and GsM .
A 3.0 GeV polarized electron beam was incident on the
target. The scattered electrons were detected in the pair
of high resolution spectrometers in Hall-A at a scattering
angle of 6 . A pair of septum magnets in front of the Fig. 4. The combination GsE + GsM for the HAPPEX, A4
spectrometers made this feasible. Beam currents of up to and G0 experiments. Also shown at Q2 = 0.6 (GeV/c)2 the
35 A were used. Total absorption detectors were used to projected uncertainty for a future HAPPEX-III measurement.
detect the scattered electrons.
The experiment on hydrogen yielded [6] the strange
form factor combination, In a future experiment HAPPEX is planning a mea-
surement at Q2 = 0.6 (GeV/c)2 . Figure 4 shows the ex-
GsE + 0.08GsM = 0.030 0.025 0.006. (7) pected uncertainty for this measurement. We expect to
run this experiment in 2008-09.
The measurement on 4 He yielded [7],
3. G0
GsE = 0.038 0.042 0.010. (8)
G0 is another ambitious JLAB parity violation experi-
Both measurements are consistent with zero. The com- ment. Its goal is to search for evidence of strange quarks
bined hydrogen and helium HAPPEX results were con- in the proton. It is designed to make both forward and
sistent with no evidence for strange quarks at Q2 = backward angle measurements in the Q2 range, 0.2
0.1 (GeV/c)2 . The results are summarized in g. 3. 1.0 (GeV/c)2 .
Following the above Phase-I measurements, a second G0 is an 8 sector superconducting toroidal spectrom-
experiment on both hydrogen and helium was completed eter. A 3.0 GeV electron beam was incident on a liquid
in 2005. These results, not yet available, when analyzed hydrogen target. The beam current was 40 A. The de-
are projected to have error bars which are 1/3 those ob- tector is designed to detect the recoil protons (g. 5). It is
tained so far. This will be a very signicant improvement. segmented along the focal plane (16) so that each segment
104 The European Physical Journal A
Fig. 5. Layout of the G0 spectrometer system. Shown are the
superconducting toroidal magnet and the segmented scintilla-
tor detector array.
measures a dierent Q2 range. The beam duty factor was
reduced to 6% allowing timing measurements to be used to
separate background from the recoiling protons. The de-
tector operates in counting mode. The A4 experiment at
Mainz also operates in counting mode. All other electron
parity violation experiments operate in integrating mode.
The results [8] are shown graphically in g. 4 together
with the results from other experiments. The results in-
dicate a non-zero, Q2 -dependent, strange quark contribu-
tion. They cover a much broader Q2 range than the other
experiments. The uncertainties for the higher Q2 points
are quite large as a result of signicant unanticipated back-
ground contributions.
G0 is planning to make backward angle measurements
in 2006. At backward angles, scattered electrons are de-
tected instead. Each measurement will be at a single Q2 .
They expect to make the rst measurements at approxi-
mately 0.2 and 0.6 (GeV/c)2 .
2.2 Neutron densities
Charge radii of nuclei are well known. For example, the
proton radius of the lead nucleus, Rp = 5.490 .002 fm.
This is very precise. A very interesting question is, What
is the neutron radius in Pb ? Experimentally, Rn is rather
poorly known. Perhaps it is known to 5% at best, where
the best estimates come from theory.
Models of a neutron star posit a solid crust over a liq-
uid core. The lead nucleus is believed by some to have a
neutron skin. Both a possible neutron skin and a neutron
star crust would be made of neutron rich matter at simi-
lar densities. A measurement of Rn to 1% accuracy would
have a major impact on nuclear theory, neutron star struc-
ture and atomic parity violation. Currently atomic par-
ity violation experiments are limited in accuracy by our
knowledge of the neutron radius in a heavy nucleus.
Donnelly, Dubach and Sick [9] suggested the possibil-
ity of using parity violating electron scattering to measure
Rn . The electromagnetic coupling to the protons is given
by QpEM = 1 while the neutrons couple by QnEM = 0.
In contrast the weak coupling of the protons is given by
QpW 1 4 sin2 W and the neutrons by QnW 1. The
neutrons have a very strong weak coupling to the electro-
magnetic probe. The resulting parity-violating asymmetry
is given by 2
GF Q2 FW (Q )
AP V = 42 F (Q2 ) , (9)
where
FW (Q2 ) = (1 4 sin2 W )Fp (Q2 ) N Fn (Q2 ), (10)
and
F (Q2 ) = ZFp (Q2 ) (11)
A measurement of AP V to 3% would provide a mea-
surement of Rn to 1%. An experiment has been approved
at JLAB to carry out such a measurement. An 850 MeV,
50 A polarized beam would be incident on a lead target
sandwiched between diamond sheets for cooling. Electrons
scattered through 60 would be detected in the pair of Hall-
A high-resolution spectrometers. At Q2 = 0.01 (GeV/c)2
the parity-violating asymmetry, AP V = 0.5 ppm.
This is a very challenging experiment. It will require
control of helicity correlated systematics to much better
than 15 ppb. It appears to be feasible.
2.3 Standard Model tests
Parity violating electron scattering has been used to test
the Standard Model (SM) since the pioneering SLAC ex-
periment in the mid-70s. Collider experiments at the Z-
pole provide our best measure of the weak mixing angle,
sin2 W . There remains much interest in exploring the run-
ning of sin2 W from the Z-pole to Q2 = 0.
At Q2 = 0, atomic parity experiments give results
consistent with the running of sin2 W . Other experi-
ments testing the SM include 12 C(e, e), 9 Be(e, e), NuTeV,
Moeller scattering and Qweak . All parity violating experi-
ments can be described in terms four fundamental quark
coupling constants,
GF  
LP V = e 5 e C1u u u + C1d d d +
2

 5 5

+ e e C2u u u + C2d d d . (12)
The SM makes predictions for the constants. These con-
stants in turn can be written in terms of isoscalar and
isovector combinations at the hadronic vertex:
= C1u + C1d = (1 2 sin2 W ), (13)
= C2u + C2d = (1 4 sin W ),
2
(14)
= C1u C1d = 2/3 sin W ,
2
(15)
= C2u C2d = 0, (16)
sin2 W = 0.23120 0.00015. (17)
 S. Kowalski: Parity violation in electron scattering 105

Fig. 7. Schematic layout of the Qweak experiment showing

the target, collimators, shielding, toroidal magnet and detector
system.
Fig. 6. Results for the SLAC and Bates parity violation ex-
periments in the space of the coupling constants and .

12
1. C / 9 Be
12
C is a spinless and isoscalar nucleus and elastic electron
scattering is described by a single form factor. The parity-
violating asymmetry may be written at the tree level [10,
11] as 1
3 2
AP V = 2 GF Q ( 2 ) . (18)
A parity violation experiment was carried out at MIT-
Bates [12] on 12 C. The results are shown in g. 6 to-
gether with the results of the SLAC experiment [13] on
deuterium. The results of both experiments are consistent
with the predictions of the SM.
2. E-158
A recent experiment testing the SM was E-158 at SLAC.
It is a purely leptonic process involving Moeller scatter- Fig. 8. Calculated running of the weak mixing angle in the
ing. The goal of the experiment was to measure the weak Standard Model. Data points are from the atomic parity vio-
charge of the electron, e = g sin W , using parity violation. lation experiment on Cs, the NuTeV experiment, the Moeller
In this experiment 50 GeV electrons were incident on experiment (E-158) at SLAC and from experiments at the Z0 -
a liquid hydrogen target. At Q2 = 0.027 (GeV/c)2 the pole. Also shown are the anticipated error bars for Qweak .
parity violating asymmetry is expected to be 150 ppb. A
quadrupole spectrometer was used to focus the scattered
Moeller electrons while at the same time defocusing e-p tons weak charge, QpW = 1 4 sin2 W , to the highest
scattering events. The ux of scattered electrons was in- precision possible.
tegrated for each beam burst. The experiment yielded an The SM makes a rm prediction of QpW based on the
asymmetry, running of the weak mixing angle sin2 W from the Z0 -
pole to low energies, corresponding to a 10 eect in our
AP V = 175 30 20 ppb. (19)
experiment. Fig. 8 shows the SM prediction for sin2 W
The extracted weak mixing angle is totally consistent together with existing data and the expected precision for
with predictions. The result [14] is shown plotted in g. 8. this experiment. This parity violating experiment is in the
semi-leptonic sector. This is in contrast to E-158 which is
3. Qweak in the purely leptonic sector.
A major new initiative is under development at JLAB. The measurement will be carried out using a 1.2 GeV
The goal of the Qweak experiment is to measure the pro- electron beam at a scattering angle of 9 and a momentum
106 The European Physical Journal A
transfer Q2 = 0.03 (GeV/c)2 . The 180 A polarized beam
will be incident on a 35 cm liquid hydrogen target. An
eight sector toroidal spectrometer is being constructed for
this measurement. The scattered electrons will be detected
by quartz Cherenkov detectors operating in integrating
mode. A schematic layout of the spectrometer is shown in
g. 7.
The measurement will take 2200 hours and will deter-
mine the protons weak charge with 4% statistical accu-
racy. Figure 8 shows the projected quality of the Qweak
results in the context of other existing data.
3 Summary
Parity-violating high-energy electron scattering is an im-
portant probe of nucleon and nuclear structure. Several
physics areas are currently under investigation. There
have been many experiments investigating the importance
of strange quarks to the structure of the proton. Results
to date indicate that the contribution of strange quarks
to nucleon structure are relatively small or vanishing. A
proposed experiment on Pb would measure the neutron
radius to 1%. Qweak is a challenging Standard Model test
which probes 5 TeV energy scales.
Such demanding measurements have been made possi-
ble by important advances in accelerator technology. We
now have high intensity CW beams with beam polariza-
tion of 85%. Control of helicity correlated beam proper-
ties allows measurement of asymmetries to an accuracy
approaching 10s ppb.
References
1. R.D. McKeown, Phys. Lett. B 219, 140 (1989).
2. D. Keyslan, A. Manohar, Nucl. Phys. B 310, 527 (1988).
3. M.J. Musolf et al., Phys. Rep. 239, 1 (1994).
4. D.T. Spangle et al., Phys. Lett. B 583, 79 (2004).
5. K.A. Aniol et al., Phys. Rev. C 69, 065501 (2004).
6. The HAPPEX Collaboration (K.A. Aniol et al.), Phys.
Lett. B 635, 275 (2006), nucl-ex/0506011.
7. The HAPPEX Collaboration (K.A. Aniol et al.), Phys.
Rev. Lett. 96, 022003 (2006), nucl-ex/0506010.
8. D.S. Armstrong et al., Phys. Rev. Lett. 95, 092001 (2005).
9. T.W. Donnelly et al., Nucl. Phys. A 503, 589 (1989).
10. G. Feinberg, Phys. Rev. D 12, 3575 (1975).
11. J.D. Walker, Nucl. Phys. A 285, 345 (1977).
12. P.A. Souder et al., Phys. Rev. Lett. 65, 694 (1990).
13. C.Y. Prescott et al., Phys. Lett. B 84, 524 (1979).
14. P.L. Anthony et al., Phys. Rev. Lett. 95, 081601 (2005).
Eur. Phys. J. A 28, s01, 107115 (2006)
DOI: 10.1140/epja/i2006-09-012-7 EPJ A direct
electronic only

Parity-violating electron scattering at the MAMI facility in Mainz

The strangeness contribution to the form factors of the nucleon
F.E. Maasa
For the A4-Collaboration
Institut de Physique Nucleaire, CNRS/IN2P3, Universite de Paris Sud, 15, rue Georges Clemenceau, F-91406 Orsay CEDEX,
France

/
Published online: 29 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. A measurement of the weak form factor of the proton allows a separation of the strangeness
contribution to the electromagnetic form factors. The weak form factor is accessed experimentally by the
measurement of a parity violating (PV) asymmetry in the scattering of polarized electrons on unpolarized
protons. We performed such measurements with the setup of the A4-experiment at the MAMI accelerator
facility in Mainz. The role of strangeness in low energy nonperturbative QCD is discussed. The A4-
experiment is presented as well as the results on the strangeness form factors which have been measured
at two Q2 -values. The plans for backward angle measurements at the MAMI facility are presented.
PACS. 12.15.-y Electroweak interactions 11.30.Er Charge conjugation, parity, time reversal, and other
discrete symmetries 13.40.Gp Electromagnetic form factors 25.30.Bf Elastic electron scattering

1 Introduction tested to an extent that the weak neutral current of the

In 1965, the discovery of weak interaction violating the
symmetry transformation of parity P : x x [1,2] has
opened a new door in studying the weak force. In contrast
electromagnetic and strong interaction conserve parity
and therefore parity non-conserving observables are an im-
portant tool to uniquely identify weak interaction. Parity-
violating correlation observables like pseudo-scalars allow
one in electron scattering to access the weak neutral cur-
rent and separate it from the electromagnetic current and
the associated photon exchange. Such an observable is for
example the cross section asymmetry in scattering right-
handed versus left-handed electrons o an unpolarized
target. The rst pioneering parity-violation (PV) exper-
iment in electron scattering at high energy at SLAC [3],
and at low energies at the MIT Bates accelerator [4] and at
the Mainz linac [5] were intended to study the standard
model. These experiments were ground breaking in de-
veloping techniques for measuring the small (order 104 )
asymmetries with order 105 errors.
Today, parity-violating observables are studied in
many areas of physics, from atomic physics to nuclear
physics, from hadron physics to high-energy physics [6].
The availability of a polarized high-intensity continuous
electron beam at the Mainz Microtron MAMI was im-
portant to increase the sensitivity in measuring a parity-
violating asymmetry by two orders of magnitude to a level
of 0.1 ppm. The standard model and its parameters are
a
e-mail: maas@ipno.in2p3.fr
nucleon can be used to study nucleon structure. An exper-
imental program to measure the strangeness contribution
to the form factors at TJLab and at the MAMI facility
has emerged. In this contribution to the proceedings we
will focus on the experimental activity of the Mainz pro-
gram and the results achieved with the A4 experiment.
The parity program at MIT-Bates and at TJLab will be
reviewed in another contribution to these proceedings (see
the contribution of S. Kowalski).
2 Strangeness in the nucleon
Quantum chromodynamics (QCD) is the quantum eld
description of strong interaction. Quarks carrying color
charge interact via the emission of gluons. The gluon is
like the photon a spin-1 particle and it couples like the
photon to the vector current of quarks. Similar to Quan-
tum Electrodynamics (QED) an expansion of the interac-
tion in a power series of the coupling constant s = QCD
is possible. The renormalization of the Lagrangian leads in
the case of QED to a running coupling QED (Q2 ), which
rises slowly from the low-energy value of QED 1/137 to
the Z-mass where it reaches QED (Q2 m2Z ) 1/128 [7].
The gluons carry color charge and this coupling leads to
a very much dierent running of the coupling QCD (Q2 ),
which is depicted in g. 1. For scales Q2 > 1 (GeV/c)2 ,
QCD (Q2 ) becomes small and perturbative treatment of
QCD (perturbative QCD) is possible [8] with the well
108 The European Physical Journal A

a value of y = 0.46 corresponding to a contribution of the

strange scalar density to the nucleon mass of 220 MeV [11].
This contribution from scalar density to the nucleon mass
is most likely cancelled by other contributions like poten-
tial energy and kinetic energy to an extend that the net
strangeness contribution to the nucleon mass might be
small [12].
The strangeness contribution to unpolarized nucleon
structure functions has been determined in deep inelas-
tic neutrino scattering. One obtains the momentum frac-
tion of strange quarks as compared to u- and d-quarks:

Fig. 1. Running coupling in QED (QED ) and QCD (QCD )
up to one loop. World data suggest that QCD 250 MeV
analyzed in the M S-renormalization scheme using 5 quark a-
vors [7].
known asymptotic freedom at higher scales [9]. An as-
sociated energy scale appears at low energies, QCD , at
which QCD (Q2 ) diverges. Any perturbative description
breaks down, the theory is here not solvable, this is the
regime of nonperturbative QCD. Virtual excitations of
pairs of quarks and antiquarks play an important role in
the range of nonperturbative QCD. Understanding QCD
in this regime is closely linked to the structure of the mat-
ter surrounding us like proton and neutron. The successful
description of a wide variety of observables by the concept
of eective, heavy ( 350 MeV) constituent quarks, which
are not the current quarks of QCD, is still a puzzle. There
are other equivalent descriptions of hadronic matter at low
energy scales in terms of eective elds like chiral pertur-
bation theory (PT) or Skyrme-type soliton models. The
eective elds in these models arise dynamically from a
= x(s(x) + s(x)) /x(u(x) + d(x)) 0.5 [13,14]. This
corresponds to the fact, that at a scale of Q2 = 5 (GeV/c)2
the strange sea carries about 3% of the nucleons momen-
tum. In a dynamical QCD evolution model one starts
with some unpolarized quark structure function at a scale
of = 0.3 (GeV/c)2 and evolutes it to the scale of
Q2 = 5 (GeV/c)2 where nally the measured parton dis-
tributions have been reproduced [15]. In this approach one
nds, that the observed value of 0.5 is compatible with
a vanishing strange sea contribution at the low scale
 .
Information on the axial charge N |s 5 s|N and on
the strangeness contribution to the spin of the nucleon
comes from the interpretation of deep inelastic scattering
data and suggests a sizeable contribution of the strange
quarks of s(Q2 = 1 (GeV/c)2 ) = 4.5 0.7 to the nu-
cleon spin from a next-to-leading order perturbative QCD
analysis of the available world data set including higher
twist eects [16].
3 Strangeness contribution to the form
factors of the nucleon
Parity-violating (PV) electron scattering o nucleons pro-
vides experimental access to the strange quark vector cur-
sea of virtual gluons and quark-antiquark pairs. rent in the nucleon N |s s|N which is parameterized in
the electromagnetic form factors of proton and neutron,
In this context the contribution of strange quarks plays
GsE and GsM [17]. Recently the SAMPLE-, HAPPEX-,
a special role since the nucleon has no net strangeness,
A4- and G0-collaborations have published experimental
and any contribution of strange quarks to nucleon struc-
results. A direct separation of electric (GsE ) and magnetic
ture observables is a pure sea-quark eect and can not be
(GsM ) contribution at forward angle has been impossible
clouded by valence quark eects. Due to the heavier cur-
so far, since the measurements have been taken at dier-
rent quark mass of strangeness (ms ) as compared to up
ent Q2 -values. The experimental details of the SAMPLE,
(mu ) and down (md ) with ms 140 MeV  mu , md
HAPPEX and G0 collaboration are discussed in a dier-
510 MeV, one expects a suppression of strangeness ef-
ent contribution to these proceedings (see the contribution
fects in the creation of quark-antiquark pairs. On the other
of S. Kowalski).
hand, the strange quark mass is in the range of the mass
A determination of the weak vector form factors of
scale of QCD (ms QCD ) so that the dynamic creation
the proton (GpE and GpM ) is done by measuring the PV
of strange sea quark pairs could still be substantial. The
asymmetry in the scattering of longitudinally polarized
consequence of the presence of virtual strangeness quark
electron o unpolarized protons. It allows the determi-
pairs on dierent matrix elements has been studied:
nation of the strangeness contribution to the electromag-
For example, the strange density of the vacuum ss is netic form factors GsE and GsM . The weak vector form
around 20 % suppressed as compared to u- and d-quark
factors GpE,M of the proton can be expressed in terms
condensates: ss /qq = 0.80.3 [10], which supports the
mechanism indicated above. The strange scalar density in of the known electromagnetic nucleon form factors Gp,n E,M
the nucleon N |ss|N is studied in connection with the and the unknown strange form factors GsE,M using isospin
pion nucleon sigma term. The scalar strangeness content symmetry and the universality of the quarks in weak and

of the nucleon is dened as y = 2p|ss|p /p|uu + dd|p , electromagnetic interaction GpE,M = (GpE,M GnE,M )
and a recent evaluation of N scattering length data yields 4 sin2 W GpE,M GsE,M . The interference between weak
 F.E. Maas: Parity-violating electron scattering at the MAMI facility in Mainz (A4)
(Z) and electromagnetic () amplitudes leads to a PV
asymmetry ALR (ep) = (R L )/(R + L ) in the elastic
scattering cross section of right- and left-handed electrons
(R and L respectively), which is given in the framework
of the Standard Model [18]. ALR (ep) is of order parts per
million (ppm). The asymmetry can be expressed as a sum
of three terms, ALR (ep) = AV + As + AA .
 
  GpE GnE + GpM GnM
AV = aeq (1 4eq sZ )
2
, (1)
(GpE )2 + (GpM )2
Gp Gs + GpM GsM
As = aeq E p E2 , (2)
(GE ) + (GpM )2

(1 4s2Z ) 1 2 (1 + )GpM GpA
AA = a . (3)
(GpE )2 + (GpM )2
AV arises from the Z0 coupling to the proton vector cur-
rent and contains the electromagnetic nucleon form fac-
tors Gp,n
E,M . A possible strangeness contribution to the
proton electromagnetic vector form factors has been sep-
arated into As . The coupling to the proton axial cur-
rent is presented by AA and contains the neutral cur-
rent weak axial form factor GpA . e is the scattering an-
gle of the electron in the laboratory and Q2 the negative
square of the four momentum transfer. = Q2 /(4Mp2 ) and
 = [1+2(1+ ) tan2 (e /2)]1 represent purely kinematical
factors with Mp the proton mass. G and represent the
Fermi coupling constant as derived from muon decay and
the ne structure constant respectively. a denotes the fac-
tor (G Q2 )/(4 2). In order to average A0 = AV + AA
over the acceptance of the detector and the target length,
we take values for the electromagnetic form factors Gp,nE,M
from a parametrization (version 1, page 5) by Friedrich
and Walcher [19] and assign an experimental error of 3 %
to GpM and GpE , 5 % to GnM , and 10 % to GnE . For evalu-
ating A0 , electromagnetic internal and external radiative
corrections to the asymmetry and energy loss due to ion-
ization in the target have been calculated and they reduce
the expected asymmetry for our kinematics by 1.3 %.
Electro-weak quantum corrections have been applied
in the M S renormalization scheme according to [20]
and are contained in the factors eq , with s2Z =
sin2 W (MZ )M S = 0.23120(15) [21]. The electro-weak
quantum corrections to AA [22] are applied and absorbed
in the value of GpA .
The largest contribution to the uncertainty of A0
comes from the uncertainty in the axial form factor GpA ,
the electric form factor of the proton GpE , and the mag-
netic form factor of the neutron GnM . For our experimental
program at backward angles, the uncertainty stemming
from GpA will be more important in the separation of GsE
and GsM . Using a dierent target like deuterium will allow
to separate GpA , GsE and GsM . The parity admixture in the
ground state of deuterium is negligible [23].
4 The A4 experimental setup and analysis
The A4 experiment at MAMI has been developed and
build in order to measure a small (order ppm) parity-
109
helicity correlated
measurement of:
beam current,
beam energy
beam position
beam angle N R - NL
luminosity: A=
monitor system N R + NL
pol. electron source
Moellerpolarimeter (A1)
P=80%, I=20A
high power transmission
MAMI liquid Compton polarimeter
E=855MeV
E/E=10-6
Compton- hydrogen
E,
and beam dump
laser target
back scatter
polarimeter
e-
luminosity
e-
e-
e- E,
calorimeter:
1022 PbF2-crystals
Fig. 2. Measurement principle of the A4 experiment. The po-
larized electrons from the source are accelerated in the MAMI
accelerator to a maximum energy of 855 MeV. Scattered elec-
trons from the 10 cm hydrogen target are detected in the homo-
geneous 1022 channel PbF2 -Cherenkov calorimeter. The sen-
sitive measurement and stabilization of all electron beam pa-
rameters is crucial for the sensitivity of the experiment.
violating cross section asymmetry in the scattering of po-
larized electrons o unpolarized protons. It is complemen-
tary to other experiments since for the rst time counting
techniques have been used in a PV electron scattering ex-
periment. Possible systematic contributions to the exper-
imental asymmetries and the associated uncertainties are
of a dierent nature as compared to previous experiments,
which use analogue integrating techniques. Figure 2 shows
the measurement principle of the A4 experiment.
The polarized 570.4 and 854.3 MeV electrons were pro-
duced using a strained layer GaAs crystal that is illu-
minated with circularly polarized laser light [24]. Aver-
age beam polarization was about 80 %. The helicity of
the electron beam was selected every 20.08 ms by set-
ting the high voltage of a fast Pockels cell according to
a randomly selected pattern of four helicity states, either
(+P P P + P ) or (P + P + P P ). A 20 ms time
window enabled the histogramming in all detector chan-
nels and an integration circuit in the beam monitoring and
luminosity monitoring systems. The exact window length
was locked to the power frequency of 50 Hz in the labora-
tory by a phase locked loop. For normalization, the gate
length was measured for each helicity. Between each 20 ms
measurement gate, there was an 80 s time window for the
high voltage at the Pockels cell to be changed. The inten-
sity I = 20 A of the electron current was stabilized to
better than I/I 103 . An additional /2-plate in the
optical system was used to rotate small remaining linear
polarization components and to control the helicity corre-
lated asymmetry in the electron beam current to the level
of < 10 ppm in each ve minute run.
From the source to the target, the electron beam de-
velops uctuations in beam parameters such as position,
energy and intensity which are partly correlated to the
reversal of the helicity from +P to P . We have used a
system of microwave resonators in order to monitor beam
110 The European Physical Journal A
Fig. 3. Floor plan of the MAMI accelerator with the three
race track microtrons (RTM). The A4 experiment is located in
experimental halls 3 and 4. Feedback stabilization systems for
energy, position, angle and current had been developed for the
A4 experiment.
current, energy, and position in two sets of monitors sep-
arated by a drift space of about 7.21 m in front of the
hydrogen target. In addition, we have used a system of 10
feed-back loops in order to stabilize current, energy [25],
position, and angle of the beam. Figure 3 shows an outline
of the MAMI oor plan with the location of the A4 exper-
iment and the beam monitoring and stabilisation systems.
The polarization of the electron beam was measured
with an accuracy of 2 % using a Mller polarimeter which
is located on a beam line in another experimental hall [26].
Due to the fact that we had to interpolate between
the weekly Mller measurements, the uncertainty in the
knowledge of the beam polarization increased to 4 %.
The 10 cm high-power, high-ow liquid-hydrogen tar-
get was optimized to guarantee a high degree of turbulence
with a Reynolds-number of R > 2 105 in the target cell
in order to increase the eective heat transfer. For the
rst time, a fast modulation of the beam position of the
intense CW 20 A beam could be avoided. It allowed us
to stabilize the beam position on the target cell without
target density uctuations arising from boiling. The total
thickness of the entrance and exit aluminum windows was
250 m. The luminosity L was monitored for each helic-
ity state (R, L) during the experiment using eight water-
Cherenkov detectors (LuMo) that detect scattered par-
ticles symmetrically around the electron beam for small
scattering angles in the range of e = 4.4 10 , where
the PV asymmetry is negligible. The photomultiplier tube
currents of these luminosity detectors were integrated
during the 20 ms measurement period by gated integra-
tors and then digitized by customized 16-bit analogue-to-
digital converters (ADC). The same method was used for
Fig. 4. The electron beam enters from the left and hits the
hydrogen target at about 2.2 m above ground. Scattered elec-
trons are detected with the 1022 channel PbF2 -calorimeter,
which covers a scattering angle of 30 < e < 40 and an az-
imuthal angle of 360 . Part of the detector has been left out in
the drawing for better visibility.
all the beam parameter signals. A correction was applied
for the nonlinearity of the luminosity monitor photomul-
tiplier tubes. From the beam current helicity pair data
I R,L and luminosity monitor helicity pair LR,L data we
calculated the target density R,L = LR,L /I R,L for the
two helicity states independently.
To detect the scattered electrons we developed a
new type of a very fast, homogeneous, total absorption
calorimeter consisting of individual lead uoride (PbF2 )
crystals [27,28]. Figure 4 shows a CAD-drawing of the
calorimeter together with the hydrogen target. The mate-
rial is a pure Cherenkov radiator and has been chosen for
its fast timing characteristics and its radiation hardness.
This is the rst time this material has been used in a large
scale calorimeter for a physics experiment. The crystals
are dimensioned so that an electron deposits 96 % of its
total energy in an electromagnetic shower extending over
a matrix of 3 3 crystals. Together with the readout
electronics this allows us a measurement
of the particle
energy with a resolution of 3.9 %/ E and a total dead
time of 20 ns. For the data taken at 854.3 MeV only 511
out of 1022 channels of the detector and the readout
electronics were operational, for the 570.4 MeV data all
 F.E. Maas: Parity-violating electron scattering at the MAMI facility in Mainz (A4)
Fig. 5. Design drawing of the A4 readout electronics. The
upper part contains the analog sum, trigger and veto circuits
together with the digitization. In the lower part, the histogram-
ming and the VMEbus access is done. The system is about
3.5 m high.
the 1022 channels were installed. The particle rate within
the acceptance of this solid angle was 50 106 s1 .
Due to the short dead time, the losses due to double hits
in the calorimeter were 1 % at 20 A. This low dead time
is only possible because of the special readout electronics
employed. The signals from each cluster of 9 crystals were
summed and integrated for 20 ns in an analogue summing
and triggering circuit and digitized by a transient 8-bit
ADC. There was one summation, triggering, and digitiza-
tion circuit per crystal. The energy, helicity, and impact
information were stored together in a three dimensional
histogram. Neighboring crystals have to go to neighboring
electronics channels in the electronics resulting a ring
shape. Analogue summation and digitization has been
galvanically separated from histogramming and VMEbus
access. Figure 5 shows a design drawing of the fast A4
experiment electronics.
Figure 6 shows an energy spectrum of scattered par-
ticles. The number of elastic scattered electrons is deter-
mined from this histogram for each detector channel by in-
tegrating the number of events in an interval from 1.6 E
above pion production threshold to 2.0 E above the elas-
tic peak in each helicity histogram, where E is the en-
ergy resolution for nine crystals. These cuts ensure a clean
separation between elastic scattering and pion production
or -excitation which has an unknown PV cross section
asymmetry. The linearity of the PbF2 detector system
with respect to particle counting rates and possible eects
111
x 102
1800
Counts
1600
1400
Elastic Peak
1200 Elastic Cut
1000
800 Elastic Cut
600
400
200
0
0 50 100 150 200 250
ADC-Channel 32
Fig. 6. The dashed histogram shows a raw energy spectrum
of accepted particles from the hydrogen target as read directly
from the hardware memory of the readout electronics of the
lead uoride calorimeter. For the solid black curve, the raw
spectrum has been corrected for the dierential nonlinearity
of the ADC, i.e. for measured variations of the ADC channel
width. The elastic scattering peak position, the 0 -production
threshold and the -resonance position are indicated as well
as the lower and upper cut position for the extraction of NeR
and NeL as described in the text.
due to dead time were investigated by varying the beam
current. We calculate the raw normalized detector asym-
metry as Araw = (NeR /R NeL /L )/(NeR /R + NeL /L ).
The possible dilution of the measured asymmetry by back-
ground originating from the production of 0 s that sub-
sequently decays into two photons where one of the pho-
tons carries almost the full energy of an elastic scattered
electron was estimated using Monte Carlo simulations to
be much less than 1 % and is neglected here. The largest
background comes from quasi-elastic scattering at the thin
aluminum entrance and exit windows of the target cell. We
have measured the aluminum quasi-elastic event rate and
calculated in a static approximation a correction factor for
the aluminum of 1.030 0.003 giving a smaller value for
the corrected asymmetry.
Corrections due to false asymmetries arising from he-
licity correlated changes of beam parameters were applied
on a run by run basis. The analysis was based on the ve
minute runs for which the counted elastic events in the
PbF2 detector were combined with the correlated beam
parameter and luminosity measurements. In the analysis
we applied reasonable cuts in order to exclude runs where
the accelerator or parts of the PbF2 detector system were
malfunctioning. The analysis is based on a total of 7.3106
histograms corresponding to 4.8 1012 elastic scattering
events for the 854.3 MeV data and 4.8106 histograms cor-
responding to 21013 elastic events for the 570.4 MeV data.
For the correction of helicity correlated beam param-
eter uctuations we used multi-dimensional linear regres-
sion analysis using the data. The regression parameters
have been calculated in addition from the geometry of the
112 The European Physical Journal A

20
Asymmetry [ppm]

15
10
5
0
-5
-10
out out out out out out out out
-15 out
in in in in in in in in
-20
0 2 4 6 8 10 12 14 16
Sample No.
5
Asymmetry (ppm)

4
3
2
1
0
-1
-2
-3
-4 out in out in in out
-5
1 2 3 4 5 6
Sample No.
Fig. 7. The top plot shows the data samples of 854.3 MeV
data with the /2-plate in and out. The lower plots represents
the data sample for the 570.4 MeV data with the /2-plate in
and out as described in the text.
precisely surveyed detector geometry. The two dierent
methods agree very well within statistics.
The experimental asymmetry is normalized to the elec-
tron beam polarization Pe to extract the physics asym-
metry, Aphys = Aexp /Pe . We have taken half of our data
with a second /2-plate inserted between the laser system
and the GaAs crystal. This reverses the polarization of
the electron beam and allows a stringent test of the un-
derstanding of systematic eects. The eect of the plate
can be seen in g. 7: the observed asymmetry extracted
from the dierent data samples changes sign, which is a
clear sign of parity violation if, as in our case, the target
is unpolarized.
Our measured result for the PV physics asymmetry in
the scattering cross section of polarized electrons on unpo-
larized protons at an average Q2 value of 0.230 (GeV/c)2
is ALR (ep) = (5.44 0.54stat 0.26syst ) ppm for the
854.3 MeV data [29] and ALR (ep) = (1.36 0.29stat
0.13syst ) ppm for the 570.4 MeV data [30]. The rst error
represents the statistical accuracy, and the second rep-
resents the systematical uncertainties including beam po-
larization. The absolute accuracy of the experiment repre-
sents the most accurate measurement of a PV asymmetry
in the elastic scattering of longitudinally polarized elec-
trons on unpolarized protons.
in out
7 8
Fig. 8. Top: The solid line represents all possible combinations
of GsE + 0.225GsM as extracted from the work presented here at
a Q2 of 0.230 (GeV/c)2 . The densely hatched region represents
the 1- uncertainty. The recalculated result from the HAPPEX
published asymmetry at Q2 of 0.477 (GeV/c)2 is indicated by
the dashed line, the less densely hatched area represents the as-
sociated error of the HAPPEX result. Bottom: The solid lines
represent the result on GsE + 0.106GsM as extracted from our
new data at Q2 = 0.108 (GeV/c)2 presented here. The hatched
region represents in all cases the one--uncertainty with sta-
tistical and systematic and theory error added in quadrature.
The dashed lines represent the result on GsM from the SAM-
PLE experiment [31]. The dotted lines represent the result of
a recent lattice gauge theory calculation for s [32]. The boxes
represent dierent model calculations and the numbers denote
the references.
5 Conclusion
From the dierence between the measured ALR (ep) and
the theoretical prediction in the framework of the Stan-
dard Model, A0 , we extract a linear combination of
the strange electric and magnetic form factors for the
570.4 MeV data at a Q2 of 0.108 (GeV/c)2 of GsE
0.106 GsM = 0.071 0.036. For the data at 854.3 MeV
corresponding to a Q2 value of 0.230 (GeV/c)2 we extract
GsE + 0.225 GsM = 0.039 0.034. Statistical and system-
atic error of the measured asymmetry and the error in the
theoretical prediction of A0 been added in quadrature. In
g. 8 the results for the 570.4 MeV data are displayed.
 F.E. Maas: Parity-violating electron scattering at the MAMI facility in Mainz (A4)
Fig. 9. A compilation of the world PV asymmetry data. The
plot shows the dierence (Aphys A0 ) between the published
measured PV physics asymmetries Aphys and the asymmetry
A0 expected in the standard model assuming no strangeness. A
signicant dierence is a direct sign for strangeness contribu-
tion. The blue points have been measured detecting the scat-
tered electron, the red points denote the G0 experiment results
where the proton has been detected under forward angles.
A recent very accurate determination of the strangeness
contribution to the magnetic moment of the proton s =
GsM (Q2 = 0 (GeV/c)2 ) from lattice gauge theory [32]
would yield a larger value of GsE = 0.076 0.036 if the
Q2 dependence from 0 to 0.108 (GeV/c)2 is neglected.
The theoretical expectations for another quenched lattice
gauge theory calculation [33], for SU (3) chiral perturba-
tion theory [34], from a chiral soliton model [35], from a
quark model [36], from a Skyrme-type soliton model [37]
and from an updated vector meson dominance model t
to the form factors [38] are included in g. 8.
Recently the HAPPEX- and the G0-collaboration pub-
lished measurements which are reviewed in a dierent con-
tribution to these proceedings. Figure 9 shows a compila-
tion of the world data in the Q2 -range up to 0.5 (GeV/c)2 .
The plot shows the dierence (Aphys A0 ) between the
published measured PV physics asymmetries Aphys and
the asymmetry A0 expected in the standard model as-
suming no strangeness. A signicant dierence is a direct
sign for strangeness contribution. The blue points have
been measured detecting the scattered electron, the red
points denote the G0 experiment results where the proton
has been detected under forward angles.
Combining the HAPPEX data at Q2 = 0.1 (GeV/c)2
and the extrapolated G0 data and with our previous data
at this Q2 value gives further constraints on GsE and GsM .
Figure 10 shows the present status of the world data at
Q2 = 0.1 (GeV/c)2 in the GsE versus sGsM plane. The
SAMPLE data have been measured at backward angle.
A4, HAPPEX and G0 at forward angle. The HAPPEX He
data have been measured using a 4 He target, which has
no magnetic form factor due to the fact that the nucleus
has spin 0. One can extract a value for both GsE and GsM
separately: GsM = 0.62 0.31 and GsE = 0.012 0.029.
While the combined value for GsE has a sensitivity which
113
Fig. 10. The combination of all available data at forward and
backward angles at from H and He target constrains the pos-
sible values for GsE and GsM . The ellipse gives the two sigma
contour plot.
is now reaching the accuracy of the electromagnetic form
factors, the error on GsM is still very large.
6 Perspectives
Further constrains on GsM can be expected from further
measurements combining H and He data at forward an-
gles, as planned by the HAPPEX collaboration. A dier-
ent approach is to combine H and D data at backward
angles. We are preparing a series of measurements of the
parity-violating asymmetry in the scattering of longitu-
dinally polarized electrons o unpolarized protons and
deuterons under backward scattering angles of 140 <
e < 150 with the A4 apparatus at a Q2 of 0.23 (GeV/c)2
in order to separate the electric (GsE )and magnetic (GsM )
strangeness contribution to the electromagnetic form fac-
tors of the nucleon. We have changed the experimental
setup. The detector has been put on a rotatable plat-
form, construction work and reinstallation of detector and
2028 cables in the experimental hall had been nished by
April 2005. We have found a large background stemming
from photons coming from 0 -decay. The energy range
of those photons cover partly the same energy interval
as the energy range of the elastic scattered electrons. Our
electromagnetic shower calorimeter has the same response
function for electrons as for photons. 0 -production has
its own unknown parity-violating asymmetry. In order to
avoid pollution of our elastic signal with photons from 0 -
decay, we have installed an electron tagging system of 72
additional plastic scintillators between scattering cham-
ber and PbF2 -crystals. Figure 11 shows a schematic of the
114 The European Physical Journal A

dence signal. They exhibit a clear elastic peak with a good

signal-to-background ratio. The count rate under the elas-
tic peak corresponds to 80 %90 % of the expected elastic
count rate. At forward angles, this ratio had been 70 %.
We expect at backward angles a higher eciency since
the radiative tail is less pronounced at the beam energy
of 315.13 MeV. The black and red histograms (only partly
shown) correspond to the large background from photons
from 0 -decay. For these signals the scintillator system
had no coincident signal. The sum of both spectra would
Fig. 11. The drawing shows a cut through the upper part of give the total spectrum. The system will be commissioned
the A4 detector system. In the upper part, one sees the alu- in December 2005. Data taking will start in January 2006.
minum frame with seven PbF2 -crystals. The scintillator sys-
We plan to accumulate 1000 h with hydrogen target cor-
tem for detecting electrons in coincidence with the calorimeter
responding to an error in GsM of 0.13, corresponding to
is located between the PbF2 -crystals and the vacuum chamber.
Two concentrical rings contain 36 scintillator each. The elec-
a factor of three over the SAMPLE result and a factor of
tron rates are low enmough, so that one scintillator can cover two improvement over the present combined world data
14 PbF2 -crystals, i.e. two aluminum frames. at Q2 = 0.1 (GeV/c)2 . In order to improve the under-
standing on the systematical uncertainty coming from the
axial form factor, we will also collect the same amount of
statistics with a deuterium target.
setup. The scintillators are arranged in two rings symmet-
rically around the scattering chamber. Only charged par-
ticle traversing the scintillator produce an output signal. If This work has been supported by the DFG in the framework
a scintillator produces a signal, it is converted to a logical of the SFB 201 and SPP 1034. We are indebted to K.H. Kaiser
level which then serves for the short coincidence time of and the whole MAMI crew for their tireless eort to provide
25 ns as an additional bit in the histogramming circuit of us with good electron beam. We are also indebted to the A1-
the readout electronics. If the bit is set, and in addition a Collaboration for the use of the Mller polarimeter. My deep
signal arrives from the PbF2 -calorimeter, the event is his- thanks got to H. Arenhovel, H. Backe, D. Drechsel, J. Friedrich,
K.H. Kaiser and Th. Walcher. During the last ten years at
togrammed into one region of the histogramming memory.
the Institut fur Kernphysik it had been a pleasure to work
If there is an event in the calorimeter without tagging bit
with them. I want to thank them not only for their continuous
from the scintillators, it is histogrammed into a dierent encouragement but also for many stimulating discussions and
memory address range. Figure 12 shows 7 typical PbFs - also for keeping a lifely, scientic spirit at the institute.
calorimeter spectra taken with the electron tagging scin-
tillator system. The blue histograms show electron spectra
where the additional scintillator system had given a coinci-
References
Run 28256: Chan 918 pol1 ProjX EntriesMean 1.99541e+01
Normal: 3.17202e+06
Run 28256: Chan 919 pol1 ProjX EntriesMean 1.95670e+01
Normal: 2.89210e+06
Run 28256: Chan 920 pol1 ProjX EntriesMean 1.85214e+01
Normal: 2.72850e+06
Run 28256: Chan 921 pol1 ProjX EntriesMean 1.94098e+01
Normal: 2.60810e+06
1. T.D. Lee, C.N. Yang, Phys. Rev. 104, 254 (1956).
EntriesIntegral
Coinc: 7.18291e+05
1.58541e+06 EntriesIntegral
Coinc: 5.80886e+05
1.44674e+06 EntriesIntegral
Coinc: 5.12944e+05
1.36421e+06 EntriesIntegral
Coinc: 5.24322e+05
1.30461e+06

Beam
norm
-----------------------
norm pol1 1 raw
Kurtosis -2.86987e-01

conditions:
pol0 1 T=300
raw s, I=19.9 muA

Expected elastic: 3.91186e+05

norm pol1 1 dnl
Beam
-----------------------
norm pol1

conditions:
norm
1 raw
Kurtosis -6.04778e-01

pol0 1 T=300
raw s, I=19.9 muA

Expected elastic: 3.56634e+05

norm pol1 1 dnl
Beam
-----------------------
norm pol1 1 raw
Kurtosis -5.92825e-01

conditions:
norm T=300 s, I=19.9 muA
pol0 1 raw

Expected elastic: 3.22953e+05

norm pol1 1 dnl
Beam
-----------------------
norm pol1

conditions:
norm
1 raw
Kurtosis -5.11082e-01

pol0 1 T=300
raw s, I=19.9 muA

Expected elastic: 2.98658e+05

norm pol1 1 dnl
2. C.S. Wu et al., Phys. Rev. 105, 1413 (1957).
norm pol0 1 dnl norm pol0 1 dnl norm pol0 1 dnl norm pol0 1 dnl

3. T.R. Prescott et al., Phys. Lett. B 84, 524 (1979).

peak edge peak edge peak edge peak edge
4. P.A. Souder et al., Phys. Rev. Lett. 65, 694 (1990).
5. J. Ahrens et al., Bucl. Phys. A 446, 377c (1985).
8000 8000 8000 8000
6. S. Kox, D. Lhuillier, F. Maas, J. Van de Wiele (Editors),
6000 6000 6000 6000

4000 4000 4000 4000 From Parity Violation to Hadronic Structure and More,
2000

0
2000

0
2000

0
2000

0
Proceedings of PAVI2004, Eur. Phys. J. A 24, s02 (2005).
0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100
7. T.R. Donoghue, E. Golowich, B.R. Holstein, Dynamics of
Run 28256: Chan 922 pol1 ProjX EntriesMean 1.80404e+01
Normal: 2.59947e+06
Run 28256: Chan 923 pol1 ProjX EntriesMean 1.93011e+01
Normal: 2.02822e+06
Run 28256: Chan 924 pol1 ProjX EntriesMean 1.78771e+01
Normal: 1.42570e+06

Beam
EntriesIntegral

norm
Coinc: 5.00314e+05
1.29944e+06

-----------------------
norm pol1 1 raw
Kurtosis -4.73486e-01

conditions:
pol0 1 T=300
raw s, I=19.9 muA Beam
EntriesIntegral
Coinc: 3.88322e+05
1.01361e+06

-----------------------
norm pol1 1 raw
Kurtosis -5.08935e-01

conditions:
norm pol0 1 T=300
raw s, I=19.9 muA Beam
EntriesIntegral

norm pol1
Coinc: 3.30970e+05

1 raw
7.12172e+05

-----------------------
Kurtosis -2.00809e-01

conditions:
norm T=300 s, I=19.9 muA
pol0 1 raw
the Standard Model, rst paperback edition (with correc-
tions) (Cambridge University Press, Cambridge, 1992).
Expected elastic: 2.72263e+05 Expected elastic: 2.48168e+05 Expected elastic: 2.24327e+05
norm pol1 1 dnl norm pol1 1 dnl norm pol1 1 dnl

norm pol0 1 dnl norm pol0 1 dnl norm pol0 1 dnl

peak edge peak edge peak edge

8. H.D. Politzer, Phys. Rev. Lett. 30, 1346 (1973).
9. D.J. Gross, F. Wilczek, Phys. Rev. Lett. 30, 1343 (1973).
8000 8000 8000 10. M. Jamin, Phys. Lett. B 538, 71 (2002).
6000

4000
6000

4000
6000

4000
11. M.M. Pavan et al., PiN Newslett. 16, 110 (2002).
2000 2000 2000 12. X.-D. Ji, Phys. Rev. Lett. 74, 1071 (1995).
0
0 20 40 60 80 100
0
0 20 40 60 80 100
0
0 20 40 60 80 100 13. M. Goncharov et al., Phys. Rev. D 64, 112006 (2001).
14. M. Tzanov et al., Nutev neutrino dis., hep-ex/0306035,
Fig. 12. The blue histograms show electron spectra where 2003.
the additional scintillator system had given a coincidence sig- 15. M. Gluck, E. Reya, A. Vogt. Eur. Phys. J. C 5, 461 (1998).
nal. They exhibit a clear elastic peak with a good signal-to- 16. E. Leader et al., Phys. Rev. D 67, 074017 (2003).
background ratio. The black and red histograms (only partly 17. D.B. Kaplan et al., Nucl. Phys. B 310, 527 (1988).
shown) correspond to the large background from photons from 18. M.J. Musolf et al., Phys. Rep. 239, 1 (1994).
0 -decay. Both spectra are taken always in parallel for better 19. J. Friedrich, Th. Walcher, Eur. Phys. J. A 17, 607 (2003).
control of systematic eects. 20. W.J. Marciano, A. Sirlin, Phys. Rev. D 29, 75 (1984).
 F.E. Maas: Parity-violating electron scattering at the MAMI facility in Mainz (A4)
21. S. Eidelmann et al., Review of particle properties, Phys.
Lett. B 592, 1 (2004).
22. S.-L. Zhu et al., Phys. Rev. D 62, 033008 (2000).
23. G. Kuster, H. Arenhovel, Nucl. Phys. A 626, 911 (1997).
24. K. Aulenbacher et al., Nucl. Instrum. Methods A 391, 498
(1997).
25. M. Seidl et al., High precision beam energy stabilisation of
the Mainz microtron MAMI, in Proceedings of the EPAC
2000 (2000) p. 1930.
26. P. Bartsch, Aufbau eines Meller-Polarimeters fur
die Drei-Spektrometer-Anlage und Messung der He-
lizitatsasymmetrie in der Reaktion p(e,e p) 0 im Bereich
der -Resonanz, Dissertation Mainz, 2001.
27. F.E. Maas et al., Proceedings of the ICATPP-7 (World
Scientic, 2002) p. 758.
115
28. P. Achenbach et al., Nucl. Instrum. Methods A 465, 318
(2001).
29. F.E. Maas et al., Phys. Rev. Lett. 93, 022002 (2004).
30. F.E. Maas et al., Evidence for strange quark contributions
to the nucleons form-factors at q 2 = 0.108 (Gev/c)2 , nucl-
ex/0412030, 2004.
31. D.T. Spayde et al., Phys. Lett. B 583, 79 (2004).
32. D.B. Leinweber et al., Phys. Rev. Lett. 94, 212001 (2005).
33. R. Lewis et al., Phys. Rev. D 67, 013003 (2003).
34. T.R. Hemmert et al., Phys. Rev. C 60, 045501 (1999).
35. A. Silva et al., Eur. Phys. J. A 22, 481 (2004).
36. V. Lyubovitskij et al., Phys. Rev. C 66, 055204 (2002).
37. H. Weigel et al., Phys. Lett. B 353, 20 (1995).
38. H.-W. Hammer et al., Phys. Rev. C 60, 045204 (1999).
Eur. Phys. J. A 28, s01, 117 127 (2006)
DOI: 10.1140/epja/i2006-09-013-6 EPJ A direct
electronic only

Virtual Compton Scattering at MAMI

N. dHosea
CEA-Saclay, DAPNIA/SPhN, F91191 Gif-sur-Yvette Cedex, France

/
Published online: 17 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. Virtual Compton Scattering (VCS) o the proton is a recent eld of investigation of the nucleon
structure. VCS at threshold gives access to the Generalized Polarizabilities (GPs) of the proton. The
qualities of both the beam and the high-resolution spectrometers available at the Mainz Microtron MAMI
allowed us to perform at rst such delicate experiments. This paper deals with dierent experiments
dedicated to the GPs measurements. They are realized without and with polarization, below and just
above pion threshold.
PACS. 13.60.Fz Elastic and Compton scattering 14.20.Dh Protons and neutrons 25.30.Rw Electro-
production reactions

1 Virtual Compton Scattering and
polarizabilities
One of the main challenges of hadronic physics in the
regime of strong (non-perturbative) QCD is to identify
the relevant degrees of freedom of the nucleon. Though
the small distance structure is rather well described by
point-like quarks and gluons, its structure at larger dis-
tance is not so well understood. There exist many models
ranging from constituent quark models to chiral models.
Polarizabilities are one of the fundamental observ-
ables to describe the internal structure of the nucleon and
they have been investigated with real Compton scattering
(RCS) since the early 1950s. As the light scattering on
atmospheric atoms which gives the well known Rayleigh
eect for blue skies and red sunsets through oscillation
of the electrons inside the atoms, real Compton scatter-
ing sheds light on the nucleon structure. This is clearly
illustrated in a common denition of the electric polariz-
ability E in a non relativistic approach at the rst-order
perturbation for an applied electric dipole moment D:
the world global average of the electric (E ) and magnetic
(M ) polarizabilities on the proton is based on an experi-
mental study investigated at MAMI with the tagged real
photon beam [1].
E =(11.9 0.5(stat.) 1.3(syst.) 0.3(mod.))104 fm3 ,
M =(1.2 0.7(stat.) 0.3(syst.) 0.4(mod.))104 fm3 .
We can note the small size of the polarizabilities which
reveals the feature that the nucleon is strongly bound.
For comparison the electric polarizability of the hydrogen
atom is of the order of the atomic volume, and the electric
polarizability of the proton E is only 0.05 per cent of its
volume. Furthermore the magnetic polarizability is still
smaller, one tenth of the electric polarizability.
Virtual Compton scattering (VCS) o the proton
refers to the reaction p p, where stands for an
incoming virtual photon of four-momentum squared Q2 .
This reaction is experimentally accessed through photon
electroproduction ep ep. The corresponding Feynman
diagram is indicated in g. 1.
In the 1960s the VCS appeared as a rather un-
wanted contribution to radiative corrections to electron
|N  |Dz |N |2
E = 2N  =
N . scattering on a proton [2]. It was mentioned as proton
EN  E N
In this formula N  indicates each nucleon resonance. The
polarizability is then sensitive to all the excitation spec-
trum of the nucleon (even if the low energy of the per-
turbating photon does not allow the real formation of the
nucleon resonances).
The Mainz laboratory has a long tradition in this eld.
Several experiments have been dedicated to the determi-
nation of proton, neutron or pion polarizabilities. Today
a
e-mail: ndhose@cea.fr
Bremsstrahlung. In 1974 Arenhovel and Drechsel [3], from
the Institut fur Kernphysik at Mainz, considered the VCS
for the rst time as a good way to measure generalized
polarizabilities (GPs). Only in 1995 with the new genera-
tion of facilities of high duty cycle to investigate exclusive
reactions, and with new theoretical concepts it regained
interest.
The general theoretical framework for VCS at thresh-
old has been extensively described by Guichon et al. [4,5]
and the Mainz theoretical group conducted by Drechsel [6,
7]. VCS reaction at threshold means that the produced
118 The European Physical Journal A

When the magnetic eld is applied, they are modied in

such a way that the induced magnetic moment is anti-
e q small
parallel to the magnetic eld (Lenz law). Quarks and pi-
e ons thus give rise to two dierent contributions: para
and diamagnetic or resonant and non resonant
( q ) contributions to the magnetic polarizability M .
* ( q ) At present the GPs have been calculated in the frame-
work of various theoretical models [9,10,11,12,13,14,15]
yielding quite dierent results with regard to both their
N , N* , , ... absolute value and their Q2 dependence. Figure 2 presents
p p dierent theoretical predictions as functions of Q2 :

Fig. 1. The VCS graph for the proton. Non relativistic constituent quark model (CQM) is
based on the assumption that baryons are composed of
three massive quarks moving within a harmonic oscil-
photon has a small enough momentum or that its electric lator conning potential and additional hyperne in-
(E) and magnetic (M) elds look constant over the size of teractions. One of its success is to explain most of the
the nucleon. In the following the three-momenta absolute observed nucleon resonance mass spectrum. Calcula-
values of the virtual and real photons in the photon-proton tions have been performed in this framework by Gui-
center of mass (c.m.) system are noted q and q , respec- chon, Liu and Thomas [4,9] and Pasquini, Scherer and
tively, and vary independently, this is in contrast with real Drechsel [10].
Compton scattering where q = q . Here the low momen- Phenomenological approach can be realized with an
tum q of the produced real photon denes the size of the eective Lagrangian model (ELM). Such a calculation
electromagnetic (EM) perturbation, while the momentum has been performed by Vanderhaeghen [11] which in-
of the virtual photon q (or the four-momentum squared cludes the eects of all the rst nucleon resonances and
Q2 ) sets the scale of the observation of the nucleon inter- 0 exchange in the t channel.
nal structure.
These two rst kinds of model describe well all the reso-
In the low momentum regime the reaction can be in-
nant contributions, but not the non-resonant one. Their
terpreted as electron scattering on a nucleon placed in a
limitation is that they have no relationship to chiral sym-
quasi-constant applied EM eld [5]. The induced motion
metry. This is an important property of QCD which gov-
of the nucleon as a whole can be eliminated thanks to a
erns much of low-energy hadron physics. The pion is the
low-energy theorem [8], so one is left with the deformation,
Goldstone boson of spontaneously broken chiral symme-
due to the applied eld, of the nucleon internal currents
try, and plays a very special and major role at low energy.
J (r) and the electron scattering measures its Fourier
The two next groups of calculations respect chiral sym-
transform J (Q). To lowest order in QED , J (Q) is
metry.
linear in the applied eld and the 6 coe cients of propor-
tionality are the GPs [4,5,6,7]. When Q2 = 0 two of them A simple model to describe interaction of Dirac parti-
reduce to the usual polarizabilities E and M measured cles with a chiral eld is the linear sigma model (LSM)
in real Compton scattering. Analogously to the form fac- in the limit of an innite sigma mass. Though this
tors for elastic scattering, which describes the charge and model is not a very realistic description of the nucleon,
magnetization distributions, VCS gives access to the de- nevertheless it fullls all the relevant symmetries like
formation of these distributions by an external EM eld, Lorentz, gauge and chiral invariance. A complete cal-
and will yield valuable information about the non pertur- culation of all the one-loop diagram contributions (for
bative structure of the nucleon. the photon interaction with a nucleon-pion system) has
This can be illustrated by a very naive picture of the been performed by Metz and Drechsel [12].
polarizabilities which are the results of an electromag- Chiral perturbation theory (ChPT) is a very system-
netic perturbation applied to the nucleon components. An atic and consistent approach with a most general La-
electric eld moves positive and negative charges inside grangian based on QCD symmetries. Heavy-baryon
the proton in opposite directions. The induced electric chiral perturbation theory allows for a systematic per-
dipole moment is proportional to the electric eld, and turbative expansion in powers of small parameters
the proportionality coe cient is the electric polarizability (no- ted p) as quark masses, inverse of hadron masses
E which measures the rigidity of the proton. A magnetic or external momenta. Hemmert, Holstein, Knocklein
eld acts dierently on the quarks and the pion cloud. The and Scherer [13] have performed a third order O(p3 )
quarks (of spin 1/2) align their magnetic moment parallel calculation for all the GPs while Kao and Vander-
to the magnetic eld giving the strong magnetic excitation heaghen [16] have performed a fourth order O(p4 ) cal-
of the (1232) resonance. The pions are at low energy, an culation but only for the spin polarizabilities which
essential element of the structure of the nucleon notably at exclude predictions for E and M . Nevertheless pre-
its surface giving the famous representation of a pion cloud vious calculations for E (Q2 ) and M (Q2 ) have been
surrounding the nucleon. The pions (of spin 0) distributed realized at Q2 = 0 by Bernard, Kaiser, Schmidt
at the surface of the proton, will generate eddy currents. and Meissner [17] including all terms to order O(p4 )
 N. dHose: Virtual Compton Scattering at MAMI
Table 1. prediction at Q2 = 0 in the heavy baryon chiral
perturbation theory.
Calculation at O(p3 ) [13] Calculation at O(p4 ) [17]
E = 12.5 10 4 fm3 E = 10.5 10 4 fm3
M = 1.25 10 4 fm3 M = 3.5 10 4 fm3
14 6 q
E (10 fm )
M (10 fm )
3
12 5 p
-4
-4
10 4
3 (a)
8
2 Fig. 3. The p(e, e p) reaction. The initial, nal electron and
6
1 initial, nal proton quadri-momenta are k, k and p, p re-
4 0
2 -1
0 -2
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
2 2 2 2
Q (GeV ) Q (GeV )
Fig. 2. Evolution of the electric and magnetic polarizabili-
ties with Q2 . Experimental results [1] at Q2 = 0 (with only
statistical errors) and ve theoretical predictions CQM [10],
ELM [11], LSM [12], ChPT [13] and DR [15] are reported. See
the text for comments.
and also (1232)-resonance contribution via counter
terms. They found that in the case of the magnetic
polarizability, a large positive contribution from the
(1232)-resonance is largely canceled by a negative pi-
onic contribution, which gives a rather small resulting
value. The results at Q2 = 0 are reported in table 1.
Another consistent and unied approach for RCS and
VCS has been given by the Mainz theoretical group
through the dispersion relation formalism (DR) [14,15]
which connects the low energy nucleon structure quanti-
ties as polarizabilities to the nucleon excitation spectrum.
A more detailed description will be given in the next sec-
tion as this formalism is also used to extract the GPs.
In g. 2 we observe a relatively sharp fall-o of the
electric polarizability with increasing momentum trans-
fer Q2 , while we can remark for the HBChPT, DR and
ELM models a rise of the magnetic polarizability at very
low transfer and then a decrease at larger transfer. This
remarkable eect has its origin in the dominance of dia-
magnetism caused by the pion cloud at long distance and
the dominance of paramagnetism due to a quark core at
short distance. It is thus clear that the GPs are sensitive
to the respective role of quark and pion degrees of free-
dom and as such they are very valuable new observables
to compare theory with experiment.
2 The unpolarized experiments at threshold
A pioneer VCS experiment has been realized at MAMI
at Q2 = 0.33 GeV2 [18] during the years 1995-6-7, and
then two experiments have been performed in two other
119
k
k
q
=
p p
k q k
k k k k q
q + +
p p p p p
(b) (c)
spectively. The nal photon quadri-momentum is q . In the
one photon exchange approximation, a) and b) correspond to
the Bethe-Heitler (BH) process. c) corresponds to the Vir-
tual Compton Scattering (VCS) process. We note q the quadri-
momentum of the virtual photon exchanged in the VCS pro-
cess, that is q = k k (Q2 = q 2 ).
complementary kinematical regimes at Jeerson Lab.
(Q2 = 1 and 2 GeV2 ) [19] in 1998 and MIT-Bates (Q2 =
0.05 GeV2 ) [20] in 2000. They are long and delicate exper-
iments and they rely on a careful analysis of the data.
2.1 Theoretical framework
As it was mentioned the general theoretical framework for
VCS is extensively presented in the following references [4,
5] and [6,7]. Only the relevant points for the analysis of
these experiments will be discussed here. In the reaction
ep ep, the nal photon can be emitted either by the
electrons, referring to the Bethe-Heitler (BH) process, or
by the proton, giving access to the VCS process (see g. 3).
The BH process dominates and interferes strongly with
the VCS process. The amplitude is the sum of the BH,
Born and Non-Born amplitudes. The two last-named refer
to the proton radiation: the Born amplitude depends only
on the static properties of the proton (charge, mass) and
elastic form factors, while the Non-Born amplitude con-
tains dynamical internal structure information in terms of
generalized polarizabilities. The dierential cross section
has the form
d5 exp
= (1)
dklab [de ]lab [dp ]CM
 
(2)5 klab q
M q M,
32m klab s
wherein klab , klab are the moduli of the incoming and out-
going electron momentum, respectively. The relevant kine-
matical variables of the problem are q and q previously
dened;  the virtual photon polarization; and the two
spherical angles indicating the CM real photon direction
on a globe with the virtual photon as a pole. is the CM
angle between the real and virtual photons while rep-
resents the angle between the two electron plane and the
photon-proton plane. In the precedent formula stands
120 The European Physical Journal A

for a phase space factor. M is the coherent sum of the Table 2. List of the 10 GPs with the corresponding electro-
dierent amplitudes: magnetic transitions. Their relation with the polarizabilities
1   BH 2 obtained in real Compton scattering are indicated. 6 GPs are
M= T + T V CS  , (2) independent. Our choice is a priori arbitrary, and is realized
4 spin
by the 6 surrounded GPs.
1   BH 2
= T + T Born + T N onBorn  . EM transition VCS GPs RCS polarizabilities
4 spin Q2 0

M(C0 M 1)S=1
P (11,00)1 0
The low energy theorem (LET) from Low [8] states that 
in an expansion in powers of the real photon energy q M (C2 M 1)S=1
P (11,02)1
8 4
27 e2
(2 + 4 )

(but xed arbitrary q), the rst term of the amplitudes M (M 1 M 1)S=0
P (11,11)0
8 4
(M )
1 3 e2
T BH and T Born is of the order q (well-known infra-
M (M 1 M 1)S=1
P (11,11)1
0
red divergence), while the rst term of T N onBorn is of the
M(C2,E2 M 1)S=1
P (11,2)1 = 0
order q :
1

b1 (q, , , ) M(C1 E1)S=1

P (01,01)1
T BH + T Born = + O(q0 ), (3)  0
q M (C1 E1)S=0
P (01,01)0
2 4
(E )
3 e2
T N onBorn = b1 (q, , , )q + O(q2 ). (4) M(M 2 E1)S=1 P (01,12)1 3
2 4
e2
(3 )
M(C1,E1 E1)S=1 P (01,1)1 = 0
Consequently in the low energy limit of the nal photon,
the cross section is independent of the dynamical nucleon M(C1,E1 E1)S=0 P (01,1)0 0
=
structure [8], and can be evaluated using only the known
BH and Born amplitudes. This can be summarized by the
where v1 (, , q), v2 (, , q) are known kinematical fac-
following equation:
tors. PLL (q), PTT (q), PLT (q) are structure functions re-
d5 exp (q, q , , , ) = d5 BH+Born (q, q , , , ) (5) lated to the GPs(q) with some kinematical factors:

+ q 0 (q, , , ) + O(q ) 2
PLL = 2 6mGE P (01,01)0 (7)
where d5 is a notation for the dierential cross sec- ! "
tion d5 /dklab [de ]lab [dp ]CM . 0 (q, , , ) is the lead- PTT = 3GM q 2
2P (01,12)1
P (11,11)1
/q0
ing term in the expansion in powers of the real photon #
3 mq
momentum q . It corresponds to the interference between PLT = GE P (11,11)0
1
the term of order q in the BH+Born amplitude and the 2 Q
 
leading order term of order q in the Non-Born amplitude.
1
3Q q2 (11,02)1
+ GM P (11,00)1
+ P
It contains the dynamical internal structure information 2 q 2
of the proton, parametrized by 6 generalized polarizabil-
ities given for the electric and magnetic dipole radiation where m stands for the proton mass, GE and GM denote
of the outgoing real photon. the form factors evaluated at Q2 and q0 is the CM virtual
We note L(L ) the initial (nal) photon orbital angu- photon energy at q = 0. (See footnote 2 .)
lar momentum, ( ) the type of multipole transition (0 The two structure functions PLL (q) PTT (q)/ and
for Coulomb, 1 for Magnetic, 2 for Electric), and S the PLT measured in an unpolarized VCS experiment are the
type of the transition at the nucleon side (non-spin-ip sum of two contributions: one coming from the scalar or
S = 0 and spin-ip S = 1). Assuming that the emitted spin-independent polarizabilities and another one coming
real photon has low energy, we may use the dipole ap- from the spin-dependent polarizabilities (which vanishes
proximation (L = 1). For a dipole transition in the nal at Q2 = 0).
state, parity and angular momentum conservations lead
to 10 GPs presented in table 2. Crossing symmetry and
charge conjugation invariance provide 4 relations between 2.2 The MAMI experiment
the 10 GPs and we are left with 6 independent GPs: 2
scalar (S = 0) and 4 spin-dependent (S = 1) polarizabil- With the high luminosity and high duty cycle provided by
ities, functions of q (or equivalently Q2 = Q2 |q =0 . See the 855 MeV Mainz Microtron MAMI it was possible to in-
footnote 1 ). The choice of 6 GPs is a arbitrary, and can be vestigate the measurement of the small photon electropro-
realized for example by the 6 surrounded GPs in table 2. duction cross section at threshold. Absolute cross sections
In an unpolarized measurement, 0 (q, , , ) can be d5 exp [18] have been measured at Q2 = 0.33 GeV2 using
written as the three-spectrometer facility [21] of the A1 collaboration
at MAMI (see g. 4). The scattered electron and the re-
0 (q, , , ) = v1 (, , q)(PLL (q) PTT (q)/) (6) coiling proton were detected in coincidence with two of the
+ v2 (, , q)PLT (q) high-resolution magnetic spectrometers. The photon pro-
 duction process was selected by a cut on the missing mass
1
Q2 = Q2 |q =0 = 2m ( m2 + q2 m) where m stands for 
the proton mass. 2
q0 = m m2 + q2 .
 N. dHose: Virtual Compton Scattering at MAMI 121

0.7
0.6
0.5
0.4

0.3

0.2
-100 0 -100 0

0.2

0.1
0.09
0.08
0.07
0.06
-100 0 -100 0 -100 0
(deg)

Fig. 6. Dierential cross sections for the reaction ep ep

Fig. 4. The 3 high-resolution spectrometer facility of the A1
collaboration at MAMI.
as a function of for xed q, , and for ve values of the
real photon momentum q . The known part of the cross sec-
tion d5 BH+Born , is presented by the solid lines. The exper-
imental data points d5 exp deviate from the solid lines as q
increases the eect of the proton polarizabilities. The dot-
ted lines represent the expected cross sections with the eect
of the polarizabilities measured by the two structure functions
deduced from this experiment.

1600
determined by the acceptance of the two spectrometers
around 0 and 180 . The spherical angles and are de-
1400 ned such that = 0 corresponds to the half plane con-
taining the electron momenta. To ease the presentation
1200 the data are plotted with ranging from 180 to +180 ;
1000
the negative values corresponding in fact to = 180 .
The wide range of from 141 to +6 covers the back-
800 ward direction relative to the incoming and outgoing elec-
trons. Here, the VCS contributions are dominant because
600 the electron radiations (BH) are emitted predominantly
in the electron directions.
400
The cross sections d5 BH+Born are presented by
200 the solid lines in g. 6. At small photon momentum
q = 33.6 MeV/c the agreement between the radiatively
0
corrected data and d5 BH+Born is excellent, and the
-5000 0 5000 10000 15000 20000 25000
2 2
Mx (MeV ) deviation from this known cross section increases when
q increases, as expected from the eect of the proton
Fig. 5. Missing mass spectrum obtained for the setup at q = polarizabilities. In order to determine accurately the
111.5 MeV/c. polarizabilities, a careful analysis of possible systematic
errors on the deviation is of particular importance. First
the BH and Born contributions rely on the knowledge of
around zero, which was possible thanks to the excellent the proton form factors. Consequently we also measured
resolution of the facility (momentum resolution of 104 the absolute elastic scattering cross section for each
and angular resolution better than 3 mrad) (see g. 5). kinematic setting of the VCS experiment. These measure-
The aim of this rst VCS experiment below pion ments validate the use of the form factor parametrization
threshold was to measure the ve-fold dierential cross from Hohler [22] at a precision better than 1%. Second
sections in a wide photon angular range, at 5 values of the the radiative corrections, which are of the order of 20% of
photon momentum q : 33.6, 45, 67.5, 90, and 111.5 MeV/c the cross section, have been evaluated by Vanderhaeghen
(presentation in g. 6). The 3 other kinematical variables et al. [23]. The systematic uncertainties are estimated
were held xed, namely the virtual photon momentum, to equal 2% for the calculation performed to order 4
q = 600 MeV/c (Q2 = 0.33 GeV2 ), the virtual photon in the VCS cross section. Third the luminosity and the
polarization  = 0.62. The out-of-plane angle range is detector e ciencies are controlled within an accuracy
122 The European Physical Journal A

5 5 BH+Bo , , -2
(d -d )/q = 0 + q 1 + ... (in GeV )
0.5 10

0 /v2 (GeV-2)
0.25 7.5
0
5
-0.25
0.5 -0.5 2.5

0.25 0
0
-2.5
-0.25
0.5
-0.5 -5

0.25
0
-0.25
0.5
-0.5
0.25
0 Fig. 8. Compilation of the complete data set (for the 14 angles)
-0.25
-0.5
,
q (MeV/c)
Fig. 7. (d5 d5 BH+Born )/q studied as a function of the
real photon momentum q for the 14 measured scattering an-
gles . The intercept at origin is 0 . In a rst method it is
determined at each scattering angle by the mean value in
the investigated real photon momentum range (solid line). The
dash-dotted, dashed and dotted lines show evolutions in the
framework of the 2nd, 3rd and 4th methods, respectively.
of 1%, the solid angles are determined within an
accuracy of 2% using a Monte Carlo [24] simulation
which reproduces perfectly the missing mass spectra. All
these uncertainties are constant over the angular range
of the real photon and are controlled by the fairly good
agreement between the radiatively corrected data and the
predicted BH and Born cross section at small q . However
small imperfections in the spectrometer optic calibration
which could provide distortion of the angular distributions
are estimated to give a variation of cross section of 2.5%.
Figure 7 shows the behavior of (d5 d5 BH+Born )/
q as a function of the real photon momentum q for the

14 measured scattering angles . The goal is to determine
the intercept at origin (noted 0 in eq. (5)), and this gure
illustrates the basic di culty of this experiment that is the
increase of the statistical errors when q decreases. Four
methods are then considered in the following.
First method based on the LET: As is apparent in g. 7,
there is no strong evolution with the real photon momen-
tum. Therefore we make the hypothesis that there is no
q dependence in (d5 d5 BH+Born )/q . 0 is then de- in table 3. This third method, mainly realized to jus-
termined at each scattering angle by the mean value of
the data at the 5 photon energies.
-7.5
-10
-0.1 0 0.1 0.2 0.3 0.4 0.5
v1/v2 ()
of 0 /v2 as a function of v1 /v2 . The data are reasonably well
aligned; the errors indicated are statistical only. This allows
to extract the two structure functions PLL PTT / and PLT
with statistical errors and the 2 given.
Figure 8 presents the complete data set (for the 14
angles) of 0 /v2 as a function of v1 /v2 (cf. eq. (6)). The
data are reasonably well aligned, which suggests that the
higher-order terms in the expansion of the cross section
(cf. eq. (5)) are not so important. This good alignment for
a wide angular range is also indicative of the consistency of
the experimental data. We extract the two structure func-
tions PLL PTT / and PLT as the slope and intercept
of a linear t to the data (according to eq. (6)) [18].
Second method: We make the hypothesis of a linear
evolution with the real photon momentum for each angle
which is tted to the data. The result is indicated by the
dash-dotted line in g. 7.
Third method: The q evolution is supposed to be gov-
erned by the interference between the complete BH and
Born amplitudes considered at all order in the q expan-
sion (complete eq. (3)) and the Non-Born amplitude trun-
cated at the rst order in the q expansion (truncated
eq. (4)). The only parameters are a priori the 6 gen-
eralized polarizabilities contained in the rst and only
term of the considered Non-Born amplitude. They are
adjusted with a best t on the complete set of 14 5
data. In order to have a better convergence, the polariz-
ability P (01,01)0 (Q2 = 0.33 GeV2 ) is xed by the result
obtained in real Compton scattering scaled by the electric
form factor and P (11,02)1 (Q2 = 0.33 GeV2 ) is xed at 0 (it
corresponds to the quadrupolar deformation of the N-
transition which is expected to be very small). The result
of the t for the 4 remaining polarizabilities is presented
tify a rather at q evolution of (d5 d5 BH+Born )/q
presented by the dashed points in g. 7, allows one to
 N. dHose: Virtual Compton Scattering at MAMI 123

Table 3. Results for the polarizabilities extracted in the third Table 4. The structure functions determined in the MAMI
method. These results are compared to the heavy-baryon chiral experiment using the four methods and compared to model
perturbation theory (HBChPT) predictions [13]. predictions at Q2 = 0.33 GeV2 and  = 0.62. The errors are
statistical only, except for the rst method where two system-
Third Method HBChPT Units
atic errors are indicated in brackets. The prediction for DR
PLL P (01,01)0 0.0626 xed 0.056 fm3 model is given for 2 values of and close to the values
PTT P (11,11)1 +0.0048 0.0034 +0.001 fm3 determined experimentally.
P (01,12)1 0.0123 0.0026 0.008 fm4
PLT P (11,11)0 0.0384 0.0186 0.034 fm3 Q2 = 0.33 GeV2 PLL PTT / PLT
P (11,00)1 0.157 0.070 0.096 fm2  = 0.62 (GeV 2 ) (GeV 2 )
method 1 [18] 23.7 2.2 5.0 .8
P (11,02)0 0. xed +0.003 fm4
(2 = 1.4) (4.3 0.6) (1.4 1.1)
method 2 23.7 8.1 7.8 3.0
(2 = 1.3)
determine some spin polarizabilities with reasonable pre- method 3 33.6 11.7 6.5 4.2
cision (notably P (01,12)1 ). (2 = 1.7)
method 4 23.2 3.0 3.2 2.0
Fourth method using Dispersion Relations: This
(2 = 1.5) ( = 1.6 0.2) ( = 0.5 0.2)
method was used after the publication [18] of the rst
HBChPT [13] 26.3 5.7
VCS MAMI experiment at the sight of the other exper-
LSM [12] 10.9 0.2
iments of JLab and MAMI where a rather at q evo-
ELM [11] 5.9 1.9
lution was not so obviously conrmed by the data. This
NRQCM [9] 11.0 3.5
method is based on the formalism of Dispersion Relations
NRQCM [10] 14.7 4.5
(DR) [14,15] for the invariant VCS amplitudes and works
DR [15] 22.0 5.5
below pion threshold as well as in the rst resonance re- ( = 1.4 GeV) ( = 0.5 GeV)
gion. Assuming analyticity, crossing symmetry and an ap-
propriate high-energy behavior, unsubtracted dispersion
relations relate the real part of VCS amplitudes to an in- a dipole parametrization has also been proposed:
tegral over the virtual photon energy of a function of their
imaginary part. The imaginary part of a VCS amplitude (E E
N
)Q2 =0
is given by the sum of N intermediate states, computed E (Q2 ) E
N
(Q2 ) = (9)
(1 + Q /2 )2
2
from N N data (in the phenomenological MAID-
2000 analysis [25]), plus higher order contributions beyond The mass scale is the second free parameter of the DR
N . Moreover asymptotic contributions have also to be formalism which can be extracted from a t to the VCS
considered for two VCS amplitudes (F1 and F5 ) which data at dierent Q2 values.
cannot fulll unsubtracted dispersion relation framework. The evolution with the real photon momentum q of
The t-channel 0 exchange and the knowledge of the the MAMI VCS data obtained at Q2 = 0.33 GeV2 is rela-
F0 form factor x the asymptotic contribution to F5 tively sensitive to the choice of the free parameter values:
and determine completely the spin-dependent GPs. = 1.6 0.2 GeV and = 0.5 0.2 GeV. The cor-
The asymptotic contribution of the amplitude F1 related responding evolution with the real photon momentum q
to the polarizability P (11,11)0 or M (Q2 ) originates from is presented by the dotted line in g 7. This prediction is
the t-channel intermediate states. In a phenomenolog- rather close to the evolution of method 3 given by the in-
ical analysis, this continuum is parametrized through the terference between the complete BH+Born amplitude and
exchange of a scalar-isoscalar particle in the t-channel, the truncated Non-Born amplitude, except for close to
i.e. an eective -meson which gives rise to a diamag- 0 . This kinematical point is only sensitive to PLT and
netic contribution. The asymptotic part and the disper- this indicates dierent results for this observable in the
sive contributions beyond N are estimated using a dipole framework of these 2 methods.
parametrization of the dierence: Results: Table 4 presents the two structure functions
in the framework of the four methods and compares them
(M MN
)Q2 =0 to theoretical predictions presented in the introduction:
M (Q2 ) M
N
(Q2 ) = (8) the heavy-baryon chiral perturbation theory calculation
(1 + Q /2 )2
2
(HBChPT) [13], the linear sigma model (LSM) [12], the ef-
fective Lagrangian model (ELM) [11], two non-relativistic
The mass scale is a free parameter related to the dia- constituent quark models (NRQCM) [9,10] and the dis-
magnetism distribution inside the nucleon. It can be ex- persive relation approach (DR) [15]. The three errors for
tracted from a t to the VCS data at dierent Q2 values. the rst method are, respectively, statistical error on the
Though unsubtracted dispersion relation is valid for data, systematic error on photon angular distributions,
the amplitude F2 related to the polarizability P (01,01)0 and systematic error on the normalization. For the other
or E (Q2 ), it is particularly relevant to wonder about the methods only statistical errors are reported. The predic-
quality of the saturation of the subtracted dispersion in- tion for the DR model is given for 2 values of and
tegrals by N intermediate states only. For this purpose close to the values determined experimentally.
124 The European Physical Journal A
80 80
60 60
40 40
20 20
0 0
0 0.2 0.4 0 0.2 0.4
Q (GeV2)
2
Q2 (GeV2)
2 2
0 0
2 2
4 4
6 6
8 8
10 10
12 12
0 0.2 0.4 0 0.2 0.4
Q2 (GeV2) Q2 (GeV2)
Fig. 9. Comparison of the unpolarized structure functions de-
termined in the VCS MAMI experiment at Q2 = 0.33 GeV2
and in the RCS results [1] with the predictions of the DR for-
malism [15] (left panel) and of the O(p3 ) HBChPT [13] (right
panel). The upper panels give the result for PLL PTT / and
the lower panel for PLT . The contributions of the scalar GPs
are indicated by the dashed (or dotted) lines and the total con-
tributions of the scalar and spin-dependent GPs are indicated
by the solid (or dashed-dotted) lines. The DR prediction for
the scalar GP E (Q2 ) is calculated for = 1.4 GeV (upper
left panel). Two values of are used to calculated the scalar
GP M (Q2 ) contribution (lower left panel) in order to show
the sensitivity. The contributions for = 0.4 and 0.6 GeV
are used in the dotted and dashed lines, respectively. Figure
extracted from [15].
The rst experimental method for which the system-
atic errors have been carefully studied, gives the two
structure functions PLL PTT / and PLT presented in
ref. [18]. The three other methods conrm the large values
of the two structure functions. The 4th method in the DR
approach is particularly in good agreement with the 1st
method for the value of PLL PTT / and gives a slightly
smaller value for PLT . Only the heavy-baryon chiral per-
turbation theory calculation (HBChPT) [13] and the dis-
persive relation approach (DR) [15] predict large values
for these two structure functions and seem relevant for
the description of the MAMI VCS experiment (see g. 9).
The structure functions measured in an unpolarized
VCS experiment are the sum of two contributions (cf.
eq. (7)) : one scalar related to the electric E and mag-
netic M polarizabilities measured in RCS and one spin-
dependent. The last contribution vanishes at Q2 = 0. Fig-
ure 9 indicates that the eect of the spin GPs is much
smaller in the DR calculation [15] than in the O(p3 )
HBChPT [13]. To go beyond, that is to measure indepen-
dently all the scalar and spin-dependent polarizabilities, it
is necessary to perform a double polarization experiment.
2.3 The two other unpolarized experiments at JLab
and MIT-Bates
The JLab experiment E93-050 [19] has been performed
in the Hall A of the Thomas Jeerson National Accel-
erator Facility at Q2 = 0.9 and 1.8 GeV2 . The values of
 are 0.95 and 0.88, respectively. Data cover the region
below the pion threshold and the resonance region up to
s = 2 GeV. The experimental analysis of the complete
experiment is presented in ref. [19].
The Bates experiment 97-03 [20] has been performed at
Q2 = 0.05 GeV2 and  = 0.90. Measurements have been
done in-plane and at 90 out-of-plane, using the OOPS
spectrometers. The experiment covers a limited range in
polar angle around 90 , so the structure functions are
determined from the -dependence of the cross section.
Data analysis is still in progress and only preliminary re-
sults [20] can be presented. This experiment represents a
laboratory achievement, having made the rst use of the
high duty factor beam in the South Hall Ring and of the
full OOPS system.
10 4
2
5
0
0
-2
0 0.5 1 1.5 2 0 0.5 1 1.5 2
2 2
Q (GeV )
2 2
Q (GeV )
Fig. 10. Compilation of the data on electric E (Q2 ) (left)
and magnetic M (Q2 ) (right) GPs of the proton. Data points
at Q2 = 0 are from ref. [1]. The other points are the analyses of
MIT-Bates [20] (still preliminary), of MAMI [18] and dierent
analyses of JLab [19]. JLab points are slightly shifted in ab-
scissa for better visibility. The inner error bars are statistical;
the outer ones are the total error. The curves show calculations
in the DR model with the values of and obtained in each
experiment ( , ) = (0.70,0.63) GeV in the JLab experi-
ment (solid curve), (1.60,0.50) GeV in the MAMI experiment
(dashed curve) and (0.60,0.51) in the MIT-Bates experiment
(dotted curve). Note that the DR predictions for M (Q2 ) for
the MIT-Bates and MAMI experiments are on the same dotted
curve. Figure done thanks to Helene Fonvieille [26].
 N. dHose: Virtual Compton Scattering at MAMI
2 0.7
0 0.5
-2
-4
-6
-8
0 0.2 0.4
2 2
Q (GeV )
Fig. 11. PTT evaluated using the DR model [15] (solid
line), the O(p3 ) HBChPT [13] (dotted line) and the O(p4 )
HBChPT [16] (dashed line).
We present in g. 10 the world results for the elec-
tric E (Q2 ) and magnetic M (Q2 ) GPs deduced from the
MAMI, JLab and MIT-Bates experiments. The value of
the electric E (Q2 ) and magnetic M (Q2 ) GPs can de-
termined directly by the coe cients and obtained
in the DR analysis and using eq. (9) and eq. (8) or indi-
rectly by the structure functions PLL PTT / and PLT
determined in the LET analysis in which the spin GPs con-
tributions are evaluated in the DR model and subtracted
using eq. (7). The agreement between these 2 methods
was reasonably controlled in the JLab experiment (see the
dierent results at Q2 = 0.9 and 1.8 GeV). The direct de-
termination allows us to use also the VCS data in the res-
onance region (see in g. 10 the result at Q2 = 0.9 GeV2
with the smallest statistical error.)
The curves in g. 10 are calculated using the DR model
and the dierent values of and obtained in each ex-
periment. By denition all the DR predictions (see eqs. (9)
and (8)) are constrained to go through the experiment
RCS point at Q2 = 0. The fact that there is no unique
DR curve going through all the data points, especially for
the electric polarizability, does not invalidate the model. It
simply means that the dipole parametrization of eqs. (9)
and (8) does not hold over the entire Q2 range. Another
fact to be aware of is the model-dependency introduced
in this gure by transforming the structure functions into
GPs. The spin-dependent GPs are evaluated using the DR
model, and as it has been pointed in g. 9, this evaluation
is quite smaller than in the O(p3 ) HBChPT [13].
It is clear that measurements of individual scalar and
spin-dependent GPs are necessary to go further. We can
note in g. 11 the very dierent predictions for PTT using
the DR model [15] or the O(p3 ) HBChPT [13] or else
the O(p4 ) HBChPT [16]. An extraction of PTT can be
achieved in a further experiment at MAMI at the same
Q2 = 0.33 GeV2 , but with an other value of . We can
take the benet of the next 1.5 GeV energy of the beam
to access a new value of  to have a comfortable lever arm
for a longitudinal-transverse separation.
125
d 5 (nb/GeV sr 2)
0.9
0.8
0.6
0.4
0.3
0.2
0.05 0.1 0.15 0.2 0.25 0.3
,
q (GeV)
Fig. 12. VCS dierential cross section as a function of the real
photon energy q in the MAMI kinematics below and above
the pion threshold. The VCS MAMI data are reported. The
BH+Born contribution is given by the dash-dotted line. Pre-
dictions for the total cross section are given in the DR ap-
proach [15] using a xed value of = 1 GeV and for three
values of : 0.6 GeV (solid line); 0.7 GeV (dotted line) and
0.4 GeV (dashed line). Fig extracted from ref. [15].
3 Single polarized experiments above pion
threshold
Figure 12 shows the DR predictions for photon energies
ranging from threshold to the (1232)-resonance region.
The deviation from the BH+Born prediction rises strongly
after pion threshold. When crossing the pion threshold,
the VCS amplitude acquires an imaginary part due to
the coupling to the N channel. Therefore single polariza-
tion observables become non-zero above pion threshold. A
particularly relevant observable is the electron single spin
asymmetry (SSA) which is obtained by ipping the elec-
tron beam helicity. For VCS this observable is mainly due
to the interference of the real BH+VCS amplitude with
the imaginary part of the VCS amplitude. As the SSA van-
ishes in-plane, its measurement requires an out-of-plane
experiment. Such an experiment has been proposed at
MIT-Bates [27] and is being realized at MAMI [28] thanks
to one of the spectrometers of the A1 collaboration moving
out-of-plane. In g. 13, the SSA is presentedfor a kinemat-
ics in the (1232) region, corresponding to s = 1.2 GeV.
The DR calculation shows that the SSA is quite sizeable,
and it is mainly sensitive to the imaginary part of the VCS
amplitude, displaying only a rather weak dependence on
the GPs (obtained for the dierent values of and ).
Therefore it provides an excellent cross-check of the dis-
persive input (MAID 2000) in the DR formalism for VCS,
in particular by comparing at the same time the pion and
photon electroproduction channels through the excita-
tion. The MAMI analysis is still in progress.
4 Double polarized experiments at threshold
A double-polarization VCS experiment is also presently
being realized at MAMI. The theoretical framework of
126 The European Physical Journal A

0 eYcm
e
*
eXcm
0.05
e p
p eZcm

Fig. 14. Kinematics for the p(e, e p ) reaction.

0.1

0.15
0 10 20 30 40 50 60 70 80 90

Fig. 13. Electron single-spin asymmetry (SSA) for the VCS

MAMI kinematics as a function of the photon scattering an-
gle. The full dispersion results are shown for the values of
= 1 GeV and = 0.6 GeV (solid curve); = 1 GeV and
= 0.4 GeV (dashed curve); = 1 GeV and = 0.7 GeV
(dotted curve) and = 1.4 GeV and = 0.4 GeV (dash-
dotted curve). Figure extracted from ref. [15].
Fig. 15. Predictions for the deviation of the double polariza-
tion asymmetry from the BH+Born contribution using the DR
such experiment has been developed by Vanderhaeghen model [15] (solid line), the O(p3 ) HBChPT [13] (dotted line)
and Guichon [5,29]. Below pion production threshold, the and the O(p4 ) HBChPT [31] (dashed line).
VCS amplitude is purely real, and all single-polarization
observables are zero. So only double polarization exper-
and PLL , P
TT a z combination of PzLT and PLT , P TT a
iment observables can disentangle the GPs. For a polar- 
combination of P LT and PLT .
ized electron of pure helicity state h = 1/2 (longitudinal
polarization e = 2hk/me ), we can measure the average
polarization P of the recoil proton along the 3 vectors PLL = aP (01,01)0 , (13)
(excm , eycm , ezcm ) associated to the virtual photon direc- PTT = c1 P (11,11)1
+ c2 P (01,12)1
,
tion in the photon-proton CM as dened in g 14. PLT = bP (11,11)0
+ c3 [P (11,00)1
+ d1 P (11,02)1 ],
We can dene for the 3 axes:
PzLT = c4 P (11,11)1 + c3 [P (11,00)1 + d1 P (11,02)1 ],
P e(i) = (10) 
z
    PLT = c5 P (11,11)1 + c6 [P (11,00)1 + d1 P (11,02)1 ],
d e = me , p = e(i) d e = me , p = e(i)
5 2hk 5 2hk 

PLT = [d2 P (11,00)1 + d3 P (11,02)1 ].
   
me , p = e(i) + d e = me , p = e(i)
d5 e = 2hk 5 2 hk
The vij in eq. (12) are kinematical factors depending
5 on and . If the out-of-plane angle remains close to
d (h, i)
= 0 or 180 , we have the following approximations:
2 d5
We obtain a similar low energy prediction as in eq. (5): v1 = v1z v1x sin ; v2 v2z v2x constant;
d5 (h, i) = d5 BH+Born (h, i) (11) v3z v3x cos ; v4x sin 0; i, viy sin 0.
+ q 0 (h, i) + O(q2 ).
The GPs can be extracted from the linear system
So in such a complete experiment we can access 4 ob- above using the angular distributions of 0 and the 3

servables: 0 (h, i). Note that P LT can only be extracted by an
out-of-plane measurement. Theoretical predictions [31] us-
0 = v1 (PLL PTT /) + v2 PLT , (12) ing the DR model, the O(p3 ) HBChPT and the O(p4 )
z
0 (h, z) = (4h) [v1z PTT + v2z PzLT + v3z P LT ], HBChPT give a few % deviation of the double polarization
asymmetry from the BH+Born contribution (see g. 15).
0 (h, x) = (4h) [v1x P x x  x 
LT + v2 PTT + v3 P TT + v4 P LT ], Such a delicate experiment is being realized using the

0 (h, y) = (4h) [v1y P y y  y 
LT + v2 PTT + v3 P TT + v4 P LT ].
polarized electron beam available at MAMI and the mea-
surement of the recoil polarization of the outgoing pro-
Only 6 structure functions are independent: PLL , PTT , ton in a focal plane polarimeter [30], and the detection of
z
PLT , PzLT , P LT and P LT . P
LT is a combination of PTT the outgoing electron in the high resolution spectrometer
 N. dHose: Virtual Compton Scattering at MAMI
moving out-of-plane. It is clearly a very challenging exper-
iment, relying on a very delicate expertise of the complete
apparatus and requiring high statistics and very reduced
systematic errors.
5 Conclusion
An ambitious program to reach the generalized polariz-
abilies of the proton has been undertaken at MAMI over
the last ten years. The ultimate Grail is the separation
between spin-independent and spin-dependent GPs which
seems very promising for the study of the nucleon struc-
ture.
All the results reported here are the fruits of the
complete A1-VCS collaboration. I would like to acknowl-
edge all the students, Luca Doria, Peter Janssens, Imad
Bensafa, Jan Friedrich, Julie Roche, David Lhuillier, Do-
minique Marchand for which the work was essential to pro-
duce reliable results on these very meticulous experiments.
I wish to underline the strong support and the synergy
given by Helene Fonvieille, Harald Merkel, Michael Dis-
tler, Luc Van Hoorebeke, Gabriel Tamas, Robert Van de
Vyver, Jorg Friedrich, Thomas Walcher for this research.
I am also very grateful to Dieter Drechsel, Pierre Guichon,
Marc Vanderhaeghen, Barbara Pasquini, Stefan Scherer,
Thomas Hemmert, Ulf Meissner for their pedagogical lec-
tures and theoretical support in the data interpretation.
It is clear that the success of the VCS MAMI exper-
iments has its origin in the coherent eort between ex-
cellent physicists as Karl-Heinz Kaiser, always concerned
with the performance and the high quality of the electron
facility, Thomas Walcher strongly supporting and man-
aging all the eorts for such a challenging experimental
program, Jorg Friedrich, expert of delicate and precise ex-
periments and Dieter Drechsel, stimulating progress in our
scientic knowledge.
References
1. V. Olmos de Leon et al., Eur. Phys. J. A 10, 207 (2001).
2. Y.S. Tsai, Phys. Rev. 122, 1898 (1961).
3. H. Arenhovel, D. Drechsel, Nucl. Phys. A 233, 153 (1974).
4. P.A.M. Guichon, G.Q. Liu, A.W. Thomas, Nucl. Phys. A
591, 606 (1995).
5. P.A.M. Guichon, M. Vanderhaeghen, Prog. Part. Nucl.
Phys. 41, 125 (1998).
127
6. D. Drechsel, G. Knochlein, A. Metz, S. Scherer, Phys. Rev.
C 55, 424 (1997).
7. D. Drechsel, G. Knochlein, A. Yu Korchin, A. Metz, S.
Scherer, Phys. Rev. C 57, 941 (1998).
8. F.E. Low, Phys. Rev. 110, 974 (1958).
9. G.Q. Liu, A.W. Thomas, P.A.M. Guichon, Austral J. Phys.
49, 905 (1996).
10. B. Pasquini, S. Scherer, D. Drechsel, Phys. Rev. C 63,
025205 (2001).
11. M. Vanderhaeghen, Phys. Lett. B 368, 13 (1996).
12. A. Metz, D. Drechsel, Z. Phys. A 356, 351 (1996); 359,
165 (1997).
13. T.R. Hemmert, B.R. Holstein, G. Knochlein, S. Scherer,
Phys. Rev. D 55, 2630 (1997); Phys. Rev. Lett. 79, 22
(1997); T.R. Hemmert, B.R. Holstein, G. Knochlein, D.
Drechsel, Phys. Rev. D 62, 014013 (2000).
14. B. Pasquini, M. Gorchtein, A. Metz, M. Vanderhaeghen,
Eur. Phys. J. A 11, 185 (2001).
15. D. Drechsel, B. Pasquini, M. Vanderhaeghen, Phys. Rep.
378, 99 (2003).
16. C.W. Kao, M. Vanderhaeghen, Phys. Rev. Lett. 89, 272002
(2002).
17. V. Bernard, N. Kaiser, A. Schmidt, U. Meissner, Phys.
Lett. B 319, 269 (1993); Z. Phys. A 348, 317 (1994).
18. J. Roche et al., Phys. Rev. Lett. 85, 708 (2000).
19. G. Laveissiere et al., Phys. Rev. Lett. 93, 122001 (2004).
20. J. Shaw, R. Miskimen, MIT-Bates Proposal 97-03, (1997)
and P. Bourgeois, PhD Thesis.
21. K.I. Blomqvist et al., Nucl. Instrum. Methods A 403, 263
(1998).
22. G. Hohler, E. Pietarinen, I. Sabba-Stefanescu, F.
Borkowski, G.G. Simon, V.H. Walther, R.D. Wendling,
Nucl. Phys. B 114, 505 (1976); private communication.
23. M. Vanderhaeghen, J.M. Friedrich, D. Lhuillier, D. Marc-
hand, L. Van Hoorebeke, J. Van de Wiele, Phys. Rev. C
62, 025501 (2000).
24. P. Janssens, L. Van Hoorebeke et al., to be published in
Nucl. Instrum. Methods.
25. D. Drechsel, O. Hanstein, S.S. Kamalov, L. Tiator, Nucl.
Phys. A 645, 145 (1999).
26. H. Fonvieille, Proceedings of the Erice School of Nuclear
Physics, 26th Course, Prog. Part. Nucl. Phys. 55, 198
(2005) and private communication.
27. N.I. Kaloskamis, C.N. Papanicolas, MIT-Bates proposal
(1997).
28. N. dHose, H. Merkel, MAMI Proposal (2001).
29. M. Vanderhaeghen, Phys. Lett. B 402, 243 (1997).
30. Th. Pospischil et al., Nucl. Instrum. Methods A 483, 726
(2002).
31. C.W. Kao, B. Pasquini, M. Vanderhaeghen, Phys. Rev. D
70, 114004 (2004).
Eur. Phys. J. A 28, s01, 129 137 (2006)
DOI: 10.1140/epja/i2006-09-014-5 EPJ A direct
electronic only

Experimental tests of Chiral Perturbation Theory

H. Merkela
Institut fur Kernphysik, Johannes Gutenberg-Universitat Mainz, D-55099 Mainz, Germany

/
Published online: 15 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. Over the last decade, a series of dedicated experiments to test heavy baryon chiral perturbation
theory was performed at MAMI. Photo production of neutral pions close to threshold with unpolarized and
polarized photon beam was performed to separate the multipole amplitudes at threshold. The extension of
this experiments to a modest photon virtuality of Q2 < 0.1 GeV2 /c2 was performed to extract additional
longitudinal multipoles and to exploit the Q2 evolution predicted by theory. An out-of-plane measurement
above + threshold with polarized electron beam gave access to the imaginary part of the s-wave amplitude.
Finally, by coherent photo and electro production from the deuteron the neutron amplitude could be
extracted.
PACS. 25.30.Rw Electroproduction reactions 13.60.Le Meson production 12.39.Fe Chiral Lagrangians

1 Introduction 2 Neutral pion photo production

Since chiral perturbation theory is an expansion in small
Chiral perturbation theory is an eective eld theory, momenta, only for a small region above threshold predic-
which utilizes the symmetry properties of the QCD La- tions can be derived. Close to threshold, the angular struc-
grangian to extract observables at low momenta in a sys- ture of the dierential cross section of photo production
tematic fashion (see, e.g., S. Scherer, this issue). of pseudoscalar mesons can be expanded in
Meanwhile, a vast amount of observables can be calcu-
q
lated in the framework of chiral perturbation theory, lead- () = A + B cos + C cos2 ,
ing to numerous ways to test this theory in experiment. k
This contribution, however, concentrates on a dedicated with q and k the pion and photon center of mass momen-
series of experiments, which were especially designed to tum and the pion center of mass production angle.
test the predictions of Heavy-Baryon Chiral Perturbation The angular coe cients can be further decomposed in
Theory (HBChPT). s- and p-wave multipoles:
The investigated reaction is the electromagnetic pro-
duction of the Goldstone Boson of the theory, the pion. 1 2
2
A = E0+ + P2 + P 3 2 ,
The choice of the electromagnetic probe ensures the pre- 2
cision, which is necessary to extract amplitudes very close B = 2 Re (E0+ P1 ) ,
to threshold. While the + production is dominated by 1 2
the charge (Kroll-Ruderman term), the 0 production is C = P1 2 P2 + P 3 2 .
2
the ideal testing ground for HBChPT.
Over the last decade, several experiments on 0 In this eld, it is common to use the p-wave combinations
threshold production were performed at MAMI and at P1 = 3E1+ + M1+ M1 ,
other electron accelerator laboratories. The formalism of
HBChPT was developed in parallel with large success in P2 = 3E1+ M1+ + M1 ,
predicting or tting of the measured threshold observ- P3 = 2M1+ + M1 .
ables, leading to an improving consistent picture. In this
article, an overview over the existing experiments and Not only the angular form of the multipoles is known,
their interpretation in terms of HBChPT is given. but also their energy dependence. The p-waves rise pro-
portional to photon and pion CMS momentum q k, while
the form of the s-wave multipole is dictated by unitary, as
will be discussed later.
The rst experiments on threshold 0 photo produc-
a
e-mail: merkel@kph.uni-mainz.de tion [1,2] were designed to test the Low-Energy Theorem
130 The European Physical Journal A

4.5 A. Schmidt et al.

series [5]. These authors calculated s- and p-waves and
M. Fuchs et al. showed, that while E0+ and P3 are slow converging, the
4.0 P1 and P2 multipoles were fast converging and should be
J. C. Bergstrom et al.
3.5 a strong prediction of HBChPT.
Figure 1 shows the total cross section of the last
3.0 MAMI experiment [3] in comparison with former data
/ b

2.5 from SAL [6] and MAMI [7]. The data were taken at
the tagged photon beam of the MAMI A2 Collaboration
2.0 (g. 2). The reaction was identied by the detection of
1.5 the two decay photons of the pion in coincidence with the
TAPS detector, an array of 504 BaF detector modules ar-
1.0 ranged in 7 blocks.
0.5 With this setup, dierential cross sections were mea-
sured up to an incoming photon laboratory energy of
0.0
144 147 150 153 156 159 162 165 168 168 MeV. By tting the angular coe cients A, B, and C to
E / MeV the dierential cross section and extrapolating to thresh-
old by using the known energy dependence, the multipole
Fig. 1. Total cross section of neutral pion photo production combinations Re E0+ , P1 , and P22 +P32 could be extracted.
(from [3]).

BaF2-modul 25 cm (12 X0) 2.1 Polarized photon asymmetry

5 cm To further decompose P22 + P32 , an additional polariza-

tion observable has to be measured. The polarized photon
asymmetry is dened as the asymmetry of the cross
section in respect to the polarization plane of a polarized
photon beam, i.e.

= 0 (1 P () cos 2)
e am
with the angle between polarization plane and produc-
B
tion plane and P the photon polarization. This asymme-
try is proportional to the dierence P22 P32 and allows
the decomposition of the modulus of these two multipole
amplitudes.
At MAMI, a polarized photon beam was prepared by
50cm
coherent Bremsstrahlung from a diamond crystal. The
LH 2Target
asymmetry was again determined by detecting the decay
photons of the pion with the TAPS detector. Figure 3

0.4

0.3

0.2
504 BaF 2 Detector Modules 0.1

Fig. 2. One BaF module and the complete setup of the photon
0.0
spectrometer TAPS at the MAMI A2-Collaboration.
-0.1
(LET) [4], which derived from general principles a value ChPT
-0.2 DR
for the s-wave amplitude E0+ at threshold.
t to the data
These rst experiments showed a serious discrepancy -0.3
from the prediction of the low-energy theorem. The de- 0 20 40 60 80 100 120 140 160 180
velopment of the formalism of Heavy-Baryon Perturba- cms
0 / deg
tion Theory resolved this puzzle by showing, that the
LET value is only the leading term in a slow converging Fig. 3. The polarized photon asymmetry (from [3]).
 H. Merkel: Experimental tests of Chiral Perturbation Theory 131

Table 1. Multipole amplitudes extracted in photoproduction. 0.0

: q k-t ChPT
E0+ P1 P2 P3 -0.2 : q-t DR
(10 3 /m ) (q k 10 3 /m3 )
-0.4

+
MAMI 1.31 0.08 10.02 0.2 10.5 0.2 13.1 0.1

ReE0+ / 10 /m
SAL 1.32 0.05 10.26 0.1

-3
ChPT 1.16 10.33 0.6 11.0 0.6 11.7 0.6 -0.6
DR 1.22 10.54 11.4 10.2
-0.8
0
-1.0
+
+
-1.2 n
Ethr =151.4 MeV
-1.4
145 150 155 160 165
p n p E / MeV
Fig. 4. Rescattering graph. Fig. 5. Real part of the s-wave multipole E0+ (see [3] for
details).

shows the measured asymmetry, averaged over the energy

range.
Combining the unpolarized and polarized measure-
ment, the four threshold amplitudes can be extracted. Ta-
ble 1 summarizes these amplitudes. For E0+ and P1 , the
values can be compared with the data from an experiment
at SAL [6], which are in agreement within the error bar.
The quoted values of HBChPT, which are based on a ret
of the low-energy constants [8] also successfully describe
the data. For comparison, also the results of a dispersion
relations calculation [9] is added. While the values for P2
and P3 are reasonable in the table, in g. 3 the remain-
ing small discrepancy is more pronounced due to the large
sensitivity of to the dierence of these two amplitudes.
2.2 Cusp eect of the s-wave amplitude
While the p-wave amplitudes rise proportional to the mo-
menta q k the energy dependence of the s-wave amplitude
is given by the unitarity of the scattering matrix.
Chiral perturbation theory alone predicts a roughly
constant s-wave amplitude. The 0 p amplitude, however,
is an order of magnitude smaller than the + n amplitude
due to the sizable Kroll-Ruderman term. Above the + n
threshold, a photon coupling to a + in the intermediate
state with a following pion charge exchange in the nal
state (see g. 4) leads to a large contribution to the s-
wave amplitude.
Figure 5 shows the real part of the E0+ amplitude in
photo production extracted from the MAMI data [3]. The
two dierent data sets are extracted from the same dier-
ential cross section by the two assumptions of the p-waves
rising proportional to q (open squares) or proportional to
q k (lled squared). The rst assumption is phenomeno-
logical valid over a larger energy range, while the second
assumption corresponds to the prediction of chiral pertur-
bation theory.
The cusp eect at the + n threshold is clearly seen.
Again, the dispersion relations calculation [9] is included
(dashed line). The strength of this eect is roughly given
by the E0+ amplitude of the + n production times the
charge exchange scattering length.
Only the real part of E0+ could be extracted by this
experiment. For the imaginary part, an additional polar-
ization experiment would be necessary, e.g., a measure-
ment of the polarized target asymmetry.
3 Electroproduction
While in photo production two amplitudes are given by
a t of low-energy constants to the data, the extension
to electro production exploits additional predictive power
of HBChPT. In addition to the dependence of the mul-
tipoles on the photon virtuality Q2 the two longitudinal
p-wave multipoles P4 and P5 and the longitudinal s-wave
multipole L0+ occur.
The ve-fold dierential cross section is given by
d(, ) !
= T +  L +  T T cos 2
d  dE  d 
+ 2(1 + ) T L cos
 "
+ h 2(1 ) T L sin
with the virtual photon ux , the transverse photon po-
larization  and the photon-proton center of mass angles
and dened as in photoproduction with respect to the
virtual photon direction.
The transverse cross section T and the longitudinal
cross section L can be disentangled by varying , for the
other structure functions the dierential cross section has
to be measured over the corresponding angular range. The
transverse-longitudinal interference T L can be extracted
by a polarized electron beam with helicity h.
132 The European Physical Journal A

Fig. 6. The three-spectrometer setup of the A1 Collaboration at MAMI.

In the threshold region, the cross section can be further A = |S0+ |2 + |P5 |2 ,
decomposed into s- and p-wave multipoles: B  = 2 Re (S0+ P4 ) ,
 
T () = p/k A + B cos + C cos2 , C  = |P4 |2 |P5 |2 ,
 

L () = p/k A + B  cos + C  cos2 , G = Im E0+ P5 + S0+ P2 ,
T L () = p/k (D sin + E sin cos ) , H = Im (P1 P5 + P4 P2 ) .

T T () = p/k F sin2 ,
# 3.1 Unpolarized experiments
Q2
T L () = p/k (G sin + H sin cos ) ,
2 A rst electroproduction experiment at NIKHEF [10] es-
timated the s-wave cross section at threshold, a second
where p/k is the phase space ratio of pion CM momentum
experiment [11] tried to extract p-wave amplitudes in ad-
and photon CM equivalent momentum, and the angular
dition by measuring the pion emission angle in plane left
coe cients are combinations of two s-wave and ve p-wave
and right from the virtual photon direction. While these
multipoles:
pioneering works were in good agreement with ChPT, the
1 predictive power was not full exploited, since the data has
A = |E0+ |2 + |P2 |2 + |P3 |2 ,
2 to be analyzed with assumptions from theory.
B = 2 Re (E0+ P1 ) , A more complete experiment was performed at
1 MAMI [12] at the three-spectrometer setup of the A1
C = |P1 |2 |P2 |2 + |P3 |2 , collaboration. In electro production, the recoil proton is
2
detected in coincidence with the scattered electron. Very
D = Re (E0+ P5 + L0+ P2 ) , close to threshold, the boost by the Lorentz transforma-
E = Re (P1 P5 + P4 P2 ) , tion from the center-of-mass system to the laboratory sys-
1 tem focuses the full solid angle of the center of mass system
F = |P2 |2 |P3 |2 , into a narrow cone.
2
 H. Merkel: Experimental tests of Chiral Perturbation Theory 133

1.0 2.0
E0+ / 10 /m

4 W = 0.5 MeV W = 1.5 MeV

-3

1.5

tot [b/sr]
0.5 1.0

2 0.5

0.0 0.0
0.00 0.05 0.10 0.00 0.05 0.10
3.0 4.0
W = 2.5 MeV W = 3.5 MeV
0 2.5
0 5 10 3.0

tot [b/sr]
W / MeV 2.0
0
L0+ / 10 /m

1.5 2.0
-3

1.0
1.0
0.5
0.0 0.0
0.00 0.05 0.10 0.00 0.05 0.10
-1
Q2 [GeV2/c2] Q2 [GeV2/c2]
Fig. 8. The total cross section [14] versus photon virtuality
(only MAMI data). The solid line shows the calculation in
HBChPT [13].

-2
0 5 10 Figure 8 summarizes the result of this experiment in
W / MeV comparison with the real photon data and the data at
Fig. 7. The extracted s-wave multipoles of the MAMI data [12] Q2 = 0.1 GeV2 /c2 . While the photon point and the higher
(solid squares) in comparison with the NIKHEF data [11] (open Q2 point was included in the t of HBChPT, the middle
circles) and a calculation in HBChPT [13] (line). Q2 point was not. As can be seen, there was a clear dis-
crepancy which has to be claried.

At MAMI, two high-resolution spectrometer of the A1

Collaboration were used for the detection of the scattered
electron and the recoil proton (g. 6). At a photon vir-
tuality of Q2 = 0.1 GeV2 /c2 up to a center-of-mass en-
ergy of 4 MeV above threshold full coverage of the angular
range was achieved. To separate transverse and longitudi-
nal cross section, data were taken at three values of the
photon polarization .
For each bin in  and energy the dierential cross sec-
tion was tted to separate the structure functions T and
L and the interference structure functions T T and T L .
Similar to photo production, the angular structure of the
structure functions was tted with the assumption of only
s- and p-waves contributing to the cross section.
Figure 7 shows the result for the s-wave multipoles in
comparison with the results from NIKHEF and with the
calculations in HBChPT [13].
As can be seen, the extracted multipoles are in agree-
ment within the error bars and can be tted by HBChPT.
Actually, this result was surprising, since at that point a
photon virtuality of Q2 = 0.1 GeV2 /c2 was expected to be
somewhat beyond the scope of HBChPT.
Therefore, a further experiment was performed at a
lower virtuality of Q2 = 0.05 GeV2 /c2 (ref. [14]). The same
experimental technique was used as by the previous exper-
iment, however the systematic errors are somewhat larger
due to the lower outgoing proton momentum.
3.2 Extended momentum range
The rst step was, to extend the momentum range to
be more sensitive to the interference structure functions.
At MAMI an extended experiment was performed at a
four-momentum transfer of Q2 = 0.05 GeV2 /c2 [15]. Due
to the extension of the momentum range to a center-
of-mass energy of 40 MeV above threshold the full an-
gular range cannot be measured in a single kinemati-
cal setting. Several settings along the in plane angular
range ( = 0 , 180 ) are required to separate the dier-
ential cross section 0 = T + L and the longitudinal-
transverse interference structure function LT .
To separate the transverse-transverse interference T T ,
which has a dependence on the out-of-plane angle of
cos 2, the according out-of-plane acceptance is required.
At MAMI, one of the high-resolution spectrometer (spec-
trometer B in g. 6) can be tilted by up to 10 . By us-
ing this spectrometer as electron detector in extreme for-
ward direction, this translates to 90 in the center-of-mass
frame.
Data were taken at 14 dierent kinematical settings
and were combined to extract the structure functions.
Figure 9 shows the structure functions versus the center-
of-mass energy above threshold. For comparison, the cal-
culation in HBChPT [13] was included (solid line). As
134 The European Physical Journal A

2.5 0.00
ChPT
Maid
2.0 DMT
0.10

LT [ub/sr]
0 [ub/sr]

1.5
0.20
1.0

0.30 ChPT
0.5 Maid
DMT
0.0 0.40
0 10 20 30 40 0 10 20 30 40
W [MeV] W [MeV]
0.40 0.5
ChPT
0.30 Maid
DMT 0.0
LT [ub/sr]

0.20
TT [ub/sr]

0.10 0.5

0.00
1.0 ChPT
0.10 Maid
DMT
0.20 1.5
0 10 20 30 40 0 10 20 30 40
W [MeV] W [MeV]
Fig. 9. Separated structure functions [15]. The solid line is the calculation in HBChPT [13], the dashed line the Maid model [16],
and the dash-dotted line the DMT model [17].

stated before, this calculation overestimates the dieren-
tial cross section. Even more striking, the derivation of the
transverse-transverse interference is considerable, which is
proportional to the dierence of the p-wave multipoles P2
and P3 , which are reproduced in good quality in photo-
production.
For comparison, the phenomenological model
MAID [16] is included (dashed line). For the struc-
ture functions, this model is able to describe the data sets
roughly. This model is basically a global t to the existing
data sets in photo- and electro-production and can be
seen as check of the consistency of the data with the other
existing data sets, however data at this photon virtuality
and energy are scarth. The dynamical Dubna-Mainz-
Taipeh model (DMT) [17] is shown as dashed-dotted line
and is in good agreement with the data.
3.3 Helicity asymmetry
tional to the imaginary part of L0+ multiplied by the large
(1232) multipole M1+ . Since it is very small, one has to
dene the helicity asymmetry
(h = 1) (h = 1)
AT L ( = 90 , = 90 ) =
(h = 1) + (h = 1)

2(1 )T L ()
=
T () + L () T T ()
to reduce the systematic errors. The expected energy
structure is given by the unitary cusp at the opening of
the + threshold.
Figure 10 shows this asymmetry. Again, the calcula-
tions in HBChPT and the models MAID and DMT are
included in the graph. This small asymmetry enhances
the dierences between the models, and as can be seen
clearly, only the DMT model is able to describe the data.

By using polarized electrons, the fth structure function 4 Coherent production from the deuteron
LT  can be extracted. This structure function has the
multipole contents T L (90 ) Im[L0+ (3 E1+ M1+ + The 0 production on the neutron is in principle a strong
M1 ) E0+ (2 L1+ L1 )], i.e. it is basically propor- prediction of HBChPT, since by the 0 production from
 H. Merkel: Experimental tests of Chiral Perturbation Theory 135

5.0% 4.2 Coherent electroproduction

DMT
4.0% Maid In contrast to the technique of coherent photo production,
ChPT in electro production the recoil deuteron is detected. This
3.0% leads to clear and unique identication of the reaction
channel by the missing mass of the pion. Without this
ALT

2.0% subtraction, also the extraction of p-waves is possible.

1.0%
0.0%
1.0%
0 10 20 30 40
W [MeV]
Fig. 10. The asymmetry ALT  versus center-of-mass energy
above threshold W . The solid line is the calculation in
HBChPT [13], the dashed line the MAID model [16] and the
dashed-dotted line the DMT model [17].
the proton nearly all low-energy constants are already
xed. Obvious, the lack of a free neutron target compli-
cates the situation.
The use of the deuteron as neutron target requires de-
tailed modeling of the deuteron structure as rst step.
Since the Fermi momenta of the nucleons in the deuteron
is of the same order as the momenta of a threshold ex-
periment, the coherent production is the most promising
channel to get the deuteron structure and nal state in-
teraction under control.
By this choice, small cross sections are to be expected
as pointed out rst in ref. [18]. Due to the restrictions
for the intermediate state of two neutrons in an s-wave
due to Pauli blocking an exact cancelation between the
+ n intermediate state in the s-wave for production from
While the theoretical advantage is obvious, the ex-
perimental challenge is large. In a threshold experiment
at a four momentum transfer of Q2 = 0.1 GeV2 /c2 the
deuteron has a kinetic energy of T 30 MeV, which makes
it di cult to achieve the necessary accuracy to identify the
reaction.
At MAMI, the spectrometer setup is optimized to de-
tect low energetic particles. By using a thin LD2 target
with 3 mm path length in the deuterium and connecting
the spectrometer vacuum system with the vacuum sys-
tem of the target chamber, the multiple scattering and
energy loss of the deuterons could be minimized. Still, the
detection e ciency for deuterons is close to 80% (in com-
parison to 98% for protons) due to hadronic interactions
of the deuteron.
Figure 11 shows the achieved missing mass resolution
of the MAMI experiment [21]. Again, a full transverse-
longitudinal separation was performed by measuring a
three dierent values of the photon polarization  and
the full center-of-mass angular range was covered up to
a center-of-mass energy of 4 MeV above threshold.
Since at that time only predictions for the s-wave am-
plitudes were available [22], a t of the dierential cross
1500
1000
Events

the proton and the rescattering graph by charge exchange

on the spectator neutron occurs. This cancellation causes
500
small s-wave cross sections on the one hand, but reduces
the systematic uncertainties by the estimation of the nal
state interaction on the other hand.
0
100 110 120 130 140 150 160
2
Missing Mass [MeV/c ]
4.1 Coherent photoproduction
2
135.4 MeV/c
A rst photoproduction experiment was performed at
SAL [19]. The decay photons of the pion were detected 1000
by the IGLOO detector, a large solid angle array of scin-
Events

tillators. By this technique, the coherent production from

the deuteron cannot be separated from the break up chan- 500 2.28 MeV/c
2

nel with proton and neutron in the nal state. The con-
tribution of this channel was calculated in a model and
subtracted from the data. The dierential cross section
was extrapolated to the threshold to extract the thresh- 0
old s-wave amplitude. 100 110 120 130 140 150 160
2
A threshold value of Ed = (1.45 0.09)103 /m was Missing Mass [MeV/c ]
extracted. This value agreed within the error bar with a
Fig. 11. Missing mass resolution for the reconstruction of the
calculation in HBChPT [20], which gave a prediction for neutral pion [21].
the coherent s-wave amplitude only.
136 The European Physical Journal A

=0.364, W=0.5 MeV =0.590, W=0.5 MeV =0.854, W=0.5 MeV

20 20 40
d/d [nb/sr]

d/d [nb/sr]

d/d [nb/sr]
fit 1
15 fit 2 15 30
10 10 20
5 5 10
0 0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
cos() cos() cos()

=0.364, W=1.5 MeV =0.590, W=1.5 MeV =0.854, W=1.5 MeV

40 40 40
d/d [nb/sr]

d/d [nb/sr]

d/d [nb/sr]
30 30 30
20 20 20
10 10 10
0 0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
cos() cos() cos()

=0.364, W=2.5 MeV =0.590, W=2.5 MeV =0.854, W=2.5 MeV

60 60 60
d/d [nb/sr]

d/d [nb/sr]

d/d [nb/sr]

45 45 45
30 30 30
15 15 15
0 0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
cos() cos() cos()

=0.364, W=3.5 MeV =0.590, W=3.5 MeV =0.854, W=3.5 MeV

80 80
d/d [nb/sr]

d/d [nb/sr]

d/d [nb/sr]

80
60 60 60
40 40 40
20 20 20
0 0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
cos() cos() cos()

Fig. 12. The MAMI data on coherent electro production from deuteron [21] for three dierent values of the photon polarization
 in comparison with the calculations in HBChPT [23]. The upper curve shows the t with xed Ld , the lower curve shows the
t with two free low-energy constants. The dashed lines show the previous results from ref. [24].

section with the usual assumptions of s- and p-waves with were extracted to
the known energy dependence was performed to extract
the threshold s-wave amplitudes. It has to be stressed, |Ed | 0.42 103 /m ,
that this t ignores completely the deuteron structure and |Ld | = (0.50 0.11) 103 /m .
nal state interaction, leading to a large systematic errors
in the extracted amplitudes. Since Ld is very large, for Ed The transverse multipole Ed was within agreement
only an upper limit could be extracted. The amplitudes with the calculations as at the photon point, while the
 H. Merkel: Experimental tests of Chiral Perturbation Theory
calculations overestimated the longitudinal multipole by
a factor of 2.
The situation improved a lot by a new calculation by
the same group [23], which now used a complete multipole
decomposition and extracted s- and p-waves. By this, the
comparison has not to be done on the level of the ques-
tionable extracted s-wave amplitudes, but can be done
directly with the dierential cross section.
Figure 12 shows the result of their calculation in com-
parison with the MAMI data. The plot shows two kind of
ts: the rst was performed by xing the Ld multipole to
the extracted value of the data set, reducing the number
of parameters to a single low-energy constant. This t,
however, does not describe the data set completely. The
second t was performed with two free parameters and
describes the data.
5 Summary and outlook
While the data basis on pion threshold production is now
quite large, there are still a number of unsolved prob-
lems. The strong Q2 -dependence of the total cross sec-
tion (g. 8) seems to be unlikely and might indicate the
large systematic errors of the data sets as quoted by the
authors.
To resolve this puzzle, a dedicated experiment was per-
formed by MAMI, where three dierent Q2 values were
measured within one experiment and with emphasis on
reducing the systematic error [25]. This experiment is un-
der analysis, but preliminary data seems to indicate, that
the Q2 = 0.1 GeV2 /c2 data set is somewhat high. Since
that data set is included in the t of HBChPT, a ret of
the theory seems to be necessary as soon as the new data
set is published.
All cited calculations are based on the heavy-baryon
formalism. Meanwhile, better regularization schemes are
on the market and an improvement of the calculations can
be expected. Especially in electroproduction, where the p-
waves are not yet calculated to the same order as in photo
production, a considerable improvement is possible.
137
The success of ChPT in the SU (2) sector encourages
one to extend this program to the strangeness threshold.
At MAMI a new kaon spectrometer is under construction,
and kaon threshold production will be among the rst
experiments with this spectrometer.
References
1. E. Mazzucato et al., Phys. Rev. Lett. 57, 3144 (1986).
2. R. Beck et al., Phys. Rev. Lett. 65, 1841 (1990).
3. A. Schmidt et al., Phys. Rev. Lett. 87, 232501 (2001).
4. P. de Baenst, Nucl. Phys. B 24, 633 (1970); I.A. Vain-
shtein, V.I. Zakharov, Nucl. Phys. B 36, 589 (1972).
5. V. Bernard, N. Kaiser, J. Gasser, U.-G. Mei ner, Phys.
Lett. B 268, 291 (1991); V. Bernard, N. Kaiser, U.-G.
Mei ner, Z. Phys. C 70, 483 (1996).
6. J.C. Bergstrom et al., Phys. Rev. C 53, R1052 (1996).
7. M. Fuchs et al., Phys. Lett. B 368, 20 (1996); A.M. Bern-
stein et al., Phys. Rev. C 55, 1509 (1997).
8. V. Bernard et al., Eur. Phys. J. A 11, 209 (2001).
9. O. Hanstein et al., Phys. Lett. B 399, 13 (1997).
10. T.P. Welch et al., Phys. Rev. Lett. 69, 2761 (1992).
11. H.B. van den Brink et al., Phys. Rev. Lett. 74, 3561 (1995).
12. M.O. Distler et al., Phys. Rev. Lett. 80, 2294 (1998).
13. V. Bernard, N. Kaiser, U.-G. Mei ner, Nucl. Phys. A 607,
379 (1996); 633, 695 (1998)(E).
14. H. Merkel et al., Phys. Rev. Lett. 88, 012301 (2002).
15. M. Weis, Doctorate Thesis, Mainz, 2003.
16. D. Drechsel et al., Nucl. Phys. A 645, 145 (1999); S.S.
Kamalov et al., Phys. Lett. B 522, 27 (2001).
17. S.S. Kamalov et al., Phys. Rev. Lett. 83, 4494 (1999);
Phys. Rev. C 64, 032201 (2001).
18. M. Rekalo, E. Tomasi-Gustafsson, Phys. Rev. C 66, 015203
(2002).
19. J.C. Bergstrom et al., Phys. Rev. C 57, 3203 (1998).
20. S.R. Beane et al., Nucl. Phys. A 618, 381 (1997).
21. I. Ewald et al., Phys. Lett. B 499, 238-244 (2001).
22. V. Bernard, H. Krebs, U.-G. Mei ner, Phys. Rev. C 61,
58201 (2000).
23. H. Krebs, V. Bernard, U.-G. Mei ner, Eur. Phys. J. A 22,
503-514 (2004).
24. H. Krebs, V. Bernard, U.-G. Mei ner, Nucl. Phys. A 713,
405 (2003).
25. J. Garc a Llongo, Diploma Thesis, Mainz, in preparation.
Eur. Phys. J. A 28, s01, 139 148 (2006)
DOI: 10.1140/epja/i2006-09-015-4 EPJ A direct
electronic only

The Bonn Electron Stretcher Accelerator ELSA: Past and future

W. Hillerta
Physikalisches Institut, Universitat Bonn, Nussallee 12, 53115 Bonn, Germany

/
Published online: 16 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. In 1953, it was decided to build a 500 MeV electron synchrotron in Bonn. It came into op-
eration 1958, being the rst alternating gradient synchrotron in Europe. After ve years of performing
photoproduction experiments at this accelerator, a larger 2.5 GeV electron synchrotron was built and set
into operation in 1967. Both synchrotrons were running for particle physics experiments, until from 1982
to 1987 a third accelerator, the electron stretcher ring ELSA, was constructed and set up in a separate
ring tunnel below the physics institute. ELSA came into operation in 1987, using the pulsed 2.5 GeV
synchrotron as pre-accelerator. ELSA serves either as storage ring producing synchrotron radiation, or as
post-accelerator and pulse stretcher. Applying a slow extraction close to a third integer resonance, external
electron beams with energies up to 3.5 GeV and high duty factors are delivered to hadron physics experi-
ments. Various photo- and electroproduction experiments, utilising the experimental set-ups PHOENICS,
ELAN, SAPHIR, GDH and Crystal Barrel have been carried out. During the late 90s, a pulsed GaAs
source of polarised electrons was constructed and set up at the accelerator. ELSA was upgraded in order to
accelerate polarised electrons, compensating for depolarising resonances by applying the methods of fast
tune jumping and harmonic closed orbit correction. With the experimental investigation of the GDH sum
rule, the rst experiment requiring a polarised beam and a polarised target was successfully performed at
the accelerator. In the near future, the stretcher ring will be further upgraded to increase polarisation and
current of the external electron beams. In addition, the aspects of an increase of the maximum energy to
5 GeV using superconducting resonators will be investigated.

PACS. 29.20.-c Cyclic accelerators and storage rings 29.20.Lq Synchrotrons 29.27.Hj Polarised beams

1 The beginning 2 The 500 MeV synchrotron

Accelerator physics started at Bonn in 1952, when Wolf-
gang Paul, who received the Nobel price 1989 for his work
on ion traps, accepted a call to Bonn and became a full
professor at the physics institute of Bonn university. Paul,
who had worked at the university of Gottingen on a 6 MeV
betatron together with his teacher Hans Kopfermann, was
deeply impressed by the recently discovered principle of
strong focusing in particle accelerators, applying trans-
verse magnetic elds with strong alternating eld gradi-
ent [1,2,3]. He was thinking about building a strong fo-
cusing 100 MeV electron synchrotron at Bonn and made
an application to the Deutsche Forschungsgemeinschaft
(DFG). Although this application covered about 10 % of
the total annual funding of the DFG, it was accepted, and
Paul was even encouraged to increase the energy of the
planned machine to 500 MeV.
a
Since 2001 in charge of the ELSA accelerator; e-mail:
hillert@physik.uni-bonn.de
In November 1953, Paul started, together with a small
group of scientic assistants, PhD and diploma students,
to work on the design and construction of the accelerator.
At this time, there was no experience in building a strong
focusing accelerator, and a lot of problems had to be over-
come. Since at that time no electronic computer was avail-
able, a pendulum as a mechanical analogue was built in
order to solve the dierential equations describing the par-
ticles motion. A eld gradient of 10 T/m at the maximum
energy of 500 MeV was proposed, actually the highest gra-
dient ever used for a combined function synchrotron. After
construction had started in 1954 and all of the nine mag-
nets of type 1/2D-F -1/2D had already been installed, it
turned out that the accelerator would operate just on top
of a non-linear stop band, which had been discovered 1956
at CERN, and would never work. This problem could be
solved by enlarging the gap between the defocusing and
focusing sectors of the magnets, using 1 cm thick pieces of
plywood. The complete radio frequency system, consisting
of the transmitter, the ampliers and the six accelerating
cavities, was built in the framework of a diploma thesis [4].
Pure ceramics was used for the vacuum chamber, which
140 The European Physical Journal A

Fig. 1. The Bonn 500 MeV strong focusing electron synchrotron.

was constructed from dierent tubes glued together with Table 1. Main parameters of the 500 MeV synchrotron.
AralditTM . A van de Graa accelerator generated a 3 MeV
electron beam, which was injected into the synchrotron. Focusing type AG
The transfer beamline was calculated and set up by a PhD Number of basic periods 9
student [5], and a 60 kV electrostatic deector served for Basic lattice 1/2D-F -1/2D
Maximum eld at design orbit 1T
particle injection. Its housing was nally completely built
Maximum eld gradient 10 T/m
out of AralditTM in order to overcome the beam deteri-
Repetition frequency 50 Hz
orating eects of eddy currents, generated by the fringe Circumference 16.45 m
elds of the bending magnets. Orbit radius 1.70 m
Figure 1 shows a photograph of the 500 MeV syn- Revolution frequency 18.12 MHz
chrotron and part of its injection beamline. The main pa- No. of betatron oscillations per turn 2.4
rameters of the machine are given in table 1 (see also [6]). Gap height at orbit 6 cm
After overcoming a couple of additional throwbacks, Momentum compaction factor 0.16
the accelerator came into operation in 1958, being the Coils, no. of turns per magnet unit 98
rst strong focusing synchrotron operational in Europe. AC voltage / current 11670 V / 215 A
This happened to some extent unexpectedly, and actually DC voltage / current 100 V / 180 A
the institute was not prepared to perform scientic ex- Accelerating frequency 163.1 MHz
periments at the accelerator at this time. So Paul sent Peak voltage per turn 2.5 kV
a member of his machine group to the USA in order Number of accelerating cavities 6
to learn about physics experiments at the weak focusing Injection energy 3 MeV
1 GeV synchrotron of the California Institute of Technol- Injector type van de Graa
ogy. In the following years, a number of pilot experiments, Vacuum chamber ceramics
mainly on the photoproduction of pions o protons and Pressure (4 oil diusion pumps) 10 6 Torr
deuterons, were carried out. For the rst time the recoil
neutron polarisation was measured. Two rotating targets
produced external photon beams by the process of brems- based on the construction of the machine and scientic
strahlung. The energy of the external photon beam was experiments at this accelerator.
measured by means of tagging counters using one bend-
ing magnet of the synchrotron as dispersive element to
determine the momentum of the scattered electrons. 3 The 2.5 GeV synchrotron
The 500 MeV synchrotron was operated until 1984.
The total operation time exceeded 100000 hours. More In 1963, it became clear to the physicists perform-
than 150 diploma and doctoral theses were carried out ing experiments at the 500 MeV synchrotron that the
 W. Hillert: The Bonn Electron Stretcher Accelerator ELSA: Past and future 141

Fig. 2. The Bonn 2.5 GeV electron synchrotron, close to its nal set-up.

experimental facilities of the Bonn accelerator laboratory Table 2. Main parameters of the 2.5 GeV synchrotron.
had to be extended in order to keep up with the develop-
ment of high-energy physics worldwide. A new electron ac- Focusing type AG
celerator with higher energy was considered, which should Number of basic periods 12
again be a university machine where scientic assistants Basic lattice O/2-F D-O/2
Field at 2.3 GeV 1.003 T
and students should participate in design and construc-
Field index n 22.26, 23.26
tion. At this time, the DESY synchrotron came close to
Repetition frequency 50 Hz
operation, the Cambridge accelerator was already operat- Circumference 69.6 m
ing and it was decided to build the NINA synchrotron. Orbit radius 7.65 m
In order to contribute to lling the gap between these Revolution frequency 4.3074 MHz
6 GeV machines and the existing ones around 1 GeV, the No. of betatron oscillations per turn 3.4
Bonn group decided to build an electron synchrotron with Gap height at orbit 6 cm
a maximum energy of 2.5 GeV. Maximum of momentum compaction 1.647 m
At the end of 1963, the state government agreed to Minimum of momentum compaction 0.81 m
support the new accelerator and the design of the ma- Coils, no. of turns per magnet unit 36
chine started. The general dimensions of the synchrotron Peak current at 1 T 1360 A
were xed by the limited area of approx. 30 60 m2 avail- Vacuum chamber ceramics
able between the two physics institute buildings. Therefore Accelerating frequency 499.67 MHz
a simple magnetic structure with twelve combined func- Voltage per turn 700 kV
RF peak power 40 kW
tion bending magnets based on a O/2-F D-O/2 lattice was
Number of accelerating cavities 2
chosen. In order to correct for chromatic eects, an addi-
Injection energy 25 MeV
tional sextupole component was introduced in the focusing
Injector type linac
and defocusing sectors of the bending magnets. To reach a Accelerating frequency (injector) 2998 MHz
short construction time the pole prole and the cross sec-
tion of these AG magnets were calculated on a computer,
using dierent two-dimensional relaxation methods which
had been developed shortly before [7,8]. In order to avoid dure at other labs, no magnet model was checked in ad-
corrections of quadrupole and higher order elds at the vance. The eects of the transition zones between the F-
time of injection, it was intended to inject at a eld of at and D-sectors and the various identical blocks of each sec-
least 0.01 T. From the beginning on a slow extraction of tor, which could not be modelled by two-dimensional com-
the electron beam was planned. puter simulations, were measured after the rst magnet
In 1964, the design was completed and the major parts had been assembled, and were compensated by well de-
were ordered. In 1965, the accelerator building and the signed exponentially shaped end blocks forming the mag-
laboratories were constructed. Manufacturing of the mag- net ends.
nets blocks was started based only on the performed com- At the end of 1966, almost all components were in place
puter simulations. In contrast to the usual design proce- and in March 1967, the rst electrons could be injected
142 The European Physical Journal A

Fig. 3. Floor plan of the synchrotrons and the associated experiments.

and, without problems, accelerated up to 2.3 GeV. Accel-
eration was performed by means of two cylindrical RF
cavity resonators, which had been developed and con-
structed in a special electroforming process by DESY,
Hamburg [9]. A conventional television transmitter, built
by Telefunken, with a modied power amplier was used
as RF power source. A linear accelerator, manufactured by
Varian (Paolo Alto, California), served as injector, deliv-
ering a pulsed 25 MeV beam with a current of 250 mA at
an energy spread of 0.5 %. In July 1967, the accelerated
electrons could be extracted with an e ciency of about
60 %. Slow beam extraction was performed by exciting a
half integer horizontal betatron resonance with the help
of the nonlinear magnetic eld of a current strip.
The main parameters of the 2.5 GeV synchrotron are
listed in table 2 (see also [10]). Figure 2 shows a pho-
tograph of the machine, close to its nal set-up. In its
rst operation period from 1967 to 1984, the synchrotron
served as accelerator for particle physics experiments and
as source for synchrotron radiation. A broad spectrum
of physics experiments has been performed at this ma-
chine, starting from photo- and electroproduction of pseu-
doscalar mesons o protons and deuterons, continuing
with photoproduction of associated strangeness in K/ 0
and nally including the measurement of the recoil nu-
cleon polarisation. Since 1970, polarised solid state targets
(starting with polarised protons and later continuing with
polarised neutrons and deuterons) became available and
were also employed.
A oor plan of the experimental hall is shown in
g. 3, presenting the experimental set-ups at the two syn-
chrotrons.
The total operation time of the 2.5 GeV synchrotron
amounted to 85000 hours for its rst operation period. In
 W. Hillert: The Bonn Electron Stretcher Accelerator ELSA: Past and future 143

PHOENICS
SAPHIR
Magnetstromver-

Synchrotron

M Tagger
sorgung

Dipole (horizontal)
T
Dipole (vertical)
extraction septa
hadron Quadrupole
Sextupole
physics ELAN
superconducting DORIS cavity
Transformatoren,

solenoid Combined-Function Magnet

experiments PETRA cavity
Filter

Solenoid
Radio Frequency
NHV1
10 kV
Trafo

stretcher ring
0.5 - 3.5 GeV bending magnet
booster beamlines for
SR experiments
synchrotron BN3
BN2
0.5 - 1.6 GeV injection septa

DESY cavity
BN1
EKS
half cell of stretcher ring BN0
LINAC 1
(20 MeV) pol. e - M Q BPM
source SR
(120 keV) beamline
Mott polarimeter
electron
gun FZK laboratory
electron detector
gun
tests
LINAC 2
(26 MeV)

0m 5m 10 m 15 m

Fig. 4. Floor plan of the ELSA accelerator laboratory, indicating the rst experimental set-up of the time period 1989 1997.

total 210 diploma and doctoral theses were carried out and for excitation of a third integer betatron resonance
during this time. needed for slow beam extraction. Two long straight sec-
tions with vanishing dispersion are equipped with the ac-
celerating cavities and beam injection and extraction el-
4 The 3.5 GeV ELSA stretcher ring ements. Depending on the maximum beam energy cho-
sen, a single-cell resonator of DORIS type or two ve-cell
After more than ten years of experimental particle physics resonators of PETRA type are driven by klystron-based
at the 2.5 GeV synchrotron, it turned out that the qual- transmitters, operating at a frequency of 500 MHz and de-
ity of the experimental data was severely limited by the livering a maximum power of 40 kW and 250 kW, respec-
low duty factor of the pulsed synchrotron, which amounts tively. The vacuum system is based on thin wall (0.3 mm)
to about 5 % and could not be increased signicantly with oval tubes of stainless steel, whose rigidity is provided by
this type of accelerator. Following the suggestions of other 1 mm thick reinforcing ribs being brazed on the tubes.
laboratories around the world, it was decided to build a
A oor plan of the ELSA facility is shown in g. 4, rep-
pulse stretcher ring, using the 2.5 GeV synchrotron as in-
resenting the rst experimental set-up (1989 1997). The
jector [11]. A proposal was made in 1979, which was re-
main parameters of the stretcher ring are listed in table 3.
vised in 1981 and nally accepted at the end of 1981. After
several years of planning and designing, the construction ELSA can be operated in three dierent modes: the
of the new ELectron Stretcher Accelerator ELSA started stretcher mode, the post accelerator mode and the storage
in 1982. mode.
To avoid additional costs, the 500 MeV synchrotron Stretcher mode: Single pulses from the booster syn-
was dismantled in 1984 in order to use the existing ac- chrotron are injected into the stretcher ring at a maximum
celerator building for experiments at the stretcher ring. rate of 50 Hz. Using a slow extraction at a third integer
The new accelerator was placed in a separate tunnel sys- tune, an external electron beam of constant intensity is
tem constructed below the physics institute. ELSA is a obtained for the time between two injections. The max-
separated function machine of simple FODO-type, which imum energy is limited to 1.6 GeV by the beam transfer
provides radiation damping of the horizontal betatron os- from the synchrotron to the stretcher ring.
cillations necessary for beam storage and allows a wide Post accelerator mode: After injection of several pulses,
range variation of the betatron tune. In addition to the the accumulated electron beam is accelerated to the re-
dipoles and quadrupoles, a total number of twelve sex- quired energy and then extracted slowly. The maximum
tupoles were installed for correction of chromatic eects energy obtainable is 3.5 GeV, limited by the dipole magnet
144 The European Physical Journal A

Table 3. Main parameters of the 3.5 GeV stretcher ring ELSA.

Menu Menu Menu Graphical user interface
Focusing type AG Computers with X Window server
Number of basic periods 16
Basic lattice F ODO TCP/IP
X Window protocol
Field at 3.5 GeV 1.073 T
Circumference 164.4 m Menu program

Orbit radius 10.88 m Presentation and

Revolution frequency 1.82 MHz
Expert programs
Distributed
data base layer
data base 3 Hewlett Packard B2000/B2600
Number of dipoles 24 Operating system: HP-UX (Unix)
Number of quadrupoles 32 Menu program Menu program

Number of sextupoles 12 Expert programs Expert programs More than 12000 parameters
Gap height (dipole magnets) 5 cm Distributed
data base
Distributed
data base in data base, 19 permanently
Momentum compaction factor 6.3 % running expert programs.
Natural emittance (3.5 GeV) 0.9 mm mrad TCP/IP
UDP, RPC
Natural energy width (3.5 GeV) 0.09 %
Current (dipoles) at max. energy 3015 A Local Local Local Equipment control layer
data base data base data base
Vacuum chamber stainless steel 32 VME computers with
real time OS VxWorks (WindRiver)
Accelerating frequency 499.67 MHz Driver Driver Driver 5 PCs with Linux
Energy loss per turn (3.5 GeV) 1.22 MeV
TCP/IP, HDLC,
Maximum RF power 250 kW BUEP64, RK512
Accelerating cavities transmitter 1 1 of type DORIS Device interface layer
Accelerating cavities transmitter 2 2 of type PETRA 12 PLCs
57 MACS computers
(monoboard computers
developed in Bonn)
IEC bus,
power supply and the RF generated acceleration voltage. analog and digital I/O...

The macroscopic duty factor depends on the ramping Power Power measuring
supplies supplies instruments Equipment
speed and the at top time and scales inversely propor- etc. etc. etc.

tional to the external current. For typical operation pa-

rameters (3.2 GeV, 1 nA external current) a duty factor of
about 60 70 % is achieved.
Storage mode: When ELSA is operated as a syn-
chrotron radiation source, a large number of pulses from
the synchrotron are accumulated in ELSA. Then the en-
ergy is ramped slowly to the desired value and the beam is
stored for hours. The beam lifetime depends on the circu-
lating beam current and energy and amounts to 1 2 hours
for typical operation parameters (2.3 GeV, 50 mA).
Until 1994, the rst set-up of the accelerator control
and timing system did not allow for a fast ramping of the
dipole magnets. ELSA was only operated in the stretcher
and storage modes [12].
In 1990, experiments utilising the synchrotron radia-
tion, which is emitted from the electrons when passing
the bending dipole magnets of ELSA, started in a sepa-
rate laboratory constructed close to the accelerator tun-
nel. Since that time, a total number of six bending magnet
beamlines are used for X-ray lithography and radiation
chemistry [13], molecular fragmentation and Auger- [14],
photoemission- and X-ray absorption spectroscopy exper-
iments [15,16] and time resolved studies [17].
For the rst years, one method of beam extraction was
based on scattering the circulating electrons o a carbon
wire and collimating the external beam to the required
beam dimensions. Since 1991, after suppressing the ripple
of the main magnet power supplies by more than a fac-
tor of 100 with the help of active lters, a slow resonance
extraction is successfully applied, using the following tech-
nique:
Fig. 5. Hierarchical structure of the ELSA control system (sta-
tus 2005).
A third integer resonance is driven by four extraction
sextupoles, located in dispersion free sections and divid-
ing the phase space into a stable triangle and an unstable
surrounding area. With the additional help of four iron-
less quadrupoles, the horizontal betatron tune is slowly
shifted to a value of 4 23 , thereby shrinking the area of the
stable triangle. When applying this procedure, part of the
electron beam becomes unstable, resulting in an increas-
ing betatron amplitude. These unstable electrons move
along the separatrix lines in phase space and are nally ex-
tracted after crossing the septa of two extraction magnets.
The major problem of the pure stretcher mode oper-
ation turned out to be the achievement of a homogenous
lling of ELSA with one injection from the synchrotron.
Due to the circumference ratio of 3 : 7 of these two ma-
chines, a sophisticated three turn shaving extraction from
the synchrotron had to be employed in order to obtain a
complete lling of ELSA. Applying this method, the phase
space structure of the ELSA beam often varied along the
closed orbit shortly after injection, even in case of careful
adjustments. This structure was not equalized by radia-
tion damping within the short extraction time and caused
small scale intensity variations of the external beam, thus
decreasing the duty factor.
In 1994, a newly developed accelerator control sys-
tem [18,19] was set up. This system is based on three
Unix workstations and in total consists of more than 100
computers (status 2005) at dierent layers (see g. 5). Its
 W. Hillert: The Bonn Electron Stretcher Accelerator ELSA: Past and future
graphical user interface combines machine steering, diag-
nostics and data analysis in one integrated environment.
In combination with the control system, a new accelerator
timing system, based on programmable delay units, was
also set up. The improved systems were the key to precise
and exible tracking of the main magnets on the energy
ramp and enabled a fast ramping operation (maximum
speed about 7 GeV/s) of ELSA. In addition they allow for
joining of several injections from the synchrotron in ELSA
with adjustable overlap, thus making a nearly perfect ho-
mogeneous lling of the stretcher ring possible. Since the
successful implementation of the post accelerator mode
at the end of 1994, ELSA was no longer operated in the
stretcher mode.
Dierent experimental set-ups were supplied with an
external electron beam.
The PHOENICS experiment was performed until the
end of 1996, utilising a polarised frozen spin target and
carrying out pion and eta photoproduction o protons and
deuterons (see [20] and references therein).
The ELAN experiment (1988 1997) performed elec-
troproduction and electrodisintegration experiments, de-
tecting the scattered electrons with a magnetic spectro-
meter [21].
It was nished in August 1997 and followed up by an
experiment requiring a beam of circularly polarised pho-
tons and a polarised proton target in order to investigate
the contributions to the GDH sum rule in the energy range
accessible at ELSA (see [22] and references therein). The
GDH experiment was carried out in the years 2000 2002
after ELSA had been successfully upgraded for the oper-
ation with polarised electrons (see sect. 5).
The SAPHIR spectrometer was operated from 1991 to
1999, carrying out photoproduction experiments and con-
centrating on the detection of charged particles. A large
variety of reactions has been studied with this detector,
reaching from photoproduction of associated strangeness
in K/ 0 to the production of vector mesons , and tals without breaking the vacuum of the gun (see g. 6).
(see [23] and references therein).
SAPHIR was disassembled in 1999 and followed up
by a new experimental set-up (CB@ELSA), based on the
4 photon detector Crystal Barrel which had been moved
in 1997 from LEAR/CERN to Bonn and equipped with
an inner scintillating ber detector for charged particle
detection and triggering. During the rst data-taking pe-
riod of the CB@ELSA experiment from 2001 to 2003, pro-
ton, deuteron and solid state targets were used in photo-
production experiments (see [24] and references therein).
A linearly polarised photon beam, produced by coherent
bremsstrahlung o a goniometer-aligned diamond crystal,
was set up in 2002 for this experiment and has been rou-
tinely available since then.
5 Polarised electrons
The production of polarised electron beams has been stud-
ied in Bonn for more than 35 years. Already in 1969, a
source of polarised electrons, based on the photoionisa-
tion of polarised lithium atoms, was set up [25]. Beam
145
intensity and polarisation were enlarged using circularly
polarised UV laser light for photoionisation of unpolarised
atoms. Based on this method (Fano eect), two sources
of polarised electrons were built in Bonn; one operating
with caesium, the other with rubidium gas [26,27]. At the
Bonn 2.5 GeV synchrotron, with the help of the rubidium
source, polarised electrons were accelerated for the rst
time worldwide in a synchrotron [28].
The work on sources of polarised electrons was con-
tinued after the construction of ELSA, using the photo-
emission of GaAs crystals pumped with circularly po-
larised laserlight. After more than ve years of work set-
ting up a suitable vacuum system and a pulsed titanium
sapphire laser of su cient power, another source, based on
a GaAs-AlGaAs superlattice photocathode was brought
into operation in 1997 [29]. This 120 keV source was used
to investigate the eects of depolarising resonances in the
stretcher ring [30]. During these studies, it turned out
that reliability, life time and beam transfer e ciency of
this source would be insu cient for hadron physics exper-
iments.
The situation improved with the construction and as-
sembling of a new source, adapted for the operation with a
second linear accelerator, which had been moved from the
university of Mainz to Bonn and installed at the 2.5 GeV
synchrotron. This 50 keV source [31] is based on an invert-
ed-geometry electron gun, operated in space charge limi-
tation in order to suppress the spiking of the free running
ashlamp-pumped titanium sapphire laser. Special care
was taken in construction and set-up of the gun and the
transfer beamline to reach and maintain a low base pres-
sure (1011 mbar) and extremely low partial pressures of
poisoning gas species (1014 mbar) by application of dier-
ential pumping [32]. To improve the gun vacuum and con-
sequently the lifetime of the photocathodes, heat cleaning
and activation of the photocathodes are carried out in a
load-lock system, which in addition allows to change crys-
The 50 keV source was operated from 2000 to 2003
with a Be-InGaAs/AlGaAs strained layer superlattice
photocathode [33] for machine studies and the GDH ex-
periment, emitting a peak current of 100 mA in rectangu-
lar 1 s long electron pulses with about 80 % beam polari-
sation and demonstrating photocathode lifetimes of more
than 2000 hours.
To prevent depolarisation during acceleration in the
circular accelerators due to spin precession around the
guiding eld of the dipole magnets, the electron spins,
originally orientated longitudinally at the source, have to
be rotated in order to point perpendicularly to the ac-
celerator plane. This rotation is performed by a 90 de-
gree electrostatic bend in the low energy beamline from
the source to the linac. An additional Larmor precession
in the linear accelerator, caused by its focusing solenoid
lenses, is compensated by additional double solenoids in
the injection beamline, which allow to vary the spin rota-
tion angle while their focusing strengths remain xed. Af-
ter beam extraction out of the stretcher ring, a supercon-
ducting solenoid rotates the spin back into the accelerator
146 The European Physical Journal A

Operation Parameters
Acceleration voltage 48 kV
Repetition rate 50 Hz
Pulse length 1 s
Pulse current 100 mA
Beam polarisation 80 %
Laser spot size (diameter) 8 mm
-11
Pressure in gun chamber 10 mbar
Beam lifetime > 2000 h

Fig. 6. Set-up of the inverted source of polarised electrons and the load-lock system.

L
0.8
Longitudinal
Horizontal
PTagger / PELSA

superconducting Vertical
solenoid 0.6 synchrotron
bending
magnets
side-band

0.4 Px
J
quadrupole
Px, Py, Pz

magnets
0.2

T 0
1 1.5 2 2.5 3 3.5
Energy / GeV

Fig. 7. Spin transfer to the tagger of the GDH experiment. Py synchrotron

side-band
Pz

plane via Larmor precession. Longitudinal polarisation at

the radiator target is obtained by Thomas precession of
the spin in the bending elds of the two dipole magnets of
the external beamline located downstream of the solenoid.
Due to the limited eld strength of the solenoid and the
xed bending angle of the dipoles, the transformation of
transverse to longitudinal polarisation is incomplete and
a transverse component is always present at the radiator
target (see g. 7).
Besides the incomplete spin transfer, severe beam de-
polarisation is caused by specic depolarising resonances
appearing at certain magic energies in the circular accel-
erators. These resonances are caused by horizontal mag-
netic elds, present in every quadrupole and combined
function bending magnet and acting on the electrons if
they are passing the magnet out of its central plane. Two
dierent types of resonances have to be distinguished:
Imperfection resonances are caused by vertical dis-
placements of the beam or misaligned focusing elements
and will depolarise the beam if the spin tune a equals
an integer number. In case of electrons this happens at
energies which are integer multiples of 440.65 MeV.
Intrinsic resonances are caused by the nite verti-
cal beam size originating from the vertical betatron os-
cillations. The resonance condition is therefore linked to
Beam Energy / MeV
Fig. 8. Depolarisation due to crossing of synchrotron side-
bands, derived from numerical simulations [35].
the machine optics and depends on the vertical betatron
tune Qz and the superperiodicity P of the accelerator:
a = kP Qz , where k is an integer and P = 2 for ELSA,
but P = 12 for the booster synchrotron, respectively.
When crossing a resonance, the depolarisation Pf /Pi
depends on the crossing speed and the resonance
strength . Neglecting the in uence of the emission of syn-
chrotron radiation, we obtain, applying the well known
Froissart Stora formalism (see, e.g., [34,30]), a conserva-
tion of the polarisation for vanishing resonance strength
and a spin ip (Pf /Pi = 1) for considerably strong val-
ues of  (|| /  1). If additional depolarisation, origi-
nating from crossing of synchrotron side-bands, is included
(see g. 8), it turns out that a total spin ip cannot be
observed at ELSA at energies higher than 1.6 GeV [36].
Therefore, it is not feasible to enhance the strengths of all
resonances su ciently in order to avoid depolarisation,
 W. Hillert: The Bonn Electron Stretcher Accelerator ELSA: Past and future 147

Horizontal orbit
10

ga+0 5
Q

x / mm
Qsp = ga 0

x -5
Qz 0 20 40 60 80 100 120 140 160
DQz s/m

Vertical orbit
t0 3
2
1
Dt1 Dt2 t

z / mm
0
-1
-2
Q
-3
0 20 40 60 80 100 120 140 160
5 s/m
10-
8-g

12-
4
6-g

ga-2
2
ga+

ga+
ga+

Fig. 10. Uncorrected (solid) and corrected (dashed) closed

a
a

ga

ga

Qz orbit.

at
Injection

4
0 1 2 3 4 5 6 7 8 ga Source
Source
0 440 880 1750 2640 3500 E / MeV
0 150 400 640 t / ms
Fig. 9. Fast tune jumping of all intrinsic resonances in ELSA.

but it is required to correct for all resonances strong
enough to cause a signicant depolarisation.
Due to the fast ramping speed and high superperiodic-
ity, no signicant depolarisation is observed in the booster
synchrotron if the beam is transferred to ELSA at ener-
gies below the third imperfection resonance at 1.32 GeV.
A dierent situation shows up for the stretcher ring. It
turned out from numerical simulations that at least nine
resonances in ELSA are strong enough to produce signi-
cant depolarisation. Three techniques are applied to avoid
depolarisation:
In case of intrinsic resonances, the crossing speed is
enhanced with the help of two pulsed betatron tune jump
quadrupoles [37], thereby shifting the vertical betatron
tune by Qz 0.1 in a t2 10 ms long triangular
pulse with a rise time of t1 = 4 s (see g. 9).
The strengths of the imperfection resonances are re-
duced with a dynamic correction of the closed orbit during
the energy ramp. Vertical and horizontal displacements of
the beam in the quadrupoles are measured by 28 mon-
itor stations [38] and corrected with 19 horizontal and
21 vertical corrector magnets [39]. The following correc-
tion scheme is used: The beam is stored at the energies
of the imperfection resonances and the optimal correc-
tions are determined. Afterwards a linear interpolation
between these corrections is applied during the energy
ramp. The remaining distortions are generally smaller
than 0.2 mm rms (see g. 10).
Further reduction of the resonance strengths is
achieved by correction of specic harmonics of closed orbit
distortions relevant for a single imperfection resonance.
This method is based on an empirical determination of
two amplitude factors for each resonance by parameter
Energy / GeV
Fig. 11. Achieved beam polarisation in the stretcher ring.
variation and measuring the polarisation response of the
extracted beam with a Mller polarimeter.
With all correction methods applied successfully, po-
larised electrons can be accelerated up to 3.2 GeV. At en-
ergies higher than 2.0 GeV some polarisation loss is ob-
served due to incomplete resonance compensation (see
g. 11).
6 Future plans
Starting in 2006, a new experimental set-up of CB@ELSA
(see g. 12) will come into operation, utilising a po-
larised beam and a polarised nucleon frozen spin target
for photoproduction experiments in the framework of the
SFB/TR 16 Subnuclear Structure of Matter , funded by
the Deutsche Forschungsgemeinschaft (DFG). With this
set-up stronger demands will be put on the quality of the
external electron beam. In order to enhance the beam po-
larisation and the external current at maximum beam en-
ergy, a number of improvements are planned in the near
future, the most important of which are the following:
The 32 old quadrupole vacuum chambers will be re-
placed by new watercooled ones, equipped with improved
monitor stations (capacitive pickups) and clearing elec-
trodes for ion clearing.
The existing 40 corrector magnets will be removed and
60 newly designed magnets will be installed, together with
four-quadrant power supplies developed at the institute.
148 The European Physical Journal A

6. H. Ehrenberg, W. Paul, Die Atomwirtschaft, 300 (1959).

7. J. Drees, Proceedings of the 2nd International Conference
To on Magnet Technology, Oxford (1967).
8. P.F. Dahl, G. Parzen, R. Christian, IEEE Trans. Nucl. Sci.
oto a NS-12, 408 (1965).
9. H. Gerke, G. Schaer, 5th International Congress on Mi-
crowave Tubes, Paris (1964) p. 511.
10. K.H. Altho et al., Nucl. Instrum. Methods 61, 1 30
(1968).
w
11. D. Husmann et al., Proceedings of the 6th European Parti-
v cle Accelerator Conference (World Scientic, Rome, 1989)
A
p. 356.
t 12. K.H. Altho et al., Part. Acc. 27, 101 (1990).
13. O. Wollersheim et al., Rad. Phys. Chem. 55, 103 (1999).
14. F. von Busch et al., Physica Scripta T 80, 401 (1999).
Fig. 12. New set-up of the CB@ELSA experiment. 15. S. Bender et al., J. Non-Crystalline Solids 298, 99 (2002).
16. J. Janssen et al., Z. Anorg. Allg. Chem. 629, 1701 (2003).
17. H. Rumpf et al., J. Phys. Chem. B 105, 3415 (2001).
With this correction system it will be possible to apply 18. T. Gotz, doctoral thesis, Bonn-IR-95-03 (1995).
three times higher magnetic correction elds with a signif- 19. M. Picard, doctoral thesis, Bonn-IR-95-01 (1995).
icant smaller response time which is needed for harmonic 20. A. Bock et al., Phys. Rev. Lett. 81, 534 (1998).
correction on the fast energy ramp at higher energies. 21. H. Arenhovel et al., Phys. Rev. Lett. 75, 21 (1995).
The rst linear accelerator will be equipped with a 22. H. Dutz et al., Phys. Rev. Lett. 94, 162001 (2005).
pulsed triode gun, a subharmonic prebuncher and a TWT 23. K.-H. Glander et al., Eur. Phys. J. A 19, 251 (2004).
buncher from the disassembled SBTF set-up [40], allowing 24. O. Bartholomy et al., Phys. Rev. Lett. 94, 012003 (2005).
for a generation of single pulses with narrow pulse widths 25. G. Baum, U. Koch, Nucl. Instrum. Methods 71, 189
(< 1 ns) and a phase synchronous operation with the syn- (1969).
26. W.v. Drachenfels et al., Z. Phys. 269, 387 (1974).
chrotron. Using this set-up a single bunch accumulation
27. W.v. Drachenfels et al., Nucl. Instrum. Methods 140, 47
in the stretcher ring is planned, allowing time resolved (1977).
measurements and detailed machine studies. 28. W. Brefeld et al., Nucl. Instrum. Methods A 228, 228
One option is to increase the maximum energy of (1985).
ELSA to 5 GeV, using superconducting cavities to com- 29. S. Nakamura et al., 12th International Symposium on High
pensate for the considerably higher energy loss per turn Energy Spin Physics (World Scientic, Singapore, 1997) p.
of about 5.1 MeV at 5 GeV. In order to produce the higher 709.
bending elds of 1.53 T, the existing dipole magnets would 30. S. Nakamura et al., Nucl. Instrum. Methods A 411, 93
have to be replaced by new ones with reduced gaps. The (1998).
existing quadrupole and sextupole magnets will allow an 31. W. Hillert, M. Gowin, B. Ne, AIP Conf. Proc. 570, 961
operation up to 5 GeV. All the aspects of this energy up- (2001).
grade will be investigated and design concepts will be 32. W. Hillert, M. Gowin, B. Ne, Proceedings of GDH2000
worked out in near future. (World Scientic, Singapore, 2001) p. 283.
33. T. Nakanishi et al., Proceedings of the Low Energy Polar-
ized Electron Workshop, St. Petersburg (SPES-Publishing,
I gratefully acknowledge the assistance of Professor K.H. Alt- St. Petersburg, 1998) p. 118.
ho. In many helpful discussions, he supplied me with detailed 34. M. Froissart, R. Stora, Nucl. Instrum. Methods 7, 297
and valuable information without I could not have completed (1960).
the sections on the Bonn synchrotrons. The ELSA stretcher 35. M. Homann et al., AIP Conf. Proc. 570, 756 (2001).
ring was constructed and operated until 2001 under the re- 36. C. Steier et al., Proceedings of the 6th European Particle
sponsibility of Professor D. Husmann. I would like to thank him Accelerator Conference, Stockholm (IOP Publishing, Bris-
and all members of the ELSA accelerator group for their contri- tol, 1998) p. 433.
bution. Accelerator development, construction, and operation 37. C. Steier, D. Husmann, Proceedings of the 17th Parti-
were continuously supported by DFG, BMBF, and MIWFT. cle Accelerator Conference, Vancouver (IEEE, Piscataway,
1997) p. 1033.
38. J. Dietrich, J. Keil, I. Mohos, Proceedings of the 18th Parti-
cle Accelerator Conference, New York (IEEE, Piscataway,
References 1999) p. 2054.
39. J. Dietrich, J. Keil, I. Mohos, Proceedings of the 4th Euro-
1. N.C. Christolos, unpublished report (1950). pean Workshop on Beam Diagnostics and Instrumentation
2. N.C. Christolos, U.S. Patent no. 2.736,799, led 1950, for Particle Accelerators, Chester (Daresbury Laboratory,
issued 1956. Daresbury, 1999) p. 153.
3. E.D. Courant, M.S. Livingston, H.S. Snyder, Phys. Rev. 40. M. Schmitz, A.D. Yeremian, Proceedings of the 1994 Inter-
88, 1190 1196 (1952). national Linac Conference, Tsukuba (IEEE, Piscataway,
4. O. Gildemeister, diploma thesis, Bonn (1955). 1994) p. 569.
5. G. Knop, doctoral thesis, Bonn (1957).
Eur. Phys. J. A 28, s01, 149 160 (2006)
DOI: 10.1140/epja/i2006-09-016-3 EPJ A direct
electronic only

The Mainz Microtron MAMI Past and future

A. Jankowiaka
Institut fur Kernphysik, Universitat Mainz, J.-J.-Becher Weg 45, D-55128 Mainz, Germany

/
Published online: 15 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. The Mainz Microtron MAMI is a cascade of three racetrack microtrons, delivering since 1991
a high-quality 855 MeV, 100 A cw-electron beam for nuclear, hadron and radiation physics experiments.
An energy upgrade of this facility to 1.5 GeV by adding a Harmonic Double-Sided Microtron (HDSM)
as a fourth stage is well underway and rst beam is expected during the rst half of 2006. A detailed
description of the multiple recirculation scheme with normal conducting accelerator structures, the basis
for the reliable operation of MAMI, is given and the historical development from MAMI A to MAMI B
is described. The natural advancement to MAMI C by realizing a polytron of the next higher order, the
HDSM, is covered in the last section and a rst glimpse into the future of MAMI is given.

PACS. 29.20.-c Cyclic accelerators and storage rings 41.75.Lx Other advanced accelerator concepts
41.85.Lc Beam focusing and bending magnets, wiggler magnets, and quadrupoles

1 Introduction quality an extraordinary amplitude and phase stability of

Since the late 1950s the electromagnetic probe had proven
to be the most successful precision tool for investigating
the internal structure of the atomic nucleus and hadrons.
A great deal of the early experiments was done by high
accelerating gradient ( 20 MeV/m) pulsed linacs, rst
with a duty factor (DF) around 0.1% and nally (Saclay,
Amsterdam, MIT [1]) up to 2%. However, because the
capability of these latter machines for coincidence exper-
iments was still limited, a strong demand came up in the
early 1970s for continuous wave (cw, 100% DF) high qual-
ity electron beams in the few 100 MeV to multi GeV range
(Lindenberger und Pinkau Ausschu 1980 in Germany,
Livingston Report, 1978 and Barnes Report 1980 in
the USA ).
Two dierent paths to satisfy this demand were dis-
cussed a) the pulse stretcher-ring [2], mostly designed
for upgrading an existing pulsed injector, with the capa-
bility to reach energies in the some GeV regime, but al-
ways limited by the maximum achievable current, and b)
the multiple recirculation of the beam through a linac. For
this path b) again two methods were suggested: In the in-
dependent orbit recirculation scheme the beam is guided
a few times through one or two linacs by a quite com-
plex, but very exible achromatic and isochronous return
optics. Clearly the cw-gradient must be as high as possi-
ble and, therefore one would prefer using superconducting
(sc) rf-technology. An advantage of these machines is their
potential for energy upgrades e.g., since 1975 the limit
for a stable cw-operation of sc-cavities increased from ca.
4 MeV/m to >20 MeV/m [3]. However, for a high beam
a
e-mail: janko@kph.uni-mainz.de
the rf-wave is necessary.
The other possibility is to guide the beam by a few
simple combined optical elements many times through
a low gradient linac, and for this the racetrack microtron
(RTM) scheme ([4,5,6]) lends itself. Rf-gradients of 1
MeV/m can be achieved very stable and reliable with
quite low power ( 15 kW/m) by conventional but highly
developed normal conducting (nc) accelerating structure
technology. Further, with the high beam load by many
recirculations one gets a high e ciency. Because of the
strong longitudinal focusing of the RTM the demands on
the stability of the rf-wave are only moderate; however,
realising the necessary excellent homogeneity of the mag-
netic eld in the two 180 -magnets B/B 104 is an
ambitious task.
With electron beams one has to aim for a precision of
better than 1% of the very small cross sections in order to
achieve meaningful physics results. Therefore, the acceler-
ator must be expected to running reliably for up to 7000
hours a year, delivering a stable beam of excellent longi-
tudinal and transverse emittance to the experiments. In
addition, for measuring small interference eects, beams
of polarised electrons are crucially important.
So in 1975 a detailed design study and the construc-
tion of an RTM-cascade MAMI was started at the Institut
fur Kernphysik (IKPh) at the University of Mainz [7], with
the goal to realise a world class accelerator facility, capable
to deliver for the rst time an excellent cw-electron beam
of up to about 800 MeV. At the same time rst consid-
erations on the development of a polarised photocathode
gun began.
150 The European Physical Journal A
Fig. 1. Basic setup of a racetrack microtron (RTM).
2 The racetrack microtron (RTM)
The basic scheme of a racetrack microtron is depicted in
g. 1. For a phase coherent acceleration one has to fulll
two conditions ( = v/c = 1 assumed):
a) The so called static coherence-condition for the length
of the rst complete circulation
2(EInj + E)
L1 = k rf = + 2d, (1)
ecB
which must be an integer multiple of the rf-wavelength
rf and has to be adjusted either by the injection en-
ergy EInj or the distance d of the 180 -dipoles (mag-
netic eld B).
b) The dynamic resonance-condition for the increase in
path length from turn to turn
Li+1 Li = 2R = n rf , (2)
which must be also an integer multiple of the rf-wave-
length and is fullled by setting the energy gain per
turn to
ecB
E = n rf . (3)
2 A problem which had shown up at the rst recircu-
Moreover, one has to consider, that the synchronous phase
range for stable longitudinal motion is given by
2 < n tg(s ) < 0, (4)
and for practical reasons (individual correction steering
in the dispersive section) that the distance of the return
paths is 2R = n rf /.
Evidently for an ample longitudinal stability range and
low rf-power consumption n = 1 is the natural choice
(32.5 < s < 0). As rf-power-source a 50 kW cw-
klystron just developed for industrial heating was avail-
able in 1975 (Thomson-CSF TH2075, rf = 2449.5 MHz,
rf = 12.24 cm), so that one got 2R = 3.9 cm, enough
distance for introducing slender correction steerers on each
return path tube.
Concerning the beam optics the vertical defocusing in
the fringe eld of the 180 -magnets is compensated by a
reversed eld stripe of 20% B, so that each magnet acts
like a R long drift. In the horizontal plane the trans-
formation through one magnet is given by the negative
unit matrix. For transverse focusing two schemes were dis-
cussed:
a) A strong focusing with quadrupoles on each return
line, their strength adapted to the increasing beam en-
ergy to e.g. stay with constant beta-functions for each
turn.
b) A weak focusing with only two quadrupole-doublets
on the linac axis with naturally decreasing strength
1/E 2 and therefore increasing beta-functions.
It was decided to realise option b), because option a)
has the disadvantage of introducing a strong transver-
sal/longitudinal phase space coupling, and also the fab-
rication of very slender quadrupoles with low higher mul-
tipole content would have been di cult. Due to pseudo
damping (emittance x,y = 1/() normalised
x,y ) the beam
size stays nearly constant with increasing number of turns;
however the phase space ellipse gets nally quite at. The
focusing of beams of all energies simultaneously in the
same quadrupole doublets is the one reason for the so
called Herminghaus rule [7], that the ratio of output
to input energy of an RTM should not exceed a factor of
about 10. The other is that at a given injection energy
there is an upper limit for the magnetic eld strength B,
in order to allow for enough space for the rst return of
the beam backwards to pass aside the linac structures.
In order to stay within reasonable limits with the nec-
essary strengths of the return path correctors and also
to avoid noticeable distortions of the dynamic resonance
condition of eq. (3), the B-eld of the end magnets must
be homogeneous to about 104 . Because the homogene-
ity of the casted iron and the available mechanical manu-
facturing precision allows only an accuracy of some parts
in 103 , an extensive mapping of the magnetic eld dis-
tribution had to be done. Based on these measurements
surface correction coils were constructed, attening the
magnetic eld distribution by more than a factor of ten
(g. 2, [8]).
lators with super conducting accelerator structures ([9,
10]) was the regenerative beam blow up (BBU). If the
bunched beam passes the accelerating structure slightly o
axis, it excites a T M110 -like deecting rf-mode at ca. 1.7-
fold the frequency of the accelerating T M010 -like mode.
Its amplitude is proportional to its oset and the beam
gets a small transversal kick (g. 3). If positive feedback
conditions for the next returns are given, this will lead to
beam loss above a certain threshold current IsBBU . Ac-
cording to a short cavity model of this process [11] a
worst case approximation is
Ez BBU 1 1
ISBBU ,
(R/Q)BBU QBBU f oc N ln(Eout /Ein )
(5)
where (R/Q)BBU is the shunt impedance / rf-quality fac-
tor of the BBU-mode, Ez the accelerating eld strength,
f oc the average beta-function of the recirculation, N are
 A. Jankowiak: The Mainz Microtron MAMI
Fig. 2. Example of a eld map of a RTM 180 -dipole magnet before and after applying the surface correction coils.
the number of turns, Eout/in is the output/input beam
energy, respectively.
Numerical BBU-calculations together with rf-measure-
ments on the MAMI biperiodic accelerating structures [12]
showed, that by staggered T M110 -detuning of the linac
sections the threshold current can be shifted distinctly
above the maximum design current of 100A.
3 From MAMI A to MAMI B
Following these design principles MAMI was realised as a
3-RTM-cascade between 1979 and 1990. Since the skep-
ticism that a machine with many recirculations could be
build a proof of principle was requested. Therefore, a
14 MeV stage (MAMI A1) was built for testing and op-
timising the rf-structure and rf-control, the B-eld cor-
rection by the surface coil technique and, quite advanced
at that time, a complete computer control using steer-
ing algorithms. It was set into operation in March 1979
and later used as the rst RTM (RTM1) of MAMI A and
MAMI B. This machine was already used for rst physics
experiments [13] from November 1979 on. Only the klys-
tron TH2075 and the Van-de-Graa injector were bought,
otherwise it consisted completely of in-house-made or used
components as, e.g., the end magnets from DESY, Ham-
burg.
Past and future 151
Originally, the second stage RTM2 was planned with
an end energy of only 100 MeV with one klystron TH2075
feeding RTM1 and RTM2 [7]. However, because of a strong
demand to surpass already with this machine distinctly
the pion production threshold, it was decided to add a
second klystron and accelerate in RTM2 by a factor of
thirteen from 14 MeV to 180 MeV. The cost increase for
the larger end magnets could be lowered by using the iron
of the Heidelberg Cyclotron in use from 1943 to 1973 [14].
This setup (Van-de-Graa + RTM1 + RTM2 MAMI A)
delivered from July 1983 to October 1987 about 70% of its
18700 hours of beam time for hadron and nuclear physics
experiments. The maximum achieved beam parameters
were 187 MeV beam energy and a current of 65 A. Its
operation was funded since 1984 by a Collaborative Re-
search Centre (SFB201, Medium energy physics with the
electromagnetic interaction ) and the main components
were bought via the HBFG funding.
A somewhat weak point of MAMI A was the cheap
Van-de-Graa injector. Its maximum usable voltage of
only 2.1 MV ( = 0.981) caused a migration of the op-
erating phase in RTM1 from +15 to 22 , resulting in
a reduced longitudinal acceptance and stability. This ad-
verse eect was enhanced by the high sensitivity of the sta-
bility of the high voltage of the Van-de-Graa to any im-
pact of background radiation. Because this background
152 The European Physical Journal A
Fig. 3. Simplied sketch of the regenerative beam blow up
(BBU) mechanism.
Fig. 4. Annual operation time of MAMI, according to ma-
chine setup (tuning and development), polarised and unpo-
larised beam time, for the years 1991 to 2005. It should be
noted that in 2001 a half year shutdown took place for the
preparation of the beam lines for MAMI C. Of the remaining
4428 hours the MAMI was operating for 4277 hours, i.e. 97%.
radiation increased with increasing beam current, the volt-
age instabilities of the Van-de-Graa were the reason for
the limit of max. 65 A beam current. Moreover, the
mediocre vacuum conditions and bad accessibility of the
high voltage terminal were evidently prohibitive for any
operation of a GaAs-photocathode source of polarised
electrons.
Therefore, when transferring MAMI A as the injector
for RTM3, the Van-de-Graa was replaced by a 3.5 MeV
linac designed and built in-house [15] with high energy
stability ( 1 keV; = 0.992). At this energy the phase
migration in RTM1 is only 12 to 22 . ation camera system. Very helpful for fast tuning of the
The nal scheme realised as MAMI B, set up in new
halls from 1987 to 1990 with rst operation in August
1990, is shown in g. 12, and its main data are given in
table 1.
Apart from its excellent beam parameters (table 1),
the machine showed an extraordinarily stable and reliable
operation. The beam time over the years of operation since
1991, classied for machine setup and polarised and un-
polarised operation for experiments, is shown in the his-
togram of g. 4. The high e ciency of MAMI is due on
one side to the inherent properties of the RTM, but also
to a considerable extent to a sophisticated monitor sys-
Fig. 5. Simplied scheme of the measurement of the absolute
beam energy (top) and beam energy uctuations (bottom) in
RTM3.
tem [16], allowing to computer control a wide diversity of
parameters and feedback loops. Among the most impor-
tant monitors are the low-Q rf-cavities on the linac axes
of each RTM. They allow by injection of 10 ns diagnos-
tic beam pulses during machine setup, actually realised
as 10 ns blackouts in the cw-operation, to supervise the
transverse positions and the phase and intensity of the
beam for each recirculation individually at the entrance
and exit of the linac. This information make quick and
e cient correcting actions possible by a machine model
implemented in the computer control system [17].
The beam prole is viewed turn by turn and in the
transfer lines between the RTMs via a synchrotron radi-
matching of the beam parameters is a synchrotron radia-
tion camera with high magnication looking through the
axes of the linacs of RTM2 and RTM3. All 51 (RTM2)
and 90 (RTM3) turns have to overlap and any mismatch
can easily be detected and globally corrected. A system of
many T M110 -rf-cavities allows a control of the beam po-
sition in diagnostic-pulse and also cw mode (above 1 A
beam current) down to a few m.
The RTM conguration easily permitted in RTM3 the
installation of two control setups extremely valuable for
precision experiments (g. 5). The distance between the
return pipes is such that small 4 2.45 GHz = 9.8 GHz
 A. Jankowiak: The Mainz Microtron MAMI Past and future 153

Table 1. Main parameters of MAMI. MAMI C comprises the injector linac, MAMI A, MAMI B and the HDSM being con-
structed.
Injector RTM1 RTM2 RTM3 HDSM
General
injection / extraction energy (total) [GeV] 0.511/3.97 10 3
3.97/14.86 10 3
14.86/180 10 3
0.180 / 0.855 0.855 / 1.5
number of turns 18 51 90 43
total power consumption [kW] 92 92 220 650 1400
Rf-System
energy resp. energy gain / turn [MeV] 3.5 0.599 3.24 7.50 16.58-13.66
frequency [GHz] 2.4495 2.4495 2.4495 2.4495 4.8990|2.4495
linac length (electrically) [m] 4.93 0.80 3.55 8.87 8.57|10.1
number of sections / klystrons 3/1 1/1 2/2 5/5 8/4|5/5
power dissipation / beam power [kW] 33.2 / 0.35 7.9 / 1.1 48.4/16.6 102.5 / 67.5 299 / 65
power consumption [kW] 90 90 180 450 1000 a
Magnet-System
ux density (within the gap) [T] 0.1026 0.5550 1.2842 1.53-0.95
gap height [cm] 6 7 10 8.5-13.9
min./max. de ection radius [m] 0.129-0.482 0.089-1.083 0.467-2.216 2.23-4.60
iron / copper weigth of the magnets [t] 4 / 0.2 90 / 2.3 900 / 11.6 1000 / 27.4
number of corrector magnets 40 72 204 360 2 172 + 2 6
number of quadrupoles and solenoids 20 2 4 4 24
power consumption [kW] 2 2 40 200 400
Beam-Parameters
energy spread (1) [keV] 1.2 1.2 2.8 13 110 b
norm. emittance hor. / vert. (1) [ 10 6 m] 0.05 / 0.04 0.07 / 0.07 0.25 / 0.13 13 / 0.84 27 b / 1.2 b
standard-energies for experiments 180MeV 195-855MeV 0.855-1.5GeV
in steps of 15MeV in steps of ca. 15MeV
a
Including the power consumption of one matching section between RTM3 and HDSM.
b
Simulation with SYTRACE, a particle tracking program including eects of stochastic emission of synchrotron radiation photons.

- T M010 - and T M110 -resonators can be inserted there
(T M110 -cavities at 720 and 855 MeV and T M010 -cavities
at 315, 420, 510, 570 and 855 MeV). So rstly, with the
T M110 -position monitor and the known distance between
this monitor and the linac axis, one can use one 180 -
dipole with its NMR-controlled eld strength and pre-
cisely measured eld map as a sensitive spectrometer with
large 2.2 m bending radius to determine the absolute beam
energy to 2 104 (140 keV at 855 MeV). Secondly,
with an additional 9.8 GHz T M010 -cavity on the extrac-
tion path and monitoring its phase dierence to one of the
return-pipe T M010 -resonators, one can precisely measure
changes of the length of the last half turn after extrac-
tion of the respective energy. The corresponding energy
change of the electron bunches is given by eq. (2) and the
sensitivity reads
rf /2 61.2 mm
= = 8.16 mm/MeV,
E/turn 7.5 MeV
which corresponds to 96 phase per MeV at 9.8 GHz. With
a resolution of 0.1 at 9.8 GHz energy changes of about
1 keV, corresponding to 1.2 106 at 855 MeV, are de-
tected. A further increase of resolution seems not to be
reasonable at the moment since uctuations of the beam
direction are producing signal levels of about the same
amount. By feeding back the energy signal to the linac
phase, it is possible to routinely provide this energy sta-
bility of 1 keV during physics experiments. Of course,
correct tuning and a su cient stability of the longitudinal
Q-value are key to the well functioning of the system [18].
This setup was of fundamental importance for the parity-
violating electron scattering experiments, where the cross
5
section change with Ebeam and must be measured with a
relative precision of 106 .
The development of the polarised source [19] after
MAMI ran with a clean linac as injector is listed in the
following time table:
1992 1995: First experiments for the A3-collabora-
tion, beam current I = 5 A, polarisation P = 30
40%. The 100 keV source was installed at ground
level meaning a 14 m long beam line to the injector
linac [20].
1995: Introduction of strained layer cathodes, I =
2 A, P = 75%.
1997: The source moves to the accelerator hall, which
allows a much easier and reproducible injection of the
beam into the linac. Installation of a 2f-prebuncher
with 4.9 GHz [21], which allows for 145 longitudinal
phase space acceptance instead of the design value of
40 [15], I = 10 A, P > 75%.
1998: Introduction of the so called synchro-laser for
(6) pulsing the source, which allows for 90% transmis-
sion of the precious polarised electrons, I = 20 A,
P = 80%.
2001: Introduction of the mask activation technique,
which strongly reduces losses of electrons starting at
the cathode due to stray light and thus improves the
vacuum conditions at the cathode. The charge ex-
tracted in one run was increased from 22 C to 115 C.
From then on several weeks of continuous operation at
high current I > 20 A were possible.
2003 2005: A Wien lter as spin-rotator at 100 keV
electron energy directly behind the polarised source
was installed. This allows for a much easier adjust-
ment of the beam polarisation at the target for all ex-
perimental stations and all energies. Before, the spin
direction at the experiments was controlled by tuning
the MAMI end energy making use of the gyromagnetic
154 The European Physical Journal A
Fig. 6. Detailed scheme of the Harmonic Double Sided Microtron (HDSM) for MAMI C.
anomaly of the electron. Currents of I > 30 A with
polarisations of P > 85% are now possible.
Apart from steady improvements of the degree of polarisa-
tion and the lifetime of the photocathodes, the installation
of the harmonic 2f-prebuncher at the injector linac and the
rf-synchronised laser for up to 90% transmission e ciency
made it possible to increase the MAMI operation with po-
larised electrons from 20% to now 60% of the total beam
time satisfying the demands of the experiments.
4 The Harmonic Double-Sided Microtron
(HDSM) as the fourth stage of MAMI
In 1999 a new Collaborative Research Centre (SFB443,
Many-body structure of strongly interacting systems )
was founded, which for its second stage physics program
demanded an electron beam of 1.5 GeV. The ideas for up-
grading the MAMI energy had to consider as boundary
conditions that the excellent beam quality and reliabil-
ity of MAMI B must be preserved, that the new fourth
stage had to t into the existing buildings and that the
research with the existing MAMI B had to go on without
any longer shutdown periods. Moreover, considering the
limited manpower capacity of the institute and the tight
time schedule envisaged, it was evident that one had to
base the new accelerator stage on the expertise of the insti-
tute. So very early the decision was taken to stay with the
well proven and tested technology applied for the RTMs:
normal conducting rf-accelerator structures and iron core
magnets with normal conducting excitation coils.
The latter point clearly implied that one could not
realise MAMI C as a fourth RTM. With iron core mag-
nets one cannot increase the eld strength very much
beyond the 1.3 T of RTM3. The size and weight of the
two such 180 end magnets would grow with the cube
of the maximum energy, i.e. to formidable weight of
450 t(1.5/0.855)3 = 2430 t each. However, the RTM with
one linac is not the only possible microtron. Already since
1979 H. Herminghaus and K.H. Kaiser developed ideas
and designs for higher order microtrons called Poly-
trons , as multi-turn recirculators with strong phase fo-
cusing ([22,23,24]).
At the bicyclotron or Double-Sided Microtron
(DSM) (see the scheme in g. 6) the pole face area is
reduced by a factor of ( 2)/ compared to an RTM.
Therefore, a 1500 MeV DSM has roughly the same mag-
nets weight as an 850 MeV RTM. As the next step one
must consider, however, that the dynamic coherence con-
dition changes and for a DSM is given by
ecB ecB
E/turn = n 2rf = n rf , (7)
2( 2) 2
and naturally one will take n = 1 for the lowest possi-
ble path lengthening of 2 rf /turn. With the parameters
of RTM3 (B = 1.28 T, rf = 0.1224 m) one would need
 A. Jankowiak: The Mainz Microtron MAMI
Fig. 7. Field gradient perpendicular to the pole edge of the
90 dipoles of the HDSM for compensation of the vertical edge-
defocusing.
Fig. 8. Development of the synchronous phases of both linacs
in DSM conguration for a phasing error between the two
linacs of 3 .
E = 41.1 MeV/turn. With the moderate well tested
MAMI rf-gradient of 1 MV/m one would need 20 m long
linacs, which would not t into the existing buildings
(g. 12) and would moreover consume about four times the
electric power of MAMI B. So it was evident that the fre-
quency of the DSM had to be 4.90 GHz (rf = 0.0612 m),
with two about 10 m long linacs and with the other pa-
rameters similar to that of RTM3.
Concerning the transverse optics, the 45 entrance and
exit angles of the beam at the four 90 magnets are with
their strong vertical defocusing a very critical point. A de-
tailed investigation [25] was done for several quadrupole
congurations on the dispersive paths (quadrupole triplet
/ two quadrupole doublets on each half recirculation). It
showed, that this way of compensation would work only in
Past and future 155
Fig. 9. Longitudinal input and output phase space for the
DSM- / HDSM-conguration (left / right) with phasing errors
of 3 @4.90GHz / 5 @4.90GHz.
principle. To avoid strong distortions of the phase space,
the quadrupoles must be extremely free of sextupole and
other higher multipole errors and their individual setting
and alignment for a dispersion free beam on the linac axis
would be in practice a very cumbersome procedure. There-
fore, it was decided to use the combined function solution,
namely to introduce a magnetic eld gradient perpendicu-
lar to the pole edge in the 90 dipoles, which compensates
for the edge-defocusing in the complete range of beam en-
ergies. The corresponding eld prole is given in g. 7.
Due to the eld decay from 100% to 60%, however, the
mean eld along the beam path decreases with increasing
energy so that the necessary synchronous energy gain be-
comes lower and lower. As a consequence, the synchronous
phase has to move away from the crest of the rf-waves in
the linac in order to full the dynamic coherence condition
(eq. (7)). This happens automatically and smoothly by the
longitudinal focusing if input energy and linac phases are
optimised with respect to minimum phase oscillations. For
an injection phase of 8 one will end at 34 for turn
43. This is well within the phase stable range of the DSM
(4 < n tg(s ) < 0, i.e. 51.9 < s < 0 for n = 1),
however, only for the ideally symmetric DSM, where one
half turn can be considered as a machine period. If the
symmetry is perturbed, e.g. by a wrong phase setting of
one of the linacs, the DSM can be considered as spliting
into two RTMs with a stable phase boundary of 32.5 .
From g. 8, assuming a phase error of only 3 between
the two linacs, one sees that only the last ca. 10 turns are
aected by this stopband. However, more detailed track-
ing calculations showed that the acceptance of the DSM
is distinctly diminished (g. 9, left). Therefore, a stan-
dard DSM with gradient dipole magnets would have to be
operated with extreme care for its symmetry, e.g. a phase
change of 1 at 4.9 GHz would mean a change of a steering
cables eective length of only 0.1 mm.
However, because the reason for this unstable region is
nally the too strong longitudinal focusing when the elec-
trons slip down on the 4.9 GHz wave, a glance at the time
table of the bunches in the two linacs of a DSM with
sub-harmonic injection showed a way out of this di culty
(g. 10). One can see, that in one of the linacs only every
second bucket is populated by the recirculated bunches.
So one could operate here with the MAMI B frequency
156 The European Physical Journal A
Fig. 10. Simplied scheme of the bunch arrival time in a DSM
operating with a sub-harmonic injection frequency.
Fig. 11. Bunch phase migration in the DSM (phase scale
shifted by 90 ) and HDSM conguration. Clearly the instable
area of the 4.9 GHz wave can be avoided only in the HDSM
scheme.
of 2.45 GHz, and it turned out, that by an appropriate
amplitude and phase relation between the 2.45 GHz and
4.9 GHz wave most of the necessary reduction of the en-
ergy gain can be overtaken by the low frequency linac with
its less steep gradient (g. 11). On the 4.9 GHz wave the
bunches migrate now only from 0 to ca. 26 and so stay
distinctly away from the dangerous 32.5 .
This Harmonic Double-Sided Microtron (HDSM, [26])
shows a good longitudinal stability, e.g. a phase error of
more than 5 between the two linacs is tolerable (g. 9,
right). Concerning the transverse optics, due to the mod-
erate total energy gain factor of the HDSM of only 1.75,
it is possible to stay with the horizontal and vertical
beta-functions below 20 m during the acceleration pro-
cess with just two quadrupole doublets on each linac
axis. Tracking calculations taking into account the ef-
fect of emittance growth due to quantum uctuations
of synchrotron radiation1 [27], show that the normalised
longitudinal/horizontal emittances increase only by a fac-
tor of 2/1.5. Therefore, the absolute emittance will stay
nearly constant and the beam sizes on the linac axes are
in the order of only some tenths of a millimetre. The nal
design parameters for the HDSM are given in table 1, its
scheme in g. 6 and its oor plan in g. 12.
The MAMI B frequency of 2.45 GHz was in a well-
established industrial heating band, with many compo-
nents available from the shelve, whereas at 4.9 GHz there
existed nearly no high power rf-components. Therefore,
here many developments had to be started. To have a
quick start of production for the ve rf-sections needed
for the 2.45 GHz linac, it was decided to make them as
copies of the well tried ones at RTM3 [28] (with a slightly
adapted length, 33 accelerating cells (AC) instead of 29).
The contract was given to the INP of the Moscow State
University collaborating with the klystron rm TORIY,
because of their great experience in developing and op-
timising the biperiodic on-axis coupled structures. How-
ever, because of many technical failures (vacuum leaks,
contaminations in the sections resulting in severe multi-
pacting eects) only three sections were nally delivered.
The contract had to be terminated, and the remaining
sections were produced by the company ACCEL Instru-
ments (Bergisch Gladbach, Germany) [29] without greater
di culties, apart from also some multipactor problems.
Concerning the 4.9 GHz accelerating structures more
time was available because of the lack of a high-power
klystron at the beginning. So an intensive optimising work
was done at IKPh. For the quite tiny cavities the relative
mechanical tolerances had to be relaxed for a promising
industrial series production. This aected mainly the cell-
to-cell coupling increasing it by a factor of two (k2.45 GHz =
4%, k4.9 GHz = 8.8%, [30]). As for their modied geom-
etry there remained uncertainties, e.g. concerning multi-
pactoring, and a prototype was successfully built and full
high power tests were performed at the IKPh [31]. The se-
ries production of these very well behaving structures was
done without any di cult problems again in cooperation
with ACCEL.
Concerning the high power rf-sources for MAMI C, the
2.45 GHz klystron posed no fundamental problem. The
tube (TH2174) was delivered by THALES Electron De-
vices as an improved and modernised version with bet-
ter electron beam focusing of the old TH2075 used at
MAMI B.
At 4.9 GHz a new klystron had to be developed, and of-
fers were asked from THALES and CPI/Varian. No prob-
lems were expected, because the power-limiting curves
given in the literature [32] asserted, that at this frequency
cw-klystrons up to several 100 kW could be built. Quite
ample specications were given to the factories (60 kW
cw for feeding two 4.9 GHz sections, 55% e ciency and
47 dB gain), also with respect to a later energy upgrade
1
These eects scale with E 5 7 and, therefore, with a beam
energy in the order of 1 GeV their inuence on the beam ac-
celeration needs to be carefully investigated.
 A. Jankowiak: The Mainz Microtron MAMI Past and future 157

Fig. 12. Floor plan of MAMI C. The installation up to RTM3 (MAMI B) has not been changed essentially since 1990 (A2:
Tagger / A4: Parity Violation / X1: Radiation Physics, till 2000 in the HDSM Hall).

of the HDSM. The contract was given to THALES. Con-
sidering both bidders would have caused an increase of the
total costs of this system of more than 50%. But obviously
the problems for fabricating a power klystron at this high
frequency were underestimated. It took a long series of
prototypes partly damaged by trivial technical failures
and 27 months compared to the anticipated 12 months
delivery time, before the desired tube was in house and
could be used for the power tests of many other 4.9 GHz
components (rf-structures, circulators, special waveguide
components). These tests were successfully performed in
2003. However, the prototype TH2166-tube showed strong
multipacting discontinuities on its transfer curve and was
therefore not qualied for precision operation at the ac-
celerator. THALES could solve the multipacting problem
at the next prototype by Ti-coating the nose cones of
the klystron resonators, but unfortunately through the
higher surface resistivity thermal problems occurred. Fi-
nally, with a total delay of 26 months all tubes needed for
MAMI C were delivered, fortunately, with a production
guarantee for the TH2166 tube till 2010 by THALES.
However, with somewhat reduced specications (50 kW,
45% e ciency and multipactor freeness only for some-
what restricted operating conditions), which is just safely
adequate for the 1.5 GeV operation of the HDSM.
All other high power components, e.g. the circula-
tors by AFT and the water loads from Spinner, worked
satisfactorily from the beginning. The two 30 kV / 27 A
klystron power supplies built by BRUKER, Wissenbourg,
were successfully operated in several longterm tests,
158 The European Physical Journal A

Fig. 13. Measured vertical B-eld of HDSM dipole No.2 normalised to the ideal eld gradient (top), and the construction
drawing of the corresponding correction coil (bottom). The dark quadratic area at the lower left of the eld map is due to a
piece of parallel pole faces necessary for an NMR-probe for precise eld regulation.

Table 2. Main parameters of the HDSM dipoles.
eld strength [T] 1.53 - 0.95
gap distance [mm] 85 - 139
mech. length of front edge [m] 7
usable length of front edge [m] 6.5
iron weight [t] 250
coils copper prole outside [mm2 ] 12 12
coils cooling duct diameter [mm] 8 casted out of high permeable iron and then machined at
number of windings 2 256
current/voltage [A/V] 212/340
copper weight [t] 6.85
especially during the high power conditioning and test-
ing of the fully installed and commissioned 2.45 GHz linac
of the HDSM.
Naturally, beside the task to design and build up the
worldwide rst 4.9 GHz cw linac, the manufacturing of the
four 90 -bending magnets with eld gradient presented
the second highly critical challenge for the completion of
the HDSM. The mechanical and magnetic design of these
dipoles was completely done at IKPh. The main goal was
to get magnets with excellent eld quality at minimum size
(existing halls) and iron consumption ([33,34]). The main
parameters of these magnets are given in table 2. The
call for tender started in 1999 and in 2000 the contract
was awarded to the French company USINOR2 . Aside the
promising manufacturing capabilities of this company the
main argument was, that only USINOR oered to pro-
duce the magnets essentially of only two symmetric pieces
(upper and lower piece), which is clearly the favourable
geometry to avoid any discontinuities perpendicular to the
pole edge. The magnet pieces, each weighing 125 t, were
the company SFAR, a subcontractor of USINOR. This
machining procedure for a high quality and precise surface
of the partly concave pole pieces was worked out in close
collaboration with IKPh. Due to the complicated pole ge-
ometry it was expected, that for the nal eld correction
of the magnets to the 104 level not only symmetric, but
also asymmetric eld errors (resulting in unwanted eld
components in the plane of beam acceleration) must to
be corrected. Based on the well proven concept of surface
correction currents [35], a procedure had been developed
which allows to extract the symmetric and asymmetric
eld components by a simultaneous measurement of the
vertical magnetic eld in and 25 mm out of the midplane
of the magnets, and to construct surface correction coils
which compensate both errors simultaneously ([36,37]).
The rst magnet was delivered end of 2001 and all four
magnets were nally in place end of 2002. The contract for
2
Today SFAR STEEL (Le Creusot, France).
 A. Jankowiak: The Mainz Microtron MAMI
the manufacturing of the excitation coils was awarded to
the company SIGMAPHI (Vannes, France). They intro-
duced a special bi-lar winding technology, which allowed
to realise optimum heat distribution within the coils, to
avoid internal brazing and to choose reasonable power sup-
ply parameters [34]. Both guarantees a high reliability over
the lifetime of the accelerator. Each magnet is fed by an
individual, highly stabilised power supply (478 V, 260 A,
short term/long term stability: 3 ppm/10 ppm) manufac-
tured by DANFYSIK (Jyllinge, Denmark). By feeding
back the reading of NMR-probes to the PS, the eld of
each magnet is stabilised to better than 105 .
It took till September 2003 to nish all magnet eld
measurements. To explore the capabilities of the mag-
nets for a later energy upgrade, these measurements were
not only done at the nominal eld of 1.53 T but also at
1.64 T = 1.61 GeV and even 1.71 T = 1.67 GeV. In g. 13
the measured eld of dipole No.2 (at the nominal eld of
1.53 T normalised to the ideal eld gradient) is plotted.
In the central area of the magnet the eld deviations are
already in the order of 104 , a clear proof of the excel-
lent work done by USINOR/SFAR. As a further result
of this high manufacturing precision, the analysis of the
asymmetric eld errors of the magnets showed, that the
transverse components are well below 1 mT. A rough es-
timation of the inuence of the resulting vertical beam
deections of 0.1 mrad to max. 0.35 mrad leads to an ac-
ceptable coupling of only a few percent between the hor-
izontal and vertical phase spaces. So it was decided to do
the nal correction only for symmetric eld errors, result-
ing in much simpler identical upper and lower correction
coils. In the lower part of g. 13 a sketch of one of this cor-
rection coils, manufactured by water jet cutting of a 3 mm
thick aluminium plate, is shown. With these pairs of coils
the desired eld accuracy of 2 104 was easily achieved
for all four magnets. One can clearly see, that most of the
correction must be done near the corners of the magnet,
because here quite large eld decays exist. This behaviour
was already predicted by TOSCA-simulations and are due
to the triangular cut necessary to t the magnets as far as
possible into the corners of the accelerator hall. It turned
out, that even at the design eld level of 1.53 T this eld
decay leads to deection errors of up to 2.2 mrad at low
electron energies. Because it reaches far into the fringe
eld region, it cannot be corrected by surface correction
coils alone. Therefore, at the entrance and exit corner of
each dipole individually designed vertical iron shims at-
tached to its front face are necessary. Together with the
steering magnets on the return paths and the linac axis
they will provide a proper angle and position correction of
the beam [37]. Because with increasing eld of the dipoles
the eld decay at the magnet corners gets much stronger,
a later energy upgrade of the HDSM, based on the ex-
periences gained during the operation at 1.5 GeV, will
most probably require the construction and installation
of a fully new set of correction coils and iron shims.
Presently, all four dipole magnets are aligned and
equipped with their individual set of correction coils and
vacuum chambers. The 2.45 GHz linac is commissioned
Past and future 159
and ready for operation, whereas the installation of the
4.90 GHz linac, after the nal delivery of the 10 needed
accelerator sections, has just started. The next step is
the installation of the two recirculation path vacuum sys-
tems and the completion of the injection and extraction
beam lines. The rst operation of the HDSM is expected
in the rst half of 2006. After a period of commission-
ing in diagnostic pulse mode with low beam power (10 ns,
high-intensity bunch trains with a repetition rate of max.
10 kHz), very soon the rst physics experiments will be
started since all upgrades of the beams lines, the photon
tagger and the spectrometers has been nished.
Many people worked together to realise the very success-
ful operation of MAMI over the last 25 years. Here I just
want to mention the certainly most important ones: Hel-
mut Herminghaus, who is the intellectual father of MAMI and
laid the strong foundations for this success story and Karl-
Heinz Kaiser, who overtook the responsibility for the MAMI
operation and development in the early 1990s and set the
guidelines for the future of MAMI: the design and installation
of the Harmonic Double-Sided Microtron. The construction
and operation of MAMI would not have been possible with-
out Hans Euteneuer who is, amongst others, responsible for
the developments in the rf-eld. His great help for the prepa-
ration of this manuscript must be stressed as my last point.
References
1. J. Haimson, Linear Accelerators, edited by P.M. Lapos-
tolle, A.L. Septier (Amsterdam, 1970) p. 415.
2. D. Husmann, IEEE Trans. Nucl. Sci. NS-30, No. 4, 3252
(1983).
3. L. Harwood, Proceedings of PAC2003, Portland, OR, USA
(2003) p. 586.
4. E.M. Moroz, Sov. Phys. Dokl. 1, 326 (1956).
5. A. Roberts, Ann. Phys. (N.Y.) 4, 115 (1958).
6. B.H. Wiik et al., Linear Accelerators, edited by P.M. La-
postolle, A.L. Septier (Amsterdam, 1970) p. 553.
7. H. Herminghaus et al., Nucl. Instrum. Methods 138, 1
(1976).
8. H. Herminghaus et al., Nucl. Instrum. Methods 187, 103
(1981).
9. P. Axel et al., IEEE Trans. Nucl. Sci. NS-24, No. 3, 1133
(1977).
10. O. Hanson, Charlottesville Conference Paper Q (1979).
11. H. Herminghaus et al., Nucl. Instrum. Methods 163, 299
(1979).
12. H. Euteneuer et al., Proceedings of LINAC84, Seeheim,
Germany (1984) p. 394.
13. M. Begemann et al., Nucl. Instrum. Methods 201, 287
(1982).
14. U. Schmidt-Rohr, Die Deutschen Teilchenbeschleuniger
(U. Schmidt-Rohr, Heidelberg, 2001) p. 144.
15. H. Euteneuer et al., Proceedings of EPAC88, Rome, Italy
(1988) p. 550.
16. H. Euteneuer et al., Proceedings of LINAC92, Ottawa,
Canada (1992) p. 356.
17. H.J. Kreidel, PhD Thesis, KPH 12/87, University of
Mainz, Mainz, Germany (1987).
160 The European Physical Journal A
18. M. Seidl, Proceedings of EPAC2000, Vienna, Austria
(2000) p. 1930.
19. K. Aulenbacher et al., Journal AIP, Vol. 675 (2002) 1088.
20. H. Steens PhD Thesis, KPH 01/94, University of Mainz,
Mainz, Germany (1994).
21. V.I. Shvedunov et al., Proceedings of EPAC96, Barcelona,
Spain (1996) p. 1556.
22. K.H. Kaiser, Proceedings of the Conference on Future
Possibilities for Electron Accelerators, Charlottesville, VA,
USA (1979) V-1.
23. H. Herminghaus et al., Proceedings of LINAC81, Santa Fe,
N.M., USA (1981) p. 260.
24. H. Herminghaus, Nucl. Instrum. Methods A 305, 1 (1991).
25. S. Ratschow, PhD Thesis, KPH 02/00, University of
Mainz, Mainz, Germany (2000).
26. A. Jankowiak et al., Proceedings of EPAC02, Paris, France
(2002) p. 1085.
27. J. Herrmann et al., Proceedings of PAC99, New York, USA
(1999) p. 2915.
28. H. Euteneuer et al., Proceedings of LINAC86, Stanford,
CA, USA (1986) p. 508.
29. H. Euteneuer et al., to be published in Proceedings of
EPAC06, Edinburgh, GB (2006).
30. H. Euteneuer et al., Proceedings of EPAC00, Vienna, Aus-
tria (2000) p. 1954.
31. A. Jankowiak et al., Proceedings of LINAC04, Lubeck, Ger-
many (2004) p. 842.
32. G. Faillon et al., Proceedings of LINAC86, Stanford, CA,
USA (1986) p. 122.
33. U. Ludwig-Mertin et al., Proceedings of EPAC98, Stock-
holm, Sweden (1998) p. 1931.
34. A. Thomas et al., Proceedings of EPAC02, Paris, France
(2002) p. 2379.
35. H. Herminghaus, Proceedings of EPAC88, Rome, Italy
(1988) p. 1151.
36. M. Seidl et al., Phys. Rev. STAB 5, 062402 (2002).
37. F. Hagenbuck et al., Proceedings of EPAC04, Lucerne,
Switzerland (2004) p. 1669.
Eur. Phys. J. A 28, s01, 161 171 (2006)
DOI: 10.1140/epja/i2006-09-017-2 EPJ A direct
electronic only

The Gerasimov-Drell-Hearn sum rule at MAMI

A. Thomasa
Institut fur Kernphysik, Universitat Mainz, 55099 Mainz, Germany

/
Published online: 23 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. An extended experimental program to investigate the Gerasimov-Drell-Hearn (GDH) sum rule
and related partial reaction cross sections on proton and neutron has been carried out by the GDH collab-
oration at the electron accelerators MAMI (Mainz) and ELSA (Bonn). The GDH sum rule connects the
helicity-dependent photoabsorption cross section with the anomalous magnetic moment of the nucleon. The
GDH collaboration has measured the total cross section of circularly polarised photons with longitudinally
polarised protons at MAMI and ELSA to check this sum rule experimentally. In addition partial reaction
channels like pion, double pion and eta production were determined. This provides new information on the
helicity-dependent excitation spectrum of the nucleon. With the help of partial wave analyses it is possible
to extract new, complementary information on the broad, overlapping resonances in this energy region.
The double polarisation observable E measured in this experiment enhances the smaller multipoles via
interference terms. The analysis of our data provides a new possibility to study the photon couplings to
the nucleon resonances, especially above the resonance where many properties of the observed states
(e.g., coupling constants, branching ratios, helicity amplitudes) are only poorly known. In this paper we
present several new results from our measurements on polarised proton and deuteron targets.

PACS. 16.60.Le Meson production 14.20 Baryon resonances with S = 0 25.20 Photoproduction

1 Introduction can be seen as an extrapolation from the Bjorken and the

The GDH sum rule connects static properties of the nu-
cleon like the anomalous magnetic moment and the nu-
cleon mass M , with the helicity dependent total absorp-
tion cross sections 1/2 and 3/2 , which are related to the
dynamics of the excitation spectrum:
 0 =
2 2 2 d 4 2 0 3
2
= (3/2 1/2 ). (1)
M 0
Eectively for the nucleon the lower integration limit is
the -production threshold energy and denotes the pho-
ton energy, is the ne-structure constant. The GDH
sum rule was derived under very general assumptions
(Lorentz and gauge invariance, causality, relativity, uni-
tarity and the no-subtraction hypothesis) in 1966 by
Gerasimov, Drell and Hearn [1,2], but it was not checked
experimentally until the pioneering experiment at MAMI
in 1998. Some authors [3,4] have calculated the right
hand side of eq. (1) using partial wave analyses of single
-photoproduction experiments and rough estimates for
the double contribution. They always obtained a dis-
crepancy with the left side of eq. (1), which yields 205 b
for the proton. The interest in the GDH sum rule was re-
newed with the measurements of the spin structure func-
tions for proton and neutron, because the GDH sum rule
a
e-mail: thomas@kph.uni-mainz.de
Ellis-Jae sum rules [5] for deep inelastic lepton scattering
to the real photon limit.
Another characteristic property of the nucleon is the
forward spin polarisability:

1 d
(3/2 1/2 ). (2)
Due to the stronger weighting of the integrand with the
third power of the inverse photon energy this integral con-
verges at much lower energies. The predictions for 0 by
several theoretical calculations reveal serious discrepan-
cies. Especially the results obtained by dispersion relations
dier signicantly from the predictions of chiral perturba-
tion theory.
Nowadays, improved technologies for polarized pho-
ton beams and polarised targets allow us to check this
sum rule performing a dedicated double polarisation ex-
periment directly. The goal of the GDH collaboration is to
measure the energy dependence of the helicity-dependent
total absorption cross section as well as partial reaction
channels on proton and neutron targets to determine the
dominant contributions to eq. (1). In addition to checking
the sum rule experimentally there is a strong motivation
to carefully measure the integrand and the helicity depen-
dence of the partial reaction channels such as single- or
double-pion photoproduction, which provides completely
162 The European Physical Journal A

Fig. 1. The Mainz Microtron MAMI. The GDH experiment was performed in the A2 tagger hall.

DAPHNE Cerenkov Star

polarized -strip
target
polarized
-beam

shower
trigger plates
1m

Fig. 2. Overview of the GDH experimental setup at MAMI. The polarised photon beam is coming along the axis of the 3 He/4 He
refrigerator of the polarised frozen spin target.

new and up to now inaccessible information on partial-
wave amplitudes. Besides the measurements with the po-
larised proton target, we have also performed an extended
series of experiments with polarised deuteron targets in or-
der to extract information on the neutron and thus on the
isospin dependence of the helicity structure.
2 Experimental setup
The MAMI accelerator (g. 1) with its source of polarised
electrons, based on the photoeect on a strained GaAs
crystal, routinely delivers polarised beams with a maxi-
mum energy of 855 MeV. In our rst successful data taking
period in 1998 (g. 3) we typically had a degree of polari-
sation of about 75%, in the 2nd period in the year 2003 for
the neutron runs this has been improved to about 80%. A
dedicated experimental apparatus (g. 2) including a po-
larised solid target and a detector with full angular accep-
tance was installed in the A2 Hall. The photons were pro-
duced by bremsstrahlung in the A2-Glasgow-Mainz tag-
ging facility (g. 4), which rstly determines the photon
energy by the help of 352 scintillation counters with a res-
olution of approximately 2MeV at 855MeV primary beam
energy and secondly measures the degree of polarisation
 A. Thomas: The Gerasimov-Drell-Hearn sum rule at MAMI
Fig. 3. The time dependence of the electron beam polarisation
during the rst data-taking period in 1998.
Fig. 4. The A2-Glasgow-Mainz tagging facility.
of the electrons by detecting the asymmetry in the Mller
process. This Mller polarimeter allows an online moni-
toring of the degree of electron beam polarisation. The
statistical error is of the order of 2% for a measuring time
of 4 hours (each point in g. 3 represents 4 hours of data
taking).
The energy-dependent helicity transfer to the pho-
ton can be calculated reliably [6]. In order to achieve a
high degree of photon beam polarisation we used 525MeV
and 855MeV as primary electron beam energies. Figure 5
shows the degree of circular polarisation of the outgoing
photons as function of longitudinal polarisation of the in-
coming electrons.
A new solid state frozen spin polarised tar-
get [7], developed and operated by the universities of
Bonn, Bochum, Mainz and Nagoya, was used. The tar-
163
Fig. 5. The helicity transfer from the incoming longitudinal
polarised electron to the photons.
Fig. 6. The end part of the 3 He/4 He refrigerator is equipped
with a superconduction holding coil.
get materials butanol (C4 H9 OH) or deuterated butanol
(C4 D9 OD), chemically doped with paramagnetic radicals
to allow the process of Dynamic Nuclear Polarisation ,
were cooled to a temperature of 50 mK in a 3 He/4 He di-
lution refrigerator. In this horizontal cryostat, which was
developed and constructed in the Bonn polarised target
group, a nearly full angular acceptance was achieved by
the integration of a thin internal holding coil on the ther-
mal radiation shields (see g. 6). This major step forward
in polarised target technology was possible by using the
inner thermal radiation shield as coil holder and cooling
it with liquid 3 He/4 He mixture from the still to a temper-
ature of 1.2K [8]. The momentum threshold for the out-
going particles is determined by the thickness of the coil
including a thermal radiation shield (coil holder), which
was equivalent to 0.8mm copper.
The polarising magnetic eld of 2.5T was produced by
an external superconducting solenoid. After 4 hours val-
ues for the degree of proton polarisation of 80% 85% were
reached by irradiation with microwaves of a frequency
near to the electron spin resonance (70GHz). The external
solenoid was moved on a rail system and the polarisation
was maintained in the frozen spin mode at 50 mK and
0.4 T (see g. 7).
164 The European Physical Journal A
Fig. 7. The experimental principle for the GDH frozen spin
target. The 3 He/4 He refrigerator is stationary, the detector
DAPHNE and the polarising solenoid are movable. The up-
per picture shows the set-up in the polarising mode, the lower
one in the data-taking mode. A high-precision rail system was
used to move the DAPHNE detector and the superconducting
polarising magnet.
Fig. 8. The time dependence of the target polarisation during
the rst data-taking period in 1998.
The relaxation time for the proton spins was about
200 hours. Consequently we could take data for typically
2 days before repolarizing the target, for example three
cycles of polarising the target and data taking during the
relaxation of the target can be seen in g. 8. The repo-
larizing time for the target (approximately 3 hours) was
used to rell the 400l liquid-helium buer containers for
the cryostat and the polarising magnet. The movement
of the polarising solenoid and the detector took around
15 minutes.
The cylindrical detector DAPHNE (Detecteur a
grande Acceptance pour la Physique photoNucleaire Ex-
perimentale) [9] was especially designed for handling multi
particle nal states by provision of a large solid angle (94%
Fig. 9. The DAPHNE detector.
of 4) particle identication and has a moderate e ciency
for neutral particles. DAPHNE was developed by a collab-
oration of Saclay and Pavia for the investigation of pho-
toreactions on light nuclei. As shown in g. 9, DAPHNE
surrounds the target and has a cylindrical symmetry. It
consists of a vertex detector (three multi-wire propor-
tional chambers, MWPC) for reconstructing the trajec-
tories of charged particles surrounded by a hodoscope for
their identication and energy determination. The outer
part forms a lead-aluminum-scintillator sandwich serving
as a calorimeter for the detection of decay photons and
protons. Below a primary beam energy of 700 MeV all
emitted protons are retained in the detector providing a
good energy determination. The threshold momentum for
the detection of charged particles emitted from a target
including holding coil amounts to 80 MeV/c for pions and
270 MeV/c for protons. Leptonic background in DAPHNE
is suppressed by the selection of appropriate discrimina-
tor thresholds. In forward direction further detection com-
ponents (silicon -strip detector, a Cerenkov detector, a
scintillation counter array, the ring shaped STAR detec-
tor) have been added to expand the angular acceptance.
3 Results
3.1 Results on the GDH-Integral and the forward spin
polarizibility integral on the proton
Figure 10 shows our data for the helicity dierence of
the total cross section measured at MAMI in the rst
phase of the GDH experiment [10]. The data are compared
to predictions from the partial wave analyses SAID and
MAID. The negative values of the cross section dierence
(3/2 1/2 ) in the threshold region are due to the dom-
inance of the E0+ multipole in the single + production
channel. The excitation of the -resonance (P33 (1232))
prefers the 3/2 cross section at energies around 300 MeV.
It is excited by a strong M1+ transition with only 2.5%
E1+ admixture. At higher energies the double pion pro-
duction plays an important role and is not represented in
the theoretical curves.
In g. 11 the experimental running GDH integral
(right-hand side of eq. (1)) is displayed and compared
to the model predictions. The integration starts at
 A. Thomas: The Gerasimov-Drell-Hearn sum rule at MAMI 165

800 250
3/2 - 1/2 (barn)

running GDH integral (b)

MAINZ This work
HDT (N) HDT
600 200
SAID (N)
SAID
UIM N+N+)
150 UIM
400

100
200
50
0
0

-200
-50
100 200 300 400 500 600 700 800 100 200 300 400 500 600 700 800
E (MeV) E (MeV)

Fig. 10. The total cross section dierence (3/2 1/2 ) on 1 H Fig. 11. The running GDH integral obtained in this work
is compared to previous results [11] (open circles) and to the starting at 200 MeV is compared to the model predictions.
predictions of the HDT [12], SAID [13] and UIM [14] analyses. Only statistical errors are shown.
Only statistical errors are shown.

Table 1. Measured values of the GDH integral in various pho-

ton energy intervals (see text for references).

[GeV] GDH Integral [b]

ELSA 0.8 2.9 27.5 2.0 1.2
MAMI 0.2 0.8 226 5 12
combined 0.2 - 2.9 253.5 5 12

Table 2. Measured value of the GDH integral and model pre-

dictions for the unmeasured energy intervals (see text for ref-
erences).

[GeV] GDH Integral [b]

MAID < 0.2 27.5
SAID < 0.2 28
Exp.(MAMI+ELSA) 0.2 2.9 253.5 5 12 Fig. 12. The helicity dependent total photoabsorption cross
Simula et al. > 2.9 13 section dierence in the second and third resonance region mea-
Bianchi and Thomas > 2.9 14 sured at ELSA.
combined 211.5 213
GDH sum rule value 205
in 254 b for the GDH integral between 0.2 and 2.9 GeV,
see table 1. In order to check the validity of the GDH sum
E = 200 MeV and the upper integration limit is taken rule the contribution of the missing low and high energy
as the running variable. The measured value of the GDH regions have to be added according to the existing models
integral between 200 and 800 MeV amounts to 226 (see table 2): The unitary isobar model MAID2002 [14]
5 (stat) 12 (sys) b. gives a contribution of (27.5 3) b for photon energies
Due to the 3 weighting in eq. (2), the 0 running below 0.20 GeV [17]. The Regge approaches from [18,19]
integral is almost saturated at E = 800 MeV. The value predict a negative contribution above 2.9 GeV of 14 b
of the 0 integral between 200 and 800 MeV amounts to and 13 b, respectively. The combination of our experi-
[187 8 (stat) 10 (sys)] 106 fm4 . mental results from MAMI and ELSA with these predic-
The data taken at ELSA [15,16] from 0.7 to 2.9 GeV tions yields an estimate of (211.5 213) b which within
(see g. 12), together with the previously measured data at the experimental errors of 5 (stat) 12 (sys) b is con-
MAMI, cover a broader photon energy interval, and result sistent with the GDH sum rule value of 205 b.
166 The European Physical Journal A

unpolarized

Fig. 13. Helicity dierence = 3/2 1/2 and unpolarised

total photoproduction cross section on the proton (combined
data set from MAMI and ELSA).

A summary of the present status is shown in g. 13,

where the helicity dierence = 3/2 1/2 for the total
cross section on the proton is compared to the unpolarised
cross section. As the large background of non-resonant
photoproduction is spin independent and therefore almost
disappears in , the helicity dierence delivers valuable
information to study the properties of nucleon resonances.
3.2 Results on partial reaction channels
The DAPHNE detector with forward components was ca-
pable to measure besides the total absorption cross section
also partial reaction channels (see g. 14). We have pub-
lished our results on the proton in a series of papers:
the total photoabsorption cross section for energies up
to 800MeV [11]
the single pion production channels up to a photon
energy of 450MeV [10] and [20]
the single 0 production up to a photon energy of
Fig. 14. Overview on the helicity dependent partial photoab-
sorption cross sections measured at MAMI.
multipoles which are responsible for the excitation of the
D13 (1520)-resonance.
Considering only partial waves with l 2, the total
unpolarised cross section () can then be written as (see,
for instance, [26]):
|E0+ |2 + |M1 |2 + 6|E1+ |2 + (3)
2|M1+ |2 + 6|M2 |2 + 2|E2 |2 + . . . ,
while the helicity dependent total cross section 31 =
(3/2 1/2 ), where the subscripts 3/2(1/2) correspond
to the (anti)parallel -nucleon spin conguration, is
31 |E0+ |2 |M1 |2 3|E1+ |2 + |M1+ |2 (4)
6E1+ M1+ 3|M2 | + |E2 | + 6E2 M2 + . . .
2 2

800MeV [21]
the single + production up to a photon energy of
800MeV [22]
the double pion production channels 0 + [23] and
0 0 [24]
helicity dependent production [25].
Besides the importance of this data to check the GDH
sum rule and to measure the forward spin polarizibility as
shown in the previous section, new information about the
nucleons excitation spectrum can be extracted. This fea-
ture can be understood from the multipole decomposition
of the N N total cross section. In the following we
use the pion multipole notation, where E and M denote
the electric or magnetic character of the incoming pho-
ton and the indices l describe the coupling of the pion
angular momentum l and the nucleon spin to the total an-
gular momentum J = l 1/2. In the (1232)-resonance
region the measured helicity asymmetry is sensitive to the
E1+
ratio of the multipoles EM R = M 1+
. In the second res-
onance region (500 MeV 900 MeV), where several
overlapping states are present, e.g., P11 (1440), D13 (1520),
S11 (1535), the helicity dependent observables are partic-
ularly sensitive to the behaviour of the electromagnetic
Since the unpolarised cross section of eq. (3) is given by
the sum of absolute squares of multipoles, only a few dom-
inant partial waves can be condently evaluated. On the
other hand, the sensitivity to some the weaker multipoles
is greatly enhanced by measuring the polarised cross sec-
tion 31 of eq. (4). In particular new interference terms
appear, e.g., between E1+ and M1+ , which are directly
related to the (1232) or P33 (1232) excitation in the low-
energy region or between E2 and M2 , which are directly
related to the D13 (1520) excitation.
3.3 The (1232)-resonance region
There has been an extended program in dierent labo-
ratories to increase our knowledge about the EMR ratio
for the (1232)-resonance [27,28]. The determination of
the double-polarisation observable E provides new, com-
plementary information to clarify this question. In gs. 15
and 16 we have compared our data with predictions from
the multipole analysis MAID2000 using EM R values of
2.5%, 0% and +2.5%. The data are well reproduced with
an EM R = 2.5% [20].
 A. Thomas: The Gerasimov-Drell-Hearn sum rule at MAMI 167

Ds[mb]
200
-2.5%
150 gp np+ 0%
+2.5%
GDH98
100

50

-50

-100

-150
200 300 400 500
Eg[MeV]

Fig. 15. Sensitivity of the GDH observable 3/2 1/2 for the
EMR ratio of the (1232) resonance for single + production.
The curves were produced using MAID2000.

Ds[mb]
350 gp pp0 -2.5%
0%
300 +2.5%
GDH98
250

200

150

100

50

0 Fig. 17. Angular dependency for the GDH observable 31 =

200 300 400 500
Eg[MeV]
Fig. 16. Sensitivity of the GDH observable 3/2 1/2 for the
EMR ratio of the (1232) resonance for single 0 production.
The curves were produced using MAID2000.
Due to its cylindrical symmetry the DAPHNE detec-
tor is ideally suited to measure angular distributions of the
outgoing particles, see for example gs. 17 and 18 for +
production [20]. Consequently the angular dependence of
the new double polarisation observable has delivered ur-
gently required input for partial wave analyses.
3/2 1/2 for single + production from detection threshold
to 290MeV incoming photon energy. The curves were produced
using the MAID and SAID partial wave analyses.
Combining the data from our double polarised experi-
ment with the new beam asymmetry data from Grenoble
and Yerewan will improve the knowledge on the higher,
strongly overlapping resonances. A detailed discussion can
be found in [21,25,22].
3.5 Double-pion production

3.4 The second resonance region
Since DAPHNE also has a moderate e ciency for neu-
tral particle detection we can distinguish in our data all
the contributing partial reaction channels in the 2nd reso-
nance region. Figure 19 shows our results for the single 0
production. The agreement of the MAID2000 partial-wave
analysis with our data could be improved signicantly by
changing the parametrisation of the multipoles E2 and
M2 that drive the D13 -excitation by approximately 20%.
In addition g. 20 shows our preliminary results [29,
22] for single + production at higher energies. Although
the + channel is mainly produced via intermediate
D13 (1520) excitation, a cusp structure can be observed
close to the production threshold in the 1/2 channel.
Additional channels are needed to disentangle the reso-
nances at higher nucleon excitation energies. Double pion
photoproduction is particularly important for the study
of the second resonance region, where the P11 (1440) state
with its unclear origin is located, since almost 50 % of the
total photoabsorption cross section can be attributed to
the N channels. In the MAMI B energy range up to
800MeV we are presently analysing our data for the he-
licity dierence 3/2 1/2 of the double-pion production
channel + . The results for the + 0 and 0 0 chan-
nels have already been published [23,24]. Comparing our
results with predictions from dierent theoretical mod-
els [30,31] gives a new insight into the double pion pro-
duction mechanism, specially the role of the D13 (1520),
the P11 (1440) and the (1700) are under discussion.
168 The European Physical Journal A

(d/d)3/2-(d/d)1/2 [b/sr]
5
2.5
0
-2.5
560 MeV 580 MeV 600 MeV
-5

4
2
0
-2
620 MeV 640 MeV 660 MeV
-4

4
2
0
-2
680 MeV 700 MeV 720 MeV
-4

4
2
0
-2
740 MeV 760 MeV 780 MeV
-4
0 100 0 100 0 100
*
[deg]

Fig. 19. The asymmetry 3/2 1/2 for selected energies

for single 0 production compared to the MAID2000 (black)
and SAID SP01 (grey) analyses. The dashed curve was pro-
duced using MAID2000 with a modied parametrisation for
the D13 (1520)-resonance.
Fig. 18. Angular dependency for the GDH observable 31 =
3/2 1/2 for single + production for photon energies be-
tween 300 MeV and 450 MeV. The curves were produced using
the MAID and SAID partial wave analyses.

A strong helicity dependence can be observed in our

data (see gs. 21 and 22), with a clear dominance of the
3/2 over the 1/2 cross section. This suggests a reso-
nant behaviour due to the intermediate excitation of the
D13 (1520)-resonance. However, the 1/2 cross section is
not negligible, with an indication of a resonance contribu-
tion, possibly from the P11 (1440) excitation. Future ex-
periments with the recently installed Crystal Ball detec-
tor with its high e ciency for neutral particles will deliver
further insights in the production mechanism.

3.6 photoproduction

The helicity dependence of the production has been

measured for the rst time at a center-of-mass angle of
= 70 in the proton energy range from 780MeV to
790MeV and was reported in [25].
The results shown in g. 23 demonstrate the impor-
tance of the S11 (1535) resonance for production. The
d
cross section ( d )3/2 is small as expected for S-wave domi- Fig. 20. Preliminary data for 3/2 and 1/2 for + production.
A pronounced cusp structure due to an interference with the
nance near threshold. Clearly, better statistics and a wider
production process at threshold can be clearly observed in the
kinematical range are required to disentangle the small
1/2 cross section [22].
contributions from resonances other than the S11 (1535).
 A. Thomas: The Gerasimov-Drell-Hearn sum rule at MAMI
Fig. 21. The helicity dependent cross-sections for 0 0 pho-
toproduction. The errors shown are statistical only.
Fig. 22. The helicity dependent dierential cross-sections for
+ 0 photoproduction. The errors shown are statistical only.
With the new accelerator stage MAMI C and the multi-
photon detector Crystal Ball in combination with TAPS
as forward detector the data set will be improved signi-
cantly.
3.7 Results on the GDH sum rule on the neutron
In 1998 there has been done a pilot experiment to inves-
tigate the feasibility to measure the GDH sum rule for
169
Fig. 23. The measured helicity dependent dierential cross-
sections for photoproduction. Left: 13 = (d d
)1/2
d d d
( d )3/2 . Right: ( d )1/2 and ( d )3/2 . The dierent lines show
the predictions of the MAID analysis for four dierent cases
including all resonances (continuous line), S11 (1535) only
(dashed line), without S13 (1520) (dash-dotted line), without
S11 (1650) (dotted line). The errors shown are statistical only.
Fig. 24. Preliminary helicity-dependent total cross section for
the deuteron compared to theoretical predictions from ref. [32,
33, 34].
the neutron. Since there is no free neutron target avail-
able, the Bonn frozen spin target had to be loaded with
deuterated butanol (C4 D9 OD).
First preliminary results are available for the deuteron
for the total cross section (see g. 24) and the partial
channels d pp and d nn + [35], based on
a small subset of the data.
The full data taking was carried out in the rst half of
2003 using the Bonn frozen-spin target [7]. In the course
of the experiment the degree of polarisation could be in-
creased from 35% to more than 70% due to a new target
material based on trityl-doped D-butanol [36]. This devel-
opment by the Bochum polarised target group is a major
170 The European Physical Journal A
Fig. 25. Progression of the target polarisation during the ex-
periment with the standard and the new target material (la-
beled Fin II).
Fig. 26. Crystal Ball detector with inner tracker and TAPS
forward wall.
step forward in the polarised target technology. The pro-
gression of the target polarisation measured during the ex-
periment with the two target materials is shown in g. 25.
The data analysis is in progress [37,29,38].
3.8 Conclusions and outlook
The international GDH collaboration, working success-
fully at the electron accelerators MAMI and ELSA, has
provided data to check the GDH sum rule experimentally
for the rst time. A newly developed polarised solid state
target with high angular acceptance and highly polaris-
able new target materials in combination with 4 detec-
tors and highly polarised continuous photon beams made
Fig. 27. The new dilution refrigerator for the Crystal Ball
detector.
the experiment possible. For the proton the sum rule has
been veried on a 10% level, the deuteron data are still
being analysed. It is a challenge that requires a tremen-
dous theoretical eort to extract the neutron properties
from the deuteron data.
The investigation of the partial reaction channels led to
an additional knowledge on the helicity dependence of the
nucleons excitation spectrum. The double-polarisation
observable E has been determined in a broad kinemati-
cal range.
With the upgrade of the MAMI accelerator to 1.5 GeV
a signicant part of the outgoing particles would escape
from the DAPHNE detector. The new standard detector
to be used with the tagged photon beam, the Crystal Ball
(CB) detector completed by TAPS as a forward wall, has
already taken data in the years 2004 and 2005 with unpo-
larised targets, and both unpolarised and polarised photon
beams. The CB consists of 672 optically isolated NaI(Tl)
crystals, 15.7 radiation lengths thick. The counters are ar-
ranged in a spherical shell with an inner radius of 25.3 cm
and an outer radius of 66.0 cm. Charged particles can be
measured by the central tracker consisting of a scintillator
barrel and modied DAPHNE cylindrical multi-wire pro-
portional chambers (two layers only). In order to achieve
a better particle identication for charged particles, an in-
ner barrel of plastic scintillators (PID-detector) has been
included. The TAPS wall is composed of 522 BaF2 detec-
tors [39] arranged in a hexagon. The apparatus is schemat-
ically shown in g. 26. The high granularity, large accep-
tance and good energy resolution make this setup a unique
instrument for the detection of multi-photon nal states.
The tagger will be upgraded to cope with the in-
creased beam energy. Linearly and circularly polarised
photons will be available. A polarised frozen-spin target is
presently under development to allow for double polarisa-
tion experiments with the new facilities. The central part
of this target will be a horizontal dilution refrigerator (see
g. 27).
The data presented in this paper have been mostly pro-
duced by the international GDH collaboration. We gratefully
 A. Thomas: The Gerasimov-Drell-Hearn sum rule at MAMI
acknowledge the excellent support of the MAMI and ELSA
accelerator groups. This work was supported by the Deutsche
Forschungsgemeinschaft (SFB 201, SFB 443, Schwerpunkt-
programm 1034, and GRK683), the INFN-Italy, the FWO
Vlaanderen-Belgium, the IWT-Belgium, the UK Engineering
and Physical Science Council, the DAAD, JSPS Research Fel-
lowship, and the Grant-in-Aid (Specially Promoted Research)
in Monbusho, Japan.
References
1. S.B. Gerasimov, Yad. Fiz. 2, 598 (1965) (Sov. J. Nucl.
Phys. 2, 430 (1966)).
2. S.D. Drell, A.C. Hearn, Phys. Rev. Lett. 16, 908 (1966).
3. I. Karliner, Phys. Rev. D 7, 2717 (1973).
4. A. Sandor, Phys. Rev. D 50, R6681 (1994).
5. Anselmino et al., Sov. J. Nucl. Phys. 49, 553 (1989).
6. H. Olsen, L.C. Maximon, Phys. Rev. 114, 887 (1959).
7. C. Bradtke et al., Nucl. Instrum. Methods A 436, 430
(1999).
8. H. Dutz et al., Nucl. Instrum. Methods A 340, 272 (1994).
9. G. Audit et al., Nucl. Instrum. Methods A 301, 473 (1991).
10. J. Ahrens et al., Phys. Rev. Lett. 87, 022003 (2001).
11. J. Ahrens et al., Phys. Rev. Lett. 84, 5950 (2000).
12. O. Hansein et al., Nucl. Phys. A 632, 561 (1998).
13. R.A. Arndt et al., Phys. Rev. C 66, 055213 (2002).
14. D. Drechsel et al., Nucl. Phys. A 645, 145 (1999).
15. H. Dutz et al., Phys. Rev. Lett. 91, 192001 (2003).
16. H. Dutz et al., Phys. Rev. Lett. 93, 032003 (2004).
171
17. L. Tiator, Proceedings of GDH2002 (World Scientic, Sin-
gapore, 2003).
18. N. Bianchi, E. Thomas, Phys. Lett. B 450, 439 (1999).
19. S. Simula et al., Phys. Rev. D 65, 034017 (2002); private
communication.
20. J. Ahrens et al., Eur. Phys. J. A 21, 323 (2004).
21. J. Ahrens et al., Phys. Rev. Lett. 88, 232002 (2002).
22. J. Ahrens et al., submitted to Phys. Lett. C.
23. J. Ahrens et al., Phys. Lett. B 551, 49 (2003).
24. J. Ahrens et al., Phys. Lett. B 624, 173 (2005).
25. J. Ahrens et al., Eur. Phys. J. A 17, 241 (2003).
26. D. Drechsel, Prog. Part. Nucl. Phys. 34, 181 (1995).
27. R. Beck, H.P. Krahn et al., Phys. Rev. Lett. 78 (1997).
28. G.S. Blanpied et al., Phys. Rev. Lett. 79, 4337 (1997).
29. T. Rostomyan, PhD Thesis (Gent) (2005).
30. J. Nacher, Proceedings of NSTAR2001 (World Scientic
Pub. Co, 2001) p. 189.
31. M. Vanderhaeghen, H. Holvoet, private communication,
Mainz 2001.
32. H. Arenhovel, The GDH for the deuteron, in Proceedings of
GDH2000, Mainz, edited by D. Drechsel, L. Tiator (World
Scientic, Singapore, 2001) p. 67.
33. A. Fix, private communication.
34. M. Schwamb, private communication.
35. M. Martinez, PhD Thesis, University of Mainz (in prepa-
ration).
36. St. Goertz et al., Nucl. Instrum. Methods A 526, 43
(2004).
37. O. Jahn, PhD Thesis, University of Mainz (2005).
38. S. McGee, PhD Thesis, Duke University (in preparation).
39. R. Novotny, IEEE Trans. Nucl. Sci. 38, 379 (1991).
Eur. Phys. J. A 28, s01, 173 183 (2006)
DOI: 10.1140/epja/i2006-09-018-1 EPJ A direct
electronic only

Experiments with photons at MAMI

R. Becka
Helmholtz-Institut fur Kern- und Strahlenphysik, Nussallee 14-16, 53115 Bonn, Germany

/
Published online: 26 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. A very successful experimental program with real photons has been achieved in 20 years of
operation at the Mainz Microtron (MAMI) facility. The dierent detector setups, like DAPHNE, TAPS
and the Crystal Ball are centered around the tagged photon facility the so-called Glasgow Tagger. From
the rich spectrum of results only a few highlights will be discussed here, the proton polarizabilities, the pion
polarizabilities, pion photoproduction close to the pion threshold and in the (1232)-resonance region.

PACS. 13.40.-f Electromagnetic processes and properties 13.60.Le Meson production 13.60.Fz Elastic
and Compton scattering 14.20.Dh Protons and neutrons

1 Introduction unique relation between a low-energy expansion of the

Experiments with real photons at MAMI have been per-
formed in the framework of the A2-collaboration. Mono-
chromatic photons from bremsstrahlung tagging by the
Glasgow Tagger [1,2] are used for all experiments. Polar-
ized photon beams, linear and circular, are available as
well as polarized targets.
Groups from several institutions and countries (see [3])
have provided dierent detector components for example
the photon spectrometer TAPS [4], the 4 charge parti-
cle tracking detector DAPHNE [5] and more recent the
photon spectrometer Crystal Ball [6]. Many data have
been taken on the proton and on light and complex nuclei
including the total photon absorption, Compton scatter-
ing, meson production, break up reactions and multi pion
production in the nal state. The experimental work was
based on 175 Diploma and PhD thesis, which are pub-
lished in more than 100 refereed articles. Here only a few
highlights of this experimental program can be adressed.
2 Low-energy Compton scattering
Next to the size and the anomalous magnetic moment,
the polarizability is a further property of a particle with a
substructure. In the present of the electromagnetic elds
(E and B), electric dipole moments are induced and mag-
netic dipole moments may be oriented (paramagnetism)
or induced according to Lenzs rule (diamagnetism). The
most precise determination of the proton polarizabilities
comes from Compton scattering experiments. These mea-
surements rely on a Low-Energy Theorem to establish a
a
e-mail: beck@hiskp.uni-bonn.de
Compton-scattering cross section and the static polariz-
abilities and . For photon energies small compared to
pion mass, this expansion reads [7,8]:
   
d d
=
d d
   2 Point
e +
 (1 + z)2 + (1 z)2 (1)
m 2 2
with z = cos( ), where and  are the energies of
d
the incident and scattered photon, respectively; d Point
is the exact cross section for a structureless proton with
an anomalous magnetic moment. The quantities and
are the static polarizabilities.
Low-energy Compton scattering from the proton in
the energy range from 55 MeV to 165 MeV was measured
using the TAPS detector set up at the photon beam at
MAMI. The energy of the incident electron beam was
chosen to be 180 MeV. The target consisted of a Kapton
cylinder of 20 cm length lled with liquid hydrogen. Data
obtained from about 200 h of beam time were analysed [9,
10]. The scattered photons were detected with 6 blocks
of the TAPS. Since the recoiling protons could not be
detected, a single-particle trigger had to be used. There-
fore, this minimum bias trigger included all kinds of back-
ground events for example cosmic ray events which have
not been suppressed by an active shield and electromag-
netic background from the beam collimation system and
from the target itself. These sources of background were
partially suppressed by time cuts and a missing-energy
cut, which is dened as the dierence between the mea-
sured incident photon energy (tagger) and the expected
incident photon, as calculated from the measured scat-
tered photon assuming Compton kinematics.
174 The European Physical Journal A
40
lab=59 lab=85
35

30
/(nb/sr)

25

20
lab
d /d

15

10

0
40
lab=107 lab=133
35

30
/(nb/sr)

25

20
lab
d /d

15

10

0
0 40 80 120 160 200
40
E /MeV
lab=155
35

30
/(nb/sr)

Fig. 2. Error contour plot in the ( )-plane for which only

25 bestt Lvov
Lvov = the statistical errors are taken. The contours correspond to
20
the values 2min + 1 of the individual ts. Also shown are the
lab

V. Olmos et al.
sum rule constraint and the value as follows from the
d /d

15

10 experiment by Zieger et al. [12]. The thick solid line shows the
5 result of the global t, eq. (4).
0
0 40 80 120 160 200
E /MeV
rule constraint leads to the following result:
Fig. 1. Measusred dierential cross-sections in the lab sys- = 12.1 0.3stat. 0.4syst. 0.3mod. ,
tem [9] compared with a dispersion relation calculation (solid (4)
line) [11]. = 1.6 0.4stat. 0.4syst. 0.4mod. ,

The dierential cross sections obtained are plotted in
g. 1. The systematic errors of 3% arise from uncer-
tainties in the photon ux (2%) and the target density
(2%) combined in quadrature. The eective solid
angles were determined with Monte Carlo simulations.
Errors from uncertainties in the experiment geometry are
estimated to be 5%. With the help of the dispersion
relation approach the electromagnetic polarizabilities
of the proton can be extracted from the experimental
cross-sections. The procedure used in the analysis was
to take and as free parameters, and sometimes as
well the constraint given by the Baldin sum rule. Using
standard 2 minimization, the result obtained, when
tting the MAMI/TAPS data alone without the sum rule
constraint, is
= 11.9 0.5(stat.) 1.3(syst.),
= 1.2 0.7(stat.) 0.3(syst.).
The Baldin sum rule obtained from this result,
+ = 13.1 0.9 is in agreement with the value
determined by the total photon absorption cross section.
A t to the existing low-energy Compton scattering
data including the new MAMI/TAPS data and the sum
where the rst error denotes the statistical, the second the
systematic and the third the model-dependent one. The
results are summarized in g. 2 (contourplot) where the
contours in the ( ) plane for 2min + 1 are plotted. In
addition, the Baldin sum and the value obtained from the
experiment by Zieger [12] are included.
3 Pion polarizability
The pion polarizabilities characterize the dynamical defor-
mation of the pion in the electromagnetic eld. The values
of the electric and magnetic pion polarizabilities de-
pend on the rigidity as a composite particle and provide
important information of internal structure. Very dierent
values for the pion polarizabilities have been calculated in
the past. All predictions agree, however, that the sum of
(2) the two polarizabilities of the meson is very small.
(3) On the other hand, the values of the dierence of the po-
larizabilities are very sensitive to theoretical models. For
example, investigations within the framework of the chiral
perturbation theory (ChP T ) predict ( ) 5.4 [13]
in one-loop calculations and 4.4 1.0 for two-loops [14]
(all values of the polarizabilities are given in units of
104 fm3 ). The calculations in the extended Nambu-Jona
 R. Beck: Experiments with photons at MAMI 175

A
B A

MWPC + FSD
-beam

LH2
C
B
TOF

1m MWPC+FSD
LH2

Fig. 3. Floor plan of the experimental setup showing the loca- Fig. 4. Enlarged view showing the details of the TAPS con-
tion of the detectors. A, B, C are TAPS blocks, MWPC+FSD guration.
show multi-wire proportional chambers and the forward scin-
tillation detector, TOF indicates the block of the neutron de-
tector bars, and LH2 stands for the liquid-hydrogen target in
The experiment discussed here has been performed at
its vacuum scattering chamber.
the continuous-wave electron accelerator MAMI B [25,
26] using Glasgow-Edinburgh-Mainz tagger photon facil-
Lasinio model with linear realization of chiral U (3) U (3) ity [1,2]. The quasi-monochromatic photon beam covered
symmetry [15] result in = = 3.0 0.6. The the energy range from 537 to 819 MeV with an intensity
application of dispersion sum rules (DSR) at xed value 6 105 /s in the tagger channel for the lowest pho-
of the Mandelstam variable u = 2 for calculation of this ton energy and average energy resolution of 2 MeV. The
parameter [16,17] leads to ( ) = 10.3 1.9. DSR at tagged photons entered a scattering chamber, containing
nite energy [8] gave the similar result: ( ) = 10.6. a 3 cm diameter and 11.4 cm long liquid hydrogen target
A calculation in the linear model with quarks and vector with Capton windows. The emitted photon  , + me-
mesons included to one loop order predicted ( ) = son, and the neutron were detected in coincidence. The
20 [18]. An evaluation in the Dubna quark connement experimental setup is shown in g. 3.
model [19] results in ( ) = 7.05. The photons were detected by the spectrometer
Because there is no stable pion target, experimental TAPS [4], assembled in a special conguration (g. 4). The
information about the pion polarizabilities is not easy to TAPS spectrometer consists of 528 BaF2 crystals. Each
obtain. One has to investigate reaction channels, like scat- hexagonally shaped crystal is 250 mm long corresponding
tering high energy pion in the Coulomb eld of a heavy to 12 radiation lengths. All crystals were arranged into
nuclei or the radiative pion photoproduction. three big blocks. Two blocks (A, B) consisted of 192 crys-
The scattering of high energy pions o the Coulomb tals arranged in 11 columns and the third block (C) had
eld of heavy nuclei [20] has resulted in = = 144 crystals arranged in 11 columns. These three blocks
6.8 1.4 1.2. This value agrees with prediction of the were located in the horizontal plane around the target
dispersion sum rules but is about 2.5 times larger than the at angles 68 , 124 , 180 with respect to the beam axis.
ChP T result. The experiment of the Lebedev Institute on Their distances to the target center were 55 cm, 50 cm and
radiative pion photoproduction from the proton [21] has 55 cm, respectively. All BaF2 modules were equipped with
given + = 20 12. This value has large error bars and 5 mm thick plastic veto detectors for the identication of
shows the largest discrepancy with regard to the ChP T charged particles.
predictions. The attempts to determine the polarizability The neutrons were detected by a wide aperture time-
from the reaction suer greatly from theoreti- of-ight spectrometer (TOF) [27]. It consisted of 111 scin-
cal [22] and experimental [23] uncertainties. The most re- tillation detector bars of 50 200 3000 mm3 and 16
cent analysis of MARK II and Crystal Ball data [24] nds counters (10 230 3000 mm3 ) which were used as veto
no evidence for a violation of the ChP T predictions. How- detectors. The bars are made from NE110 plastic scin-
ever, even changes of polarizabilities by 100% and more tilator and each bar is read out on both ends by two 3
are still compatible with the present error bars. phototubes XP2312B. All bars were assembled in 8 planes
176
d/ds1dt (nb/ )
1.2
4
1 12
0.8
0.6
0.4
0.2
0
-10 -9 -8 -7 -6 -5 -4 -3
Fig. 5. The dierential cross section of the process p + n
averaged over the full photon beam energy interval and over s1
from 1.5m2 to 5m2 . The solid and dashed lines are the predic-
tions of model-1 and model-2, respectively, for ( )+ = 0.
The dotted line is a t to the experimental data (see text).
of a special conguration with 16 detectors in each, fol-
lowing one after another (g. 3). Such a neutron detector
allows to detect the neutrons in the energy region 10
100 MeV with e ciency 30 50% and to determine their
energy with a resolution 10% using the neutron time
of ight and the angle of the neutron emission measured
with a precision 2 3%.
To detect the + meson two two-coordinate multi-
wire proportional chambers (MWPC) and a forward scin-
tillator detector (FSD), for getting a fast trigger signal,
have been developed and constructed. The MWPCs over-
lap angles in the laboratory system were = 2 20 ,
= 0
360
and were located under 0
with respect to
the beam direction.
The cross section of the process p + n has been
calculated in the framework of two dierent models. In
the rst model (model-1) the contribution of all the pion
and nucleon pole diagrams is taken into account using
pseudoscalar pion-nucleon coupling [28].
In the second model (model-2), the nucleon and the
pion pole diagrams without the anomalous magnetic mo-
ments of the nucleons, and in addition the contributions
of the resonances (1232), P11 (1440), D13 (1520), and
S11 (1535) are included.
To control the model dependence of the result the kine-
matic regions were limited to regions where the dierence
between model-1 and model-2 does not exceed 3% when
( )+ is constrained to zero. First, a kinematic region
where the contribution of the pion polarizability is negligi-
ble, i.e. the region 1.5m2 s1 < 5m2 was analysed, where
s1 is the squared pion-photon center-of-mass energy.
In g. 5, the experimental data for the dierential cross
section, averaged over the full photon beam energy in-
terval from 537 MeV up to 817 MeV and over s1 in the
indicated interval, are compared to predictions of model-
1 (dashed curve) and model-2 (solid curve). The dotted
curve is the t of the experimental data in the region of
10m2 < t < 2m2 , where t is the squared pion mo-
The European Physical Journal A
(nb)
14
10
8
6
4
2
-2 -1 0
2 0 550 600 650 700 750 800
t/
E (MeV)
Fig. 6. The cross section of the process p + n integrated
over s1 and t in the region where the contribution of the pion
polarizability is biggest and the dierence between the predic-
tions of the theoretical models under consideration does not
exceed 3%. The dashed and dashed-dotted lines are predic-
tions of model-1 and the solid and dotted lines of model-2 for
( )+ = 0 and 14 10 4 fm3 , respectively.
mentum transfer. As seen from this gure, the theoretical
curves are very close to the experimental data. This means
that the dependence of the dierential cross section on the
square of the four-momentum transfer t which is basically
the kinetic energy of the neutron is well reproduced by us-
ing the mentioned GEANT simulations for the e ciency.
In a second step, the kinematic region where the po-
larizability contribution is maximal was investigated. This
is the region 5m2 s1 < 15m2 and 12m2 < t < 2m2 .
In the considered region of the phase space, the cross sec-
tions of the process p + n integrated over s1 and t
are calculated according to model-1 and model-2 for two
dierent values of (). The obtained experimental cross
sections and their theoretical predictions for ()+ = 0
and 14 104 fm3 are presented in g. 6. The error bars
are the quadratic sum of statistical and systematic errors.
For each model, we obtain
( )+ = (12.2 1.6stat 3.3syst ) 104 fm3
(model 1), (5)
( )+ = (11.1 1.4stat 2.8syst ) 104 fm3
(model 2). (6)
Averaging over the results of the two models, the nal
result is obtained [29]:
( )+ = (11.6 1.5stat 3.0syst 0.5mod ) 104 fm3 .
(7)
4 Pion photoproduction in the threshold
region
The photoproduction of pions near threshold has been a
topic of considerable experimental and theoretical activi-
ties over the past years, ever since the results of the ex-
periments, performed in Saclay [30], Mainz ([31,32]) and
 R. Beck: Experiments with photons at MAMI
Saskatoon [33], were at variance with the prediction of
a low energy theorem (LET), which was derived in the
early 70s [34,35]. Being based on fundamental principles,
this LET predicted the value of the S-wave threshold am-
plitude E0+ in a power series in = m /mN , the ratio of
the masses of the pion and nucleon.
The discrepancy could be explained by a calculation in
the framework of heavy-baryon chiral perturbation theory
(ChPT) [36], which showed that additional contributions
due to pion loops in 2 have to be added to the old LET.
Rened calculations within heavy-baryon ChPT [37] led
to descriptions of the four relevant amplitudes at threshold
by well-dened expansions up to order p4 in the S-wave
amplitude E0+ and p3 in the P -wave combinations P1 ,
P2 and P3 , where p denotes any small momentum or pion
mass, the expansion parameters in heavy-baryon ChPT.
To that order, three low-energy constants (LEC) due to
the renormalization counter terms appear, two in the ex-
pansion of E0+ and an additional LEC bP for P3 , which
have to be tted to the data or estimated by resonance
saturation.
However, two combinations of the P -wave amplitudes,
P1 and P2 , are free of low-energy constants. Their expan-
sions in converge rather well leading to new LETs for
these combinations. Therefore, the P -wave LETs oer a
signicant test of heavy-baryon ChPT. However, for this
test the S-wave amplitude E0+ and the three P -wave com-
binations P1 , P2 and P3 have to be separated. This separa-
tion can be achieved by measuring the photon asymmetry
using linearly polarized photons, in addition to the mea-
surement of the total and dierential cross sections.
The dierential cross sections can be expressed in
terms of the S- and P -wave multipoles, assuming that
close to threshold neutral pions are only produced with
angular momenta l of zero and one. Due to parity and
angular momentum conservation only the S-wave ampli-
tude E0+ (l = 0) and the P -wave amplitudes M1+ ,
M1 and E1+ (l = 1) can contribute and it is conve-
nient to write the dierential cross section and the pho-
ton asymmetry in terms of the three P -wave combinations
P1 = 3E1+ + M1+ M1 , P2 = 3E1+ M1+ + M1 and
P3 = 2M1+ + M1 . The c.m. dierential cross section is
d() q
= (A + B cos() + C cos2 ()), (8)
d k
where is the c.m. polar angle of the pion with re-
spect to the beam direction and q and k denote the
c.m. momenta of pion and photon, respectively. The co-
e cients A = |E0+ |2 + |P23 |2 , B = 2 Re(E0+ P1 ) and
C = |P1 |2 |P23 |2 are functions of the multipole ampli-
tudes with P232
= 12 (P22 + P32 ). Earlier measurements of
the total and dierential cross sections already allowed
determination of E0+ , P1 and the combination P23 .
In order to obtain E0+ and all three P -waves sepa-
rately and to test the new LETs of ChPT, it is necessary
to measure, in addition to the cross sections, the photon
asymmetry ,
d d
= , (9)
d + d
177
4.5 this work
Ref. [3]
4.0
Ref. [4]
3.5
3.0
/ b
2.5
2.0
1.5
1.0
0.5
0.0
144 147 150 153 156 159 162 165 168
E / MeV
Fig. 7. Total cross sections for 0 photoproduction close to
threshold with statistical errors (without systematic error of
5%) as function of incident photon energy (solid squares, this
work ref. [38], open circles, ref. [33], open diamonds ref. [32]).
where d and d are the dierential cross sections for
photon polarizations perpendicular and parallel to the re-
action plane dened by the pion and proton. The asym-
metry is proportional to the dierence of the squares of
P3 and P2 :
q d()
() = (P32 P22 ) sin2 ()/ . (10)
2k d
A measurement of the reaction p( , 0 )p [39] was per-
formed at the Mainz Microtron MAMI [40] using the
Glasgow/Mainz tagged photon facility [1,2] and the pho-
ton spectrometer TAPS [4]. The MAMI accelerator de-
livered a continuous wave beam of 405 MeV electrons.
Linearly polarized photons were produced via coherent
bremsstrahlung in a 100 m thick diamond radiator [41,
42] with degrees of polarization of up to 50%. The neu-
tral pion decay photons were detected in TAPS [43], an
array of 504 BaF2 detectors, which was built up around a
liquid-hydrogen target.
The total and dierential cross sections were measured
over the energy range from 0 threshold to 168 MeV. Fig-
ure 7 shows the results for the total cross section in com-
parison to ref. [33] and [32]. The results for the photon
asymmetry are shown in g. 8 in comparison to the values
of ChPT [37] and to a prediction of a dispersion theo-
retical calculation (DR) by Hanstein, Drechsel and Tia-
tor [44]. The photon asymmetry was determined from all
the data between threshold and 166 MeV for which the
mean energy was 159.5 MeV. The theoretical predictions
are shown for the same energy.
The values for the real and imaginary part of E0+
and the three P -wave combinations were extracted via
two multipole ts to the cross sections and the photon
asymmetry simultaneously. The two multipole ts dier
in the energy dependence of the real parts of the P -wave
combinations. For the rst t the usual assumption of a
behaviour proportional to the product of q and k was
adopted (qk-t, 2 /dof = 1.28). The assumption made
178 The European Physical Journal A

Table 1. Results of both ts (qk-t and q-t) for Re E0+ at the 0 - and + -threshold (unit: 10 3 /m+ ), for the parameter
of Im E0+ (unit: 10 3 /m2+ ) and for the three combinations of the P -wave amplitudes (unit: q 10 3 /m2+ ) with statistical and
systematic errors in comparison to the predictions of ChPT [37, 45] (O(p3 )) and of a dispersion theoretical approach (DR, [44]).
This work ChPT DRa
qk-ta q-t
0
p
E0+ (Ethr ) 1.23 0.08 0.03 1.33 0.08 0.03 1.16 1.22
n +
E0+ (Ethr ) 0.45 0.07 0.02 0.45 0.06 0.02 0.43 0.56
2.43 0.28 1.0 5.2 0.2 1.0 2.78 3.6
P1 9.46 0.05 0.28 9.47 0.08 0.29 9.14 0.5 9.55
P2 9.5 0.09 0.28 9.46 0.1 0.29 9.7 0.5 10.37
P3 11.32 0.11 0.34 11.48 0.06 0.35 10.36 9.27
P23 10.45 0.07 10.52 0.06 11.07 9.84
a
Values of the P -wave combinations converted into the unit q 103 /m2+ .

0.4 0.0
this work: q k-t ChPT [8]
0.3 -0.2 this work: q-t DR [17]

0.2 -0.4
+
ReE0+ / 10 /m
-3

0.1 -0.6

0.0 -0.8

-0.1 -1.0
ChPT [8] +
-0.2 -1.2 n
DR [17] Ethr =151.4 MeV
this work t to the data
-0.3 -1.4
0 20 40 60 80 100 120 140 160 180 145 150 155 160 165
cms
0 / deg E / MeV

Fig. 8. Photon asymmetry for 0 photoproduction at 159.5
MeV photon energy with statistical errors (without systematic
error of 3%) as a function of the polar angle (solid line: t
to the data) in comparison to ChPT [37] (dotted line) and
DR [44] (dashed line).
Fig. 9. Results for Re E0+ with statistical errors as a function
of incident photon energy E for an assumed energy depen-
dence of the P -wave amplitudes proportional to q k (solid
squares) and q (open squares) in comparison to ChPT [37]
(dotted line) and DR [44] (dashed line).

to the new MAMI data presented in this letter, shows,

for the second t is an energy dependence of the P -wave
amplitudes proportional to q (q-t, 2 /dof = 1.29). This
is the dependence which ChPT predicts for the P -wave
amplitudes in the near-threshold region, but at higher en-
ergies the prediction is in between the q and qk energy
dependence.
The results of both multipole ts for Re E0+ as a func-
tion of the incident photon energy are shown in g. 9 and
compared with the predictions of ChPT and of DR. The
results for the threshold values of Re E0+ (at the 0 - and
+ -threshold), for the parameter of ImE0+ and for the
values of the threshold slopes of the three P -wave com-
binations of the qk-t and the q-t are summarized in
table 1, for more details see [38].
For both ts the low-energy theorems of ChPT (O(p3 ))
for P1 and P2 agree with the measured experimental re-
sults within their systematic and statistical errors. The ex-
perimental value for P3 is higher than the value of ChPT,
which can be explained by the smaller total and dier-
ential cross sections of ref. [32], used by ChPT to deter-
mine the dominant low-energy constant bP for this multi-
pole [45]. A new fourth-order calculation in heavy-baryon
ChPT by Bernard et al., introduced in [46] and compared
that the potentially large -isobar contributions are can-
celled by the fourth-order loop corrections to the P -wave
low-energy theorems. This gives condence in the third-
order LET predictions for P1 and P2 , which are in agree-
ment with the present MAMI data. With the new value of
bP [46], tted to the present MAMI data, the ChPT cal-
culation is in agreement with the measured photon asym-
metry.
In a recent work, pion photoproduction on the nu-
cleon is evaluated by dispersion relation at constant t [47].
The extension to the unphysical region provides a unique
framework to determine the low-energy constants of chiral
perturbation theory by global properties of the excitation
spectrum. See also the most recent work for pion produc-
tion at threshold in the framework of covariant baryon
chiral perturbation theory [48].
5 The N (1232) transition and the
E2/M1 ratio
Low-energy electromagnetic properties of baryons, such as
mass, charge radius, magnetic and quadrupole moments
 R. Beck: Experiments with photons at MAMI 179

are important observables for any model of the nucleon 0.8

E =280 MeV E =300 MeV
structure. In various constituent-quark models a tensor
0.6
force in the inter-quark hyperne interaction, introduced
rst by de Rujula, Georgi and Glashow [49], leads to 0.4
a d-state admixture in the baryon ground-state wave-
0.2
function. As a result the tensor force induces a small vio-
lation of the Becchi-Morpurgo selection rule [50], that the 0.0
MAMI / TAPS
N (1232) excitation is a pure M 1 (magnetic dipole) MAMI / DAPHNE
-0.2
transition, by introducing a non-vanishing E2 (electric LEGS (E =275 MeV)
(E =298 MeV)
Hanstein (TAPS)
quadrupole) amplitude. For chiral quark models or in the -0.4
Skyrmion picture of the nucleon, the main contribution to 0.8 E =320 MeV E =340 MeV
the E2 strength stems from tensor correlations between
0.6
the pion cloud and the quark bag, or meson exchange cur-
rents between the quarks. To observe a static deformation 0.4
(d-state admixture) a target with a spin of at least 3/2 0.2
(e.g. matter) is required. The only realistic alternative
is to measure the transition E2 moment in the N 0.0
3/2
transition at resonance, or equivalently the E1+ partial -0.2
wave amplitude in the N decay. The experimental -0.4
(E =322 MeV)

quantity of interest to compare with the dierent nucleon

3/2 3/2 0.8 E =360 MeV E =380 MeV
models is the ratio REM = E2/M 1 = E1+ /M1+ of the
electric quadrupole E2 to the magnetic dipole M 1 am- 0.6
plitude in the region of the (1232)-resonance. In quark 0.4
models with SU (6) symmetry, for example the MIT bag
0.2
model, REM = 0 is predicted. Depending on the size
of the hyperne interaction and the bag radius, broken 0.0
SU (6) symmetry leads to 2% < REM < 0 [51,52,53,54]. -0.2
Larger negative values in the range 6% < REM < 2.5%
have been predicted by Skyrme models [55] while re- -0.4
0 40 80 120 160 0 40 80 120 160
sults from chiral bag models [56] give values in the range CMS
0 (Grad)
CMS
0 (Grad)
2% to 3%. The rst Lattice QCD result is REM =
Fig. 10. Photon asymmetries in the -resonance region
(+3 9)% [57] and a quark model with exchange currents
(solid circles, this work ref. [64], open diamonds ref. [59] and
yields values of about 3.5% [58]. crosses ref. [60]).

The determination of the quadrupole strength E2 in

the region of the (1232) resonance has been the aim of
a considerable number of experiments and theoretical ac-
tivities in the last few years. Experimental results have
been published for the dierential cross section and pho-
ton asymmetry of pion photoproduction o the proton
from the Mainz Microtron MAMI and the laser backscat-
tering facility LEGS at Brookhaven National Laboratory,
with the results REM = (2.5 0.2stat 0.2sys )% from
the Mainz group [59] and REM = (3.0 0.3stat+sys
0.2mod )% from the LEGS group [60]. These new REM
results have started intense discussions about the cor-
rect way to extract the E2/M 1 ratio from the new ex-
perimental data. In particular the large variation in the
REM values obtained in theoretical analysis of these
data at RPI [61] (REM = (3.2 0.25)%), VPI [62]
(REM = (1.5 0.5)%) and Mainz [63] (REM = (2.5
0.1)%) was quite unsatisfactory. Since small dierences
in the dierential cross section occur in the mentioned
MAMI/DAPHNE and LEGS experiments, a new experi-
ment on neutral pion photoproduction o the proton has
been performed at the Mainz Microtron covering the full
polar angle range of the pion. The new enlarged set of ex-
perimental results should allow a determination of REM
more accurately.
Figure 10 shows the new results for the photon asym-
metry for six dierent energies in the -resonance region.
For the rst time this new experiment delivers data in the
full polar angle range. The new results are in good agree-
ment with the experimental data of MAMI/DAPHNE and
LEGS. In addition, the photon asymmetries of all three
experiments are compared to the dispersion theoretical
analysis of Hanstein [63,65] and good agreement is found.
The unpolarized dierential cross sections for the same
six photon energies in the -resonance region are shown
in g. 11. The new results are in agreement with the
MAMI/DAPHNE, the LEGS data dier not only in the
absolute values of the dierential cross section but show
as well a dierent angular distribution. In addition, the re-
sults of the Hanstein analysis for the MAMI/TAPS data
are shown.
In the angular momentum expansion of the neutral
pion photoproduction it is su cient to take into account
s- and p-waves, i.e. l = 0 or 1 only. The angular distribu-
tions for the unpolarized cross section d0 /d, the paral-
lel part d /d (pion detected in the plane dened by the
photon polarization and the photon momentum vector),
and perpendicular part d /d can be expressed in the
180 The European Physical Journal A

0.06
MAMI / TAPS (E =298 MeV) 3/2 3/2
35.0 REM=(Im E1+ / Im M1+ ):
MAMI / DAPHNE 0.04 0 +
Multipolanalysis (p -, n -data)
30.0 LEGS (E =280 MeV)
/ ( b/sr)

0.02 Hanstein(DR)
Hanstein (TAPS)
25.0
0.0
20.0

R
-0.02
15.0
d /d

10.0 -0.04

5.0 -0.06 0
E =280 MeV E =300 MeV R: p -Data only
0.0 -0.08 R=REM MAMI/TAPS
40.0 Hanstein(DR)
LEGS (E =322 MeV) -0.1
35.0 250 300 350 400
E / MeV
/ ( b/sr)

30.0
25.0 3/2 3/2
Fig. 12. The energy dependence of the ratio E1+ /M1+ is
20.0 shown as solid diamonds. In addition, the energy dependence
d /d

15.0 of R = C /(12A )is shown as solid squares.

10.0
5.0
E =320 MeV E =340 MeV 3/2 3/2
0.0 can be identied with the ratio REM = E1+ /M1+ at the
30.0 (1232)-resonance (33 = 90 )

Im E1+ 
25.0 3/2
/ ( b/sr)

20.0 R  REM =  . (16)

Im M 
3/2
1+ W =M
15.0
d /d

10.0 This is the crucial point of our analysis [66]. This method
oers the advantage of being independent of absolute nor-
5.0
malization and insensitive to many systematic errors, be-
E =360 MeV E =380 MeV
0.0 cause REM is extracted from the ratio of the coe cients
0 40 80 120 160 0 40 80 120 160 C and A tted to the angular distribution of d /d.
CMS CMS
0 (Grad) 0 (Grad) Further, the following identity can be derived [64]:
Fig. 11. Dierential cross sections in the -resonance region.

MAMI/TAPS results are shown with statistical (1 2 %) and 1 C 1 C
A + ( = 90 )
systematic errors (solid circles, this work ref. [64], open dia- R= =
REM , (17)
12 A 12 1 ( = 90 )
monds ref. [59] and crosses ref. [60]).
which depends only on the shape (C/A) of the dieren-
tial cross section d/d and the photon asymmetry
s- and p-wave approximation by the parameterization at CM S = 90 . Using eq. (17), the ratio REM can be
extracted [64]:
dj () q
= (Aj + Bj cos() + Cj cos2 ()), (11) REM = (2.4 0.16stat. 0.24sys ).% (18)
d k
3/2
where q and k denote the center-of-mass momenta of the According to the Fermi-Watson theorem the E1+ and
3/2
pion and the photon, respectively, and j indicates the par- M1+ partial waves have the same phase 33 and the ratio
allel (), perpendicular () and unpolarized (0) compo- 3/2 3/2
E1+ /M1+ is a real quantity. As shown in g. 12, this ratio
nents. The coe cients Aj , Bj and Cj are quadratic or is strongly dependent on the photon energy and varies
bilinear functions of the s- and p-wave amplitudes. In par- from 8% at E = 270MeV to +2% at E = 420MeV.
ticular, d /d is sensitive to the E1+ amplitude, because
of interference with M1+ in the terms
6 Future plans
A = | E0+ |2 + | 3E1+ M1+ + M1 |2 , (12)
B = 2 Re[E0+ (3E1+ + M1+ M1 ) ], (13) The rst round of experiments with the Crystal Ball is
C = 12 Re[E1+ (M1+ M1 ) ]. (14) centered on the rst measurement of the magnetic dipole
moment of the + (1232)-resonance. The magnetic dipole
moment, b , provides us with a simple way for testing the
Furthermore, the ratio validity of the theoretical hadron description in the non-
perturbative sector of QCD. This includes quark soliton
1 C Re(E1+ (M1+ M1 ) ) models, the standard quark models, various eective La-
R= = (15)
12 A | E0+ |2 + | 3E1+ + M1+ M1 |2 grangians and lattice QCD calculations. Our experimental
 R. Beck: Experiments with photons at MAMI 181

technique takes advantage of the very short lifetime by

having the radiatively decay to itself. This method has
been successfully pioneered for the ++ using the reac-
tion + p  ++  + p [67]. We propose to deter-
mine b [+ (1232)] using radiative 0 photoproduction:
p +  +  0 p. A rst pilot experiment
p  0 p has been performed
with the TAPS calorime-
ter at MAMI for energies s = 12211331 MeV. Angular
and energy dierential cross section have been determined
for all particles in the nal state in three bins of the exci-
tation energy [68]. The theoretical aspects have been dealt
with in detail already by the theory groups at MAMI [69]
and Tuebingen [70]. b can be determined from the dier-
ential cross section d 5 /d d dE and from the asym-
metry, , for linearly polarized photons.
The broad spectrum of MAMI bremsstrahlung pho-
tons from Emin 100 MeV to Emax 1500 MeV to-
gether with the 4 acceptance of the experimental ap-
paratus allows the simultaneous survey of 0 , 2 0 , 3 0
and production at all energies and for the full angular Fig. 13. The Crystal Ball detector and TAPS as forward wall.
range. Such measurements will be perform with LH2 and
LD2 targets using linearly and circularly polarized pho- Meson_phLadd_v_IM2phot_px

Entries 4428
ton beams. A unique frozen spin target lled with 1 H,  Mean 203.4
Counts

900 RMS 148.3

2
or H will be used in the second stage of the experiment
800
(MAMI-C). The target makes possible new high precision,
high statistics measurements of the cross sections for the 700
 0 N and  N
 N  0 0 N processes at incident pho- 600
ton energies up to 1.5 GeV. In particular it provides a 500
unique opportunity to measure the partial contributions
400
to the GDH sum rule on a neutron target in the reactions
n 0 n and n 0 0 n. Our measurements will also 300

provide new information on the photon coupling of low- 200

mass baryon and hyperon resonances. An incomplete list 100
of other possible measurements includes: i) threshold pho-
toproduction of 0 and at MAMI-B as well as  , and
0
100 200 300 400 500 600
Ks0 at MAMI-C with polarized and unpolarized beams Inv.Mass, MeV

and targets; ii) measurements of the N (1535) magnetic
dipole moment using p  p; iii) a new measurement
of the mass.
The new experimental apparatus is shown in g. 13.
The Crystal Ball with TAPS as the forward wall will be
used for detection of photons and nucleons. In addition
the polar and azimuthal angles of the outgoing proton for
lab > 20 will be measured by the central tracker which is
based on the DAPHNE cylindrical multiwire proportional
chamber. The chamber will be inserted into the Crystal
Ball beam cavity.
The Crystal Ball was build at SLAC and used in
J/ measurements at SPEAR and b-quark physics at
DESY [71]. The CB is constructed of 672 optically iso-
lated NaI(Tl) crystals, 15.7 radiation lengths thick. The
counters are arranged in a spherical shell with an inner ra-
dius of 25.3 cm and an outer radius of 66.0 cm. The hygro-
scopic NaI is housed in two hermetically sealed evacuated
hemispheres. Each crystal is shaped like a truncated trian-
gular pyramid, 40.6 cm high, pointing towards the center
of the Ball. The sides on the inner end are 5.1 cm long
and 12.7 cm on the far end. Electromagnetic showers in
Fig. 14. Invariant mass of two-cluster events for beam photons
with energy above 700 MeV after requiring the missing mass to
be equal to the mass of proton. The peaks are due to 0 2
and 2 decays.
the spectrometer are measured with an energy resolution
E /E 1.7%/(E (GeV))0.4 ;
the angular resolution for photon showers at energies of
0.05 0.5 GeV is = 2 3 in the polar angle and =
2 / sin in the azimuthal angle.
High granularity and a large acceptance make the
Crystal Ball a unique instrument for measuring reactions
with multiphoton nal states. The CB detects neutrons
with an e ciency of 35% at En = 150 MeV [72].
The rst production run of the CB@MAMI program,
a measurement of the photon asymmetry in 0 photopro-
duction at threshold, was accomplished in July-August
2004. In October 2004 we have started a 600 hours long
production run for the measurements of the + (1232)
magnetic dipole moment. Figures 14 17 illustrate the
182 The European Physical Journal A

7 0
p ( p)
total ( p p), arb. units

6
0.2

5 0.18

0.16
4
0.14

3 0.12

0.1
2 0.08

0.06
1
0.04
0 0.02
620 640 660 680 700 720 740 760 780 800 820
ELab
, MeV 0
-1 -0.5 0 0.5 1
Fig. 15. The total cross section of p p in arbitrary units cos(CM
)
from 2 decay modes (solid circles) is compared to the
p 3 0 p total cross section (open circles). Above the Fig. 17. Preliminary results for the photon asymmetry times
threshold the p 3 0 p is mainly from 3 0 decay. The beam polarization shown as a function of cos 0 in c. m. for
total cross sections are relatively normalized at E = 730 MeV. E = 375 MeV (triangles), 405 MeV (squares), and 436 MeV
(circles), compared to MAID predictions. The MAID curves
0 are normalized to the experimental data at cos = 0.
p ( p)
0.2
the mass of the proton. Figure 15 shows an excitation func-
0.15 tion for p ()p in arbitrary units. The total cross
section for p 3 0 p is shown on the same gure for com-
0.1 parison. Below threshold 3 0 events are produced via se-
quential decay of resonances, while above the -threshold
0.05 most of the events are produced by the 3 0 decay.
The experimental setup made up of the Crystal Ball
-0 and TAPS is almost perfectly -symmetric. Together with
the good quality polarized MAMI beam it allows high
-0.05 statistics, low systematics uncertainty, polarization mea-
surements. Figure 16 shows the -dependence of the beam
-0.1 photon asymmetry, (), for the reaction p 0 p.
The asymmetry is not corrected for the beam polariza-
-0.15 tion. The data represent about 5 % of the statistics ob-
tained in the course of our most recent ((1232)) run.
-0.2 The photon asymmetry as a function of cos 0 is shown
-150 -100 -50 0 50 100 150
0
( ), deg
in g. 17 for beam photon energies of 375 MeV, 405 MeV,
and 436 MeV in comparison with MAID [73]. The results
Fig. 16. -dependence of the beam photon asymmetry times are very preliminary. The measured distributions are not
polarization, p (), for the reaction p 0 p, where corrected for the beam photon polarization therefore the
() ()
() =  ()+ ()
. The function shows clear cos(2) behav- MAID curves are normalized to the data at cos = 0.

ior over the full angular range. The data covers the beam en- The photon asymmetries obtained in our experiment show
ergy interval of 360 450 MeV and is integrated over 0 . good agreement with the MAID evaluations for cos > 0
and slightly deviate in the backward angles. The dierence
quality of the data, showing some characteristic distribu- between our data and MAID gets larger at higher beam
tions. energies.
The invariant mass of two photons for an incident
beam with energy above 700 MeV is shown in g. 14. The I would like to thank the organizers Hartmuth Arenhovel, Hart-
two peaks of the spectra are due to the reactions p mut Backe, Dieter Drechsel, Karl-Heinz Kaiser and Thomas
0 ()p and p ()p. The two-gamma invariant Walcher of the symposium 20 Years of Physics at the Mainz
mass is shown for events with the missing mass equal to Microtron MAMI .
 R. Beck: Experiments with photons at MAMI
References
1. I. Anthony et al., Nucl. Instrum. Methods A 301, 230
(1991).
2. S.J. Hall et al., Nucl. Instrum. Methods A 368, 698 (1996).
3. URL: http://wwwa2.kph.uni-mainz.de/A2.
4. R. Novotny, IEEE Trans. Nucl. Sci. 38, 379 (1991).
5. G. Audit et al., Nucl. Instrum. Methods A 301, 473 (1991).
6. A. Partridge et al., Paper presented to the IEEE Meeting,
San Francisco, October 1977 (Stanford Linear Accelerator
Center, Stanford, 1977).
7. V.A. Petrunkin, Sov. Phys. JETP 13, 808 (1961).
8. V.A. Petrunkin, Sov. J. Part. Nucl. 12, 278 (1981).
9. V. Olmos de Leon, PhD Thesis, Mainz University (2000).
10. V. Olmos de Leon et al., Eur. Phys. J. A 10, 207 (2001).
11. A. Lvov et al., Sov. J. Nucl. Phys. 34, 597 (1981).
12. A. Zieger et al., Phys. Lett. B 278, 34 (1992).
13. J.F. Donoghue, B.R. Holstein, Phys. Rev. D 40, 2378
(1989); J. Bijnens, F. Cornet, Nucl. Phys. B 296, 557
(1988); B.R. Holstein, Comments Nucl. Part. Phys. 19,
221 (1990); S. Belluci, J. Gasser, M.E. Sainio, Nucl. Phys.
B 423, 80 (1994).
14. U. Burgi, Nucl. Phys. B 79, 392 (1997).
15. A.N. Ivanov, M. Nagu, N.I. Troitskaya, Mod. Phys. Lett.
A 7, 1997 (1992).
16. L.V. Filkov, I. Guiasu, E.E. Radescu, Phys. Rev. D 26,
3146 (1982).
17. L.V. Filkov, V.L. Kashevarov, Eur. Phys. J. A 5, 285
(1999).
18. V. Bernard, B. Hiller, W. Weise, Phys. Lett. B 205, 16
(1988).
19. M.A. Ivanov, T. Mizutani, Phys. Rev. D 45, 1580 (1992).
20. Yu.M. Antipov et al., Phys. Lett. B 121, 445 (1983).
21. T.A. Aybergenov et al., Sov. Phys. Lebedev Inst. Rep. 6,
32 (1984); Czech. J. Phys. B 36, 948 (1986).
22. J. Portoles, M.R. Pennington, The Second DAN E
Physics Handbook, Vol. 2 (1995) p. 579, hep-ph/9407295.
23. J. Boyer et al., Phys. Rev. D 42, 1350 (1990).
24. F. Donoghue, B. Holstein, Phys. Rev. D 48, 137 (1993).
25. T. Walcher, Prog. Part. Nucl. Phys. 24, 189 (1990).
26. J. Ahrens et al., Nucl. Phys. News 4, 5 (1994).
27. P. Grabmayer et al., Nucl. Instrum. Methods A 402, 85
(1998).
28. Ch. Unkmeir, PhD Thesis, Mainz University (2000).
29. J. Ahrens et al., Eur. Phys. J. A 23, 113 (2005).
30. E. Mazzucato et al., Phys. Rev. Lett. 57, 3144 (1986).
31. R. Beck et al., Phys. Rev. Lett. 65, 1841 (1990).
32. M. Fuchs et al., Phys. Lett. B 368, 20 (1996).
33. J.C. Bergstrom et al., Phys. Rev. C 53, R1052 (1996); 55,
2016 (1997).
34. P. de Baenst, Nucl. Phys. B 24, 633 (1970).
35. I.A. Vainshtein, V.I. Zakharov, Nucl. Phys. B 36, 589
(1972).
36. V. Bernard, J. Gasser, N. Kaiser, U.-G. Mei ner, Phys.
Lett. B 268, 291 (1991).
183
37. V. Bernard, N. Kaiser, U.-G. Mei ner, Z. Phys. C 70, 483
(1996).
38. A. Schmidt, Phys. Rev. Lett. 87, 232501 (2001).
39. A. Schmidt, Doktorarbeit, University Mainz (2001).
40. H. Herminghaus, K.H. Kaiser, H. Euteneuer, Nucl. In-
strum. Methods A 138, 1 (1976).
41. D. Lohmann, J. Peise et al., Nucl. Instrum. Methods A
343, 494 (1994).
42. A. Schmidt, Diplomarbeit, University Mainz (1995).
43. R. Novotny, IEEE Trans. Nucl. Sci. 43, 1260 (1996).
44. O. Hanstein, D. Drechsel, L. Tiator, Phys. Lett. B 399, 13
(1997).
45. V. Bernard, N. Kaiser, U.-G. Mei ner, Phys. Lett. B 378,
337 (1996).
46. V. Bernard, N. Kaiser, U.-G. Mei ner, Eur. Phys. J. A 11,
209 (2001).
47. B. Pasquini, D. Drechsel, L. Tiator, Eur. Phys. J. A 23,
279 (2005).
48. V. Bernard, B. Kubis, U.-G. Mei ner, arXiv:nucl-th/
0506023v1 (2005).
49. A. Rujula, H. Georgi, S.L. Glashow, Phys. Rev. D 12, 147
(1975).
50. C. Becchi, G. Morpurgo, Phys. Lett. 17, 352 (1965).
51. R. Koniuk, N. Isgur, Phys. Rev. D 21, 1868 (1980).
52. S.S. Gershteyn et al., Sov. J. Nucl. Phys. 34, 870 (1981).
53. D. Drechsel, M.M. Giannini, Phys. Lett. B 143, 329
(1984).
54. S. Capstick, Phys. Rev. D 46, 2864 (1992).
55. A. Wirzba, W. Weise, Phys. Lett. B 188, 6 (1987).
56. K. Bermuth et al., Phys. Rev. D 37, 89 (1988).
57. D.B. Leinweber, Proceedings of the International Confer-
ence Baryons92 (1992) p. 29.
58. A. Buchmann et al., Phys. Rev. C 55, 448 (1997).
59. R. Beck, H.-P. Krahn et al., Phys. Rev. Lett. 78, 606
(1997).
60. G.S. Blanpied et al., Phys. Rev. Lett. 79, 4337 (1997).
61. R.M. Davidson, N.C. Mukhopadhyay, Phys. Rev. Lett. 79,
4509 (1997).
62. G. Keaton, R.L. Workman, Phys. Rev. Lett. 79, 4511
(1997).
63. O. Hanstein et al., Phys. Lett. B 385, 45 (1996).
64. R. Leukel, PhD Thesis, University Mainz, 2001.
65. O. Hanstein et al., Nucl. Phys. A 632, 561 (1998).
66. R. Beck et al., Phys. Rev. C 61, 035204 (2000).
67. B. Nefkens et al., Phys. Rev. D 18, 3911 (1978).
68. M. Kotulla et al., Phys. Rev. Lett. 89, 272001 (2002).
69. D. Drechsel, M. Vanderhaeghen, Phys. Rev. C 64, 065202
(2001).
70. A.I. Machavariani, A. Faessler, arXiv:nucl-th/0202060
(2002).
71. E.D. Bloom, C.W. Peck, Annu. Rev. Nucl. Sci. 33, 143
(1983).
72. T.D. Stanislaus et al., Nucl. Instrum. Methods A 462, 12
(2001).
73. D. Drechsel et al., Nucl. Phys. A 645, 145 (1999).
Eur. Phys. J. A 28, s01, 185195 (2006)
DOI: 10.1140/epja/i2006-09-019-0 EPJ A direct
electronic only

Coherent X-rays at MAMI

W. Lautha , H. Backe, O. Kettigb , P. Kunz, A. Sharafutdinov, and T. Weber

Institut fur Kernphysik der Universitat Mainz, D-55099 Mainz, Germany

/
Published online: 31 May 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. Coherent radiation in the range from soft X-rays up to hard X-rays, produced by the low-
emittance electron beam of MAMI, can be used for various applications. Novel types of interferometers
have been developed for the measurement of the complex index of refraction of thin self-supporting foils.
For the vacuum ultraviolet and soft X-ray region the interferometer consists of two collinear undulators,
and a grating spectrometer. A foil placed between the undulators causes a phase shift and an attenuation of
the oscillation amplitude. The complex index of refraction has been measured at the L2,3 -absorption edges
of nickel. A novel method is described for the measurement of the X-ray magnetic circular birefringence.
For the hard X-ray region the interferometer consists of two foils at which the 855 MeV electron beam
produces transition radiation. Distinct interference oscillations have been observed as a function of both,
the photon emission angle and the distance between the foils. The refractive index decrement () of a
2 m thick nickel sample foil has been measured at X-ray energies around the K absorption edge at 8333 eV
and at 9930 eV with an accuracy of better than 1.5 %. The line width of parametric X radiation (PXR) was
measured in backward geometry with a Si single-crystal monochromator. Upper limits of the line width of
42 meV, 50 meV, and 44 meV, have been determined for the (333), (444) and (555) reections at photon
energies of 5932 eV, 7909 eV, and 9887 eV, respectively. Small angle scattering of the electrons in the crystal
leads to a stochastic frequency modulation of the exponentially damped wave train which results in the
line broadening. To elucidate the quest if the production of PXR is a kinematical or a dynamical process
the radiation from silicon single-crystal targets, emitted close to the electron direction, has been studied.
The observed interference structures and the narrow-band radiation in forward direction shows that PXR
is produced in a dynamical process.
PACS. 07.60.Ly Interferometers 78.20.Ci Optical constants (including refractive index, complex di-
electric constant, absorption, reection and transmission coecients, emissivity) 41.60.-m Radiation by
moving charges 41.50.+h X-ray beams and X-ray optics

1 Introduction also considered at that time to be of interest for the

Immediately after MAMI B became fully operational for
nuclear physics experiments, in the early nineties also a
research program was launched to explore the potential of
the high-quality, low-emittance electron beam for applica-
tions. It was our conviction that it ought to be possible
to use X-rays produced with the 855 MeV beam of MAMI
in various elds of physics, material science, medicine and
biology. Of course, attention focused at that time on the
third-generation synchrotron radiation sources like ESRF,
APS and Spring8 which were on the horizon and, in par-
ticular, on the production of brilliant soft X-ray ashes
in a single pass of high-current electron bunches through
an undulator by the process of self-amplied spontaneous
emission (SASE). However, various other processes were
a
e-mail: lauth@kph.uni-mainz.de
b
Present address: Arcor AG & Co. KG, 65760 Eschborn,
Germany.
production of soft and hard X-ray beams with external
high-quality electron beams. The most important ones
are schematically depicted in g. 1. These are transition
radiation (TR), channeling radiation (CR), parametric
X-radiation (PXR), undulator radiation (UR), and Smith-
Purcell radiation (SPR). There are potential advantages
of such X-ray sources over synchrotron radiation sources
or free electron lasers on the basis of SASE. First of all,
since accelerators may become relatively inexpensive in
the future, they could meet the radiation requirements of
research laboratories or hospitals on the spot. Secondly,
the X-ray beam can be triggered and its time structure
adapted to nearly any experimental requirement. In par-
ticular, the electron beam can easily be turned o if the
X-ray beam is not used minimizing power consumption
and radiation production in the beam dump.
At MAMI with UR and TR brilliant photon beams
can be produced with energies covering the range between
186 The European Physical Journal A
Undulator Radiation Transition Radiation Parametric
X-Radiation (PXR)
N S N S
S N S N
Channeling Radiation Smith-Purcell
(CR) Radiation (SP)
Fig. 1. Processes for the generation of coherent radiation with
relativistic electrons.
some 100 eV and up to about 50 keV. Brilliance means
that a large number of photons from a small source spot
size down to the m range are emitted with high direc-
tionality in space. In particular, the hard TR X-ray beam
turned out to be comparable in photon ux and brilliance
with second-generation synchrotron radiation sources [1].
Taking advantage of this fact a novel K-edge imaging sys-
tem [2] was developed. In addition, X-ray phase contrast
imaging has been accomplished. For the latter we refer to
the contribution ref. [3] in this issue.
SPR is generated when a beam of charged particles
passes close to the surface of a periodic structure. This
type of radiation has been investigated in the optical spec-
tral range with the 855 MeV beam of MAMI [4]. A detailed
discussion of the emitted photon number per electron led
to the conclusion that a SPR source is not advantageous in
comparison with an UR source for ultrarelativistic beam
energies. At present experiments are being performed at
the 3.4 MeV injector LINAC of MAMI to explore the gen-
eration of intense SPR in the THz region of the electro-
magnetic spectrum [5].
In channeling, the charged particle directions are
closely aligned with an atomic row or with crystal planes,
and their motion is governed by many correlated collisions
with crystal atoms. As a result, the particles are steered
along strings or planes and CR is emitted. Channeling
experiments have been taken up at MAMI only very re-
cently, see ref. [6]. They are connected with the feasibility
of a crystalline undulator with positrons which was inves-
tigated recently in great detail [7].
In sect. 2 of this contribution our results obtained with
a novel interferometry principle will be reviewed. The in-
terferometer consists of two spatially separated, phase-
correlated radiation sources in the soft and hard X-ray
ranges. For soft X-rays the radiation sources are undula-
tors with small period length, for hard X-rays they are
foils in which the electron beam produces TR. It will be
shown in this section 2 that the optical properties of foils
can be determined in the soft and hard X-ray region.
If the electron beam strikes a crystal, it emits quasi-
monochromatic PXR close to a Bragg angle. This kind
of radiation source is amazing for its compactness, since
production and monochromatisation of the radiation take
place in the same crystal. The expected small spectral
line width of PXR would promise an abundance of ap-
plication possibilities, e.g. within the eld of solid-state
physics. However, line broadening by multiple scattering
of the electrons in the crystal may spoil the superb line
width. In sect. 3 our experiments addressing this question
are reviewed, including fundamental aspects like the ques-
tion whether the process of PXR production is of kinemat-
ical or of dynamical nature.
The paper closes with a conclusion and an outlook.
2 X-Ray interferometry
Resonant anomalous X-ray scattering plays an increas-
ingly important role in many disciplines of physics, biol-
ogy, and material sciences. Using the brilliant and tune-
able X-ray beams from modern synchrotron radiation
sources, it is now possible to fully exploit the informa-
tion in the strong energy and polarisation dependence
of the atomic scattering amplitude f (, q) = f0 (q) +
f  () + if  () near absorption edges [8,9]. This micro-
scopic description can be translated into a macroscopic
description with the complex index of refraction n() =
1 () + i() by the relations for the refractive index
decrement () = (1/2) (p /)2 (f0 (0) + f  ())/Z and the
absorption index () = (1/2) (p /)2 f  ()/Z. In these
expressions Z is the atomic number, p the plasma fre-
quency with p 2 = 4r0 c2 na Z, r0 the classical electron
radius, na the number of atoms per volume, and f0 (0) = Z
neglecting relativistic corrections.
The imaginary part of the scattering amplitude f 
can be directly determined from the total photon cross
section () by employing the optical theorem: f  () =
()/(4r0 c). The total cross section is well approxi-
mated by the absorption cross section which can be mea-
sured by a transmission experiment. The real part f  can
be calculated from f  by means of Kramers-Kronig disper-
sion relations. However, this method is suited for a relative
comparison only, since it requires precise absorption data
for all frequencies from zero to innity [10]. If precise ab-
solute values are needed, a direct measurement of f  () is
required. Direct measurements are based on X-ray inter-
ferometry with the Bonse Hard-X-ray interferometer [11],
refraction through a prism [12,13], diraction from perfect
crystals and pendellosung fringes [14,15], determination of
the angle of total reection [10,16], and Fresnel bi-mirror
interferometry [17]. Whereas most of these methods are
based on splitting of either wave amplitudes or wave fronts
the novel type of interferometer which is described here
uses two spatially separated coherent X-ray emitters.
The basic idea of the interferometer will be explained
by means of the schematic experimental setup shown in
g. 2. Relativistic electrons create two wave trains in
source 1 and source 2, the relative distance of which
is given in leading order by (, d) = 12 ( 2 + 2 )d. Here
d is the distance between the sources, the Lorentz factor
of the electron, and the observation angle with respect
to the electron beam direction. The monochromator serves
 W. Lauth et al.: Coherent X-rays at MAMI 187

X-ray sources Mono- Detector

chromator 2000 with sample =Z = 847.3 eV
Sample
S 2 foil S 1 e- A2 A1 1500
T1 T2
e-
1000

500
d '
Variable distance 0
1000
with sample =Z = 854.3 eV
Fig. 2. Interferometry with spatially separated coherent X-ray

counts
emitters. 800
600
as a Fourier analyser of the wave trains. The two result- 400
ing plane waves have a phase dierence of = v (, d) 200
(v is the velocity of the electron) and interfere in the de-
tector, resulting in oscillations of the intensity I(d), if the 0
distance d is varied. A sample foil placed between the two 2000 with sample =Z = 874.1 eV
sources produces an additional phase shift and attenuation 1500
of wave 2. Consequently, both quantities, i.e. the refrac-
tive index decrement and the absorption index , can 1000
be extracted from the measured interference oscillations
I(d) with and without the foil between the sources. This 500
holds independently of the nature of the emission process,
provided that the produced X-rays remain coherent. 0
210 220 230 240 250 260

2.1 The soft X-ray interferometer
For photon energies in the range of about 100 eV up to
2 keV we use two identical undulators (period length LU =
12 mm, number of periods 10, undulator parameter K =
1.1) and a grating spectrometer. The recorded intensity
with a foil between the undulators is given by
I(d) = |A1 |2 + |A2 |2 e2 c ()t0 + 2|A1 | |A2 |e c ()t0

 tions are shown in g. 3. The extracted optical constants
K2
cos (, d) + ()t0 + 2 LU (1)
c 4
with A2 being the amplitude of the upstream undulator,
A1 that of the downstream one and t0 the thickness of the
foil.
The interferometer has been developed at the Mainz
Microtron MAMI and its performance was demonstrated
with measurements at the K absorption edge of carbon
at 284 eV. Details of this experiment can be found else-
where [18]. The visibility (coherence), dened by C =
(Imax Imin )/(Imax + Imin ) without sample foil, is close
to its maximum value C = 1. No loss of coherence was
observed over the scanning distance of 15 cm. Therefore,
the optical constants and could be extracted by a t
with simple cosine functions.
Measurements were also performed at the L2 -
absorption and L3 -absorption edges of nickel at 871 eV
and 855 eV, respectively. The experimental setup was sim-
ilar with that described in ref. [18] with the following
modications: the third harmonics of the undulator was
chosen as radiation, which was generated at an electron
energy of 766.3 MeV. The radiation was analyzed with a
variable line spacing (VLS) grating [19] with an energy
d [ mm ]
Fig. 3. Intensity oscillations as a function of the distance d
between the undulators with and without the self-supporting
83.2 g/cm2 nickel sample foil at three dierent photon ener-
gies at the L2,3 absorption edges. Note the change of sign of
the phase shift at the photon energy of 854.3 eV.
resolution of 0.44 eV. Typical measured intensity oscilla-
at the Ni L2,3 are shown in g. 4. A high accuracy has
been reached, even in the region where in which the
optical constants can be determined from reectivity mea-
surements only with large uncertainties [20]. The remark-
able fact about this measurement is the hight of the L3
resonance with its maximum value = (5.640.24)103 .
This value corresponds to an imaginary scattering factor
f  = 65.7 2.8. In ref. [21] a mass attenuation coecient
= 2 104 cm2 /g was determined from which, with the
relation = (4), an f  = 24 can be calculated with an
estimated uncertainty of 10 %. The dierence may orig-
inate from the better resolution in our experiment. Cor-
related with this L3 absorption resonance is a change of
sign of the refractive index decrement (). It is interest-
ing to notice that for () < 0 the real part of the refrac-
tive index 1 () is larger than 1 and a monochromatic
Cherenkov radiation with an energy of 855 eV will be emit-
ted. This fact has been pointed out in the literature.
The strong absorbtion line at the L3 edge of Ni is
the result of an allowed dipole transition between the
2p3/2 core state and empty 3d valence states above the
Fermi energy. This transition exhibits a strong X-ray
Magnetic Circular Dichroisim (XMCD) eect which can
188 The European Physical Journal A
6
4
[ 10 ]
-3
2 Movable Gap
E
0 Beam tube
L3 L2
2 3 mm
0
[ 10 ]
-3
-2
G
-4
840 850 860 870 880
=Z [ eV ]
Fig. 4. Absorption index and refraction index decrement
of an 83 g/cm2 Ni sample foil as obtained from the analysis of
the intensity oscillation measurements as shown in g. 3. The
photon energy resolution was 0.44 eV.
be used to probe the magnetic properties of the mate-
rial [22]. With XMCD spectroscopy, pioniered by Schutz
and coworkers [23], the dierence in the absorption of left
and right circular polarized light is measured. Connected
with the XMCD is the X-ray Magnetic Circular Birefrin-
gence (XMCB) also known as Faraday Magneto-Optical
Rotation (MOR). Both eects can be described with the
complex index of refraction n () = 1 () + i ()
for the two helicity states of the radiation. Thickness
variation in the transmission measurements hampered the
accuracy in the determination of () with the XMCD
spectroscopy, especially near the strong-absorption lines
where the XMCD eect is the largest. It has been shown
in ref. [24], that such thickness eects are less important
for the measurement of the refractive index (). How-
ever, this measurement is dicult because of the lack of
polarization analyzers in the soft X-ray region.
Our interferometer has been developed further to mea-
sure () without any polarization lter. The magne-
tized sample foil with the magnetization direction par-
allel or antiparallel to the electron beam axis was po-
sitioned between the two undulators, see g. 5. Due to
the XMCB and XMCD eect, the linear polarized light
from the rst undulator suers a helicity-dependent phase
shift and absorption resulting in elliptical polarized light.
The major semi-axis of the ellipse is rotated by an an-
gle with respect to the plane in which the impinging
linear polarized E eld vector oscillates. The second un-
dulator which can be rotated around the electron beam
axis acts as the analyzer. The maximum visibility deter-
mines the rotation angle of the ellipse on which the
+/- 90
e-
Undulator 1
Undulator 2
1m
Fig. 5. Experimental setup of the new undulator interferom-
eter. The undulator 2 can be both moved along and rotated
around the electron beam axis. These possibilities allow pro-
duction of radiation with a well-dened polarization state be-
tween linear and circular. In addition, the undulator gap can
be changed for an online variation of the photon energy via the
undulator parameter K. Note that the electron beam traverses
a vacuum tube of only 3 mm in diameter over a length of about
1 m.
10
5
0
\ []
-5
L3 L2
-10
840 850 860 870
=Z [ eV ]
Fig. 6. Preliminary results of a measurement of the rotation
angle of the ellipse on which the E eld vector rotates behind
a 72.5 nm self-supporting Ni foil, which was illuminated with a
linear polarized undulator radiation. Shown are measurements
at the L3 and L2 absorption edges.
E eld vector moves behind the foil. This angle is given
by = 2c
t0 (+ () ()). The rst measurements of
the angle of a 72.5 nm magnetized Ni foil, placed in a
magnetic eld of 1.0 T, are shown in g. 6. In this ex-
periment the photons were detected in the energy disper-
sive plane of the grating spectrometer with a windowless
CCD-detector [25]. The visibility for rotation angles be-
tween 105 and +105 of the second undulator was ex-
tracted from the interference oscillation as a function of
the distance between the undulators. Furthermore, also
the refractive index decrement () can be extracted si-
multaneously from this measurement which allows the full
determination of the helicity-dependent complex index of
 W. Lauth et al.: Coherent X-rays at MAMI
Fig. 7. Experimental setup of the transition radiation inter-
ferometer. The monochromator is a at silicon single crystal,
cut with the (111) plane parallel to the surface, at a distance
of 5.5 m from the foils. It acts as energy dispersive mirror. A
silicon 1 3 cm2 pn CCD with active thickness of 300 m and
a pixel size of 150 150 m2 was used as detector [26]. The
CCD is located at a distance of 5.5 m from the monochroma-
tor crystal.
refraction n () without knowledge of the degree of the
polarization of the light.
2.2 The hard X-ray interferometer
When a charged particle traverses the interface of two me-
dia with dierent dielectric polarizability, transition radi-
ation (TR) is emitted. The TR radiation is sharply peaked
into forward direction with a characteristic opening angle
of about 2/ and features broadband characteristics with
a cut-o energy at about 40 keV. It is well known that the
TR amplitudes from two interfaces of a single foil inter-
fere coherently. The same holds for the amplitudes from
two or more foils (interfoil interference), see, e.g., ref. [27].
It was suggested more than fteen years ago by Moran
et al. [28] that the interfoil coherence of TR, generated
by relativistic electrons, constitutes a new technique for
the measurement of the refractive index decrement ()
in the X-ray region. A number of experiments were per-
formed in the soft and hard X-ray region in which more
or less clear interference structures were observed [28,29,
30,31,32,1]. However, that () really can be measured by
such a technique has been demonstrated rst by the X-ray
interferometer described in this work. It consists of two
foils at which the 855 MeV MAMI beam produces tran-
sition radiation, a single crystal spectrometer with a at
crystal in Bragg geometry, and a pn CCD X-ray detector,
see g. 7. Details of the experiment have been reported
elsewhere [33,34,35].
Measurements have been performed on nickel around
the K absorption edge at 8333 eV as well as around
9930 eV, well above the K absorption edge, where ex-
tended diraction anomalous ne structures (EDAFS) in
the dispersion spectra are supposed to be negligible. A
189
60 447
335
40
row
223
20
111
0 0
0 20 40 60 80 100 120 140 160 180 200
col mn
Fig. 8. Measured interference pattern at a xed foil distance
d = 475.0(2) m. The central photon energy h = 9929 eV is
well above the K absorption edge of nickel. Grey levels indicate
the intensities.
Fig. 9. Examples of interference oscillations as a function of
foil distance d at a xed photon energy h and a xed observa-
tion angle as indicated. The ordinates are counts/(ms mrad2 ).
The measurements (dots) are well met by the simulation cal-
culation (full line).
typical example of a measured TR interference pattern is
shown in g. 8.
There are two possibilities to extract 1 of the down-
stream nickel foil from the TR interference patterns shown
in g. 8. In the rst one the information is obtained from
the interference oscillations as a function of the observa-
tion angle at xed distance between the foils. Since the
pn CCD detector was arranged horizontally the interfer-
ence oscillations were observed essentially along the en-
ergy dispersive angular coordinate x . Reliable results can
be obtained if 1 () can be approximated with reasonable
accuracy by a linear expansion as a function of the photon
energy h. Close to an absorption edge such an approxi-
mation is not valid. In such a case, as a second possibility,
1 () can be extracted from the interference oscillations
observed as a function of foil distance d, as shown in g. 9.
The rapid damping of the visibility after about three pe-
riods originates mainly from the small-angle scattering of
the electrons in the upstream foil which was made of the
low-Z element beryllium to minimize the eect. The re-
sults of the refractive index decrement 1 () around the
K absorption edge of Ni are shown in g. 10. Details of
the analysis procedure can be found in refs. [34,36].
The measured 1 () agrees at the K absorption edge
within the total error of 1 /1 1.5%, with the
Kramers-Kronig transformation of 1 () of refs. [37,38].
190 The European Physical Journal A
2.10
1 105
2.00
1.90
1.80
8300 8320 8340 8360 8380
h [eV]
Fig. 10. Refractive index decrement 1 (h) at the K absorp-
tion edge as obtained from interference oscillations. The error
bars represent pure statistical errors of the experiment. The full
line is a Kramers-Kronig transform of the absorption data [37,
38].
In comparison, the measurements of 1 () by Bonse and
co-workers [39,40] with Bonse-Hart interferometers [11]
are systematically too low by about 1.5%. This small de-
viation may originate from a systematical error in the foil
thickness measurements in those experiments.
3 Parametric X-radiation
Parametric X-radiation (PXR) or quasi-Cherenkov radi-
ation is produced when a relativistic electron traverses
a single crystal, and the wave vector kv of the virtual
photon associated with the electron eld, the reciprocal
lattice vector H of a specic crystal plane and the wave
vector kr of the emitted X-ray nearly full the well-known
diraction condition kv + H = kr . For ultra-relativistic
electrons the wave vector kv nearly coincides with the
electron velocity vector v and the emission process can
also be understood as diraction of virtual photons by
the crystal. The production mechanism of PXR in such a
medium with three-dimensional periodic permittivity was
extensively studied both theoretically by Baryshevsky and
Feranchuk [41,42], Garibyan and Yang [43,44], and Ter-
Mikaelian [45] and experimentally by a large number of
researchers. For an overview of the theoretical and exper-
imental work up to the year 1997 see, e.g., ref. [46] and
also ref. [47,48,49,50] for the recent works.
PXR features a sharp quasi-monochromatic X-ray
beam close to the Bragg angle with a very narrow line
width. The angular distribution consists of one peak
above and one below the symmetry plane of the crys-
tal. Their spatial widths are characterized by the angle
ph = (1/ 2 + (p /)2 )1/2 , with the Lorentz factor of
the moving particle, the angular frequency of the emit-
ted photon, and p the plasma frequency of the crystal,
with h = 31 eV for Si. The theoretical description of
the intensity distribution of PXR [51,52,45,53,54], suit-
ably modied for self-absorption and multiple-scattering
eects, has been tested for a broad range of electron ener-
gies extending from 3.5 MeV [55,56] to about 1 GeV [46].
It was found to be accurate within 12%.
The line width Wnat of PXR is in essence determined
by the photo absorption in the crystal, if the electron beam
divergence, the small-angle electron scattering in the crys-
tal and the solid angle of the X-ray detector can all be
neglected, and the crystal has a perfect lattice structure.
An exponentially damped wave train is emitted of which
the Fourier transform is a Lorentzian with width [54]
0
Wnat = h0 . (2)
2 sin2 0
The quantity 0 is the imaginary part of the mean dielec-
tric susceptibility 0 = 0 + i0 . This line width Wnat
is called natural line width. For example, the (444) re-
ection of silicon at h0 = 7908 eV yields 0 = 3.74
107 [57]. With this value a natural line width Wnat =
1.48 meV results from eq. (2) for backward emission, i.e.
for 0  /2. It is interesting to note that the correspond-
ing Darwin-Prins curve has a width of WDP = 38.5 meV
and is a factor of 26 broader, see also g. 13. In view of the
fact that such a narrow bandwidth source could be of ex-
treme interest for many applications, a number of experi-
ments were performed to determine the line width of PXR.
Measurements at the low electron beam energy of 6.8 MeV
result in a line width of 48 eV for a 55 m thick diamond
crystal at a photon energy of 8.98 keV [58]. This rather
large line width originates from the multiple scattering of
electrons in the crystal. With the critical absorber tech-
nique experiments have been performed at MAMI at an
electron beam energy of 855 MeV [59]. Upper limits of the
line width of 1.2 eV and 3.5 eV have been determined for
the (111) and (022) reections of silicon single crystals
at photon energies of 4966 eV and 8332 eV. These limits
originate mainly from geometrical line broadening eects.
In backward geometry geometrical line broadening
contributions, originating from the angular spread of the
electron beam and small-angle scattering of the electrons
in the crystal, minimize. To investigate whether under
these conditions line widths as small as the Darwin-Prins
values can be reached, experiments have been performed
at MAMI which will be described in the next subsection.
3.1 Measurements of the line width of PXR
Measurements of the line shape of PXR were done with the
855 MeV electron beam of MAMI [60]. The experimental
setup is shown in g. 11. The backward-emitted radiation
of a reection from an (nnn) plane was analyzed with a
silicon single-crystal monochromator in Bragg geometry at
the same (nnn) reection. A vertical slit of 2 mm width in
front of the analyzer crystal reduces the divergence of the
X-rays to 0.07 mrad. The PXR radiation was separated
from the backward-diracted transition radiation (DTR)
by tilting the target crystal to x /2 = 5 mrad for which
PXR emission dominates in comparison to DTR.
The measured line shapes are shown in g. 12 for var-
ious reections. The lines are convolutions of the PXR
emission spectra, which are broadened by multiple scatter-
ing of the electrons, and the response function of the ana-
lyzer crystal. The data were analyzed assuming a Gaussian
distribution for the broadened PXR line and the Darwin-
Prins curve of the analyzer crystal. The best ts are shown
 W. Lauth et al.: Coherent X-rays at MAMI 191

1000
Target Crystal Analyzer

(111)
(T1 = 145 K) Crystal (T2 = 296 K)

(333)
1, x 2 Slit

(444)

(555)
2, x 2

Line Width [meV]

Aperture 100

 
H1 H2 1.12m 10
855 MeV Si-Drift
Electron Beam Detector
15.3m
1

Fig. 11. Experimental setup. The target crystal was cooled 0 2 4 6 8 10

down to a temperature of T1 = 145 K, while the analyzer
crystal was kept at room temperature T2 = 296 K. Since the
lattice parameter a(T ) of the crystal diminishes by cooling,
an enlarged photon energy and an enlarged deection angle
2,x = 45.6 mrad at 1,x = 0 result. The angle is large enough
for a lateral displacement of the Si-drift detector that it does
not shadow the radiation to be analyzed. A 1 mm thick sili-
con single crystal with an area of 30 mm 20 mm, cut with the
(111) plane parallel to the surface, was used as analyzer crystal.
=Z [keV]
Fig. 13. Measured and calculated line widths (FWHM) for
various (nnn) reections for a silicon single crystal of 525 m
thickness. Open circles: measured PXR line width, full squares:
natural PXR line widths for straight electron trajectories, open
squares: Darwin-Prins width, triangles: line widths calculated
with the analytical model of ref. [48], and stars: Monte Carlo
simulations.

3 It has been discussed in ref. [48] that the small-angle

scattering of the electron in the Coulomb potential of the
2
(333) crystal atoms results in a stochastic change of the electron
direction which leads to a stochastic frequency modula-
1 tion of the exponentially damped wave train of PXR. As
a consequence, the PXR line broadens. The scattering dis-
0
3 tribution function was approximated by a Gaussian. This
NPh [10 /e]

approximation may not be anymore valid if the electron

(444)
-12

2 enters the crystal close to a channeling axis or a channeling

plane. In such cases Monte Carlo simulations of the scat-
1 tering process must be applied to obtain the PXR wave
train. However, it is interesting to notice that calculations
0
according to ref. [48] as well as also Monte Carlo simulated
0.3 line shapes, which are both shown also in g. 13, are in
(555)
0.2 good agrement with the measurements.

0.1
0.0
-0.4 -0.3 -0.2
'=Z [eV]
Fig. 12. Line shape measurements of PXR at 1,x /2 = 5 mrad
for dierent reections as indicated by (nnn) values; h =
h h0 , with h0 the energy of the DTR reection.
in g. 12 by the full lines. The resulting PXR widths are
shown in g. 13 together with the natural line width ac-
cording to eq. (2) and the width of the Darwin-Prins curve.
Only for the (333) reex the observed line width is smaller
than the width of the Darwin-Prins curve. It could not be
excluded that imperfections of the analyzer crystal itself
broaden the higher reections. Therefore, the real PXR
line could be somewhat smaller as the t results shown in
g. 13.
3.2 Measurement of forward-diracted PXR
-0.1 0.0 Up to now it could not be decided experimentally whether
PXR emission is a kinematical or a dynamical process.
The reason has been discussed by Nitta in a recent pa-
per [61]. He showed that the rst-order approximation
of the dynamical calculation gives the kinematical ex-
pression. Extremely accurate absolute intensity measure-
ments would be required to gure out a dierence. Bary-
shevsky [62] proposed to search for the predicted forward-
diracted wave (FDPXR) which is associated to PXR and
emitted close to the direction of the electron propagation.
Similar proposals have also been communicated by Na-
sonov [63,64]. In ref. [65] the observation of narrow FD-
PXR structures from a 410 m thick tungsten single crys-
tal at photon energies of 28.3 and 40 keV is reported.
At the Mainz Microtron MAMI experiments were per-
formed for the search of the forward-diracted wave (FD-
PXR) in single silicon crystals of various thicknesses [66].
192
x 0
A 2
0 x n 1.5
e- Beam H0
x PXR
z
HA
O
pn CCD
Column 200
Fig. 14. Schematic experimental setup for the search of FD-
PXR.
1000
5000
500
0 0
1000
5000
Events
500
0 0
1000
5000
500
0 0
0 50 100 150 200 0 50
Column
Fig. 15. Measurements at photon energy of 10.554 keV and
target thickness of 58 m (left) and 1000 m (right). Shown
are intensity distributions summed over all 64 rows of the
pn CCD detector as a function of the column number. From
the upper to the lower panel the rotation angle x of the
target crystal was varied in steps x = 0.5894 mrad. Beam
current: 53.5 nA, exposure time: 600 s. Left panel: Beam spot
size about 500 m (FWHM) horizontally and 434 m (FWHM)
vertically. The destructive interference fringes can clearly be
recognized. Right panel, upper curves: Beam spot size about
500 m (FWHM) horizontally and 434 m (FWHM) vertically.
Right panel, lower curves: Reduced beam spot size 114 m
(FWHM) horizontally and 200 m (FWHM) vertically.
The basic idea of the experiment will be explained by
means of g. 14. A silicon single-crystal target was po-
sitioned in such a way that the PXR reex at a photon
energy h0 = 10.554 keV is located at twice the Bragg
angle 0 = 10.797 in the horizontal plane of drawing.
The radiation in forward direction close to the electron
direction was analyzed with a at silicon single-crystal
monochromator in combination with a pn CCD camera as
a position-sensitive and energy-resolving photon detector.
The quasi-monochromatic FDPXR peak energy matches
with the energy of the analyzer crystal at only one specic
The European Physical Journal A
Ne pixel 1012
Ne pixel 1012
0.2
1 0.1
0.5
0 0
0 50 100 150 200 0 50 100 150 200
x Column Number Column Number
Ne pixel 1012 
Ne pixel 1012 
2 0.2
1.5
0
1 0.1
1 0.5
x 0
0 50 100 150 200
0
0 50 100 150 200
Column Number Column Number
Ne pixel 1012
Ne pixel 1012
2 0.2
1.5
1 0.1
0.5
0 0
0 50 100 150 200 0 50 100 150 200
Column Number Column Number
Fig. 16. Results of simulation at photon energy of 10.554 keV
and target thickness of 58 m (left) and 1000 m (right). Shown
are the number of photons N per pixel and electron for one
row of the pn CCD detector as a function of the column num-
ber. From the upper to the lower panel the rotation angle of the
target crystal x was varied in steps of x = 0.5894 mrad.
The beam spot sizes of the experiment and scattering of the
electron beam were taken into account. The residual interfer-
ence oscillations originate from the interference of the remain-
ing 4% amplitude created at the entrance interface with the
amplitude at the exit interface of the crystal. However, these
100 150 200
oscillations are smoothed out in a real experiment in which
Column summation is made over several or all rows of the pn CCD
detector.
observation angle x . Since its reecting power of the crys-
tal monochromator exhibits energetically a narrow-band
characteristics, quasi-monochromatic intensity structures
emitted from the target crystal can be detected by this
experimental arrangement.
Experiments were performed with target crystals of
varying thickness and for dierent photon energies. As an
example the intensity distributions of the experiment with
58 m and 1 mm crystal thicknesses are shown in g. 15.
The most striking features are the structures which move
across the pn CCD detector if the rotation angle x of the
target crystal around the vertical y-axis is varied. These
structures are for the thin targets interferences of the ra-
diation amplitudes created at the entrance and exit inter-
faces of the crystal which originate from a resonance in
the dispersion surface of the electron in the crystal. The
pronounced peak structures observed for the thick target
which are clearly correlated to the interference structures
of the thin targets are interpreted as FDPXR contribu-
tions to the smooth transition radiation background from
the downstream interface of the target crystal.
The interference structures can quantitatively be ex-
plained in the framework of the well-known TR production
 W. Lauth et al.: Coherent X-rays at MAMI
mechanism utilizing a generalized formation length for
crystalline matter [66]. Within this model, which is based
on the formalism described in ref. [67], the resonance is
connected to forward-diracted PXR (FDPXR). Calcu-
lated emission spectra of this model, including multiple
scattering of the electrons, are displayed in g. 16. The
simulations are in good agreement with the experimental
observation.
4 Conclusion and outlook
A novel interferometer for soft X-rays has been developed.
Intensity oscillations with a high degree of coherence have
been observed, not only at the K absorption edge of car-
bon as described in ref. [18] but with the third harmonics
of the undulator radiation also at the L absorption edge of
Ni around 865 eV. The polarized undulator radiation can
be used to investigate the magneto-optical properties of
3d transition metals. First experiments to determine the
X-ray magnetic circular circular dichroism look promis-
ing. The energy band between 50 eV and about 1500 eV,
which can be covered with the interferometer at MAMI,
allows the investigation of many elements throughout the
periodic table with samples of masses as low as 10 ng.
A novel X-ray interferometer has been developed for
hard X-rays. The good agreement within experimental er-
rors of our measurements with that of Bonse et al. [39,40]
proves that it is fully operational. A simultaneous preci-
sion measurement of both, 1 () and 1 () of the sample
foil, should be possible if the foil thicknesses are optimized
properly. A possible application of this type of interferom-
eter originates from the transient state of matter which
can be produced by multi GeV bunches of about 1 nC
charge and a duration in the 400 fs range. Quite unusual
properties are connected with such bunches if focused to
a radius " of a few m, e.g. [68]. A strong static electric
eld of E0 = 16/("m) V/A is present at the periphery
of the charge distribution. In addition, the high electrical
current of 2500 A produces a strong magnetic eld B = G. Wagner, Th. Walcher, M. Kretzschmar, Phys. Rev. E
540/("m) T. During the passage of such a bunch through
a thin foil the matter is put in a transient state which is
characterized by a high dielectric polarization in the pres-
ence of a strong magnetic eld at probably a rather large
non-equilibrium electron temperature. This state can, in
principle, be investigated by our novel interferometer.
PXR emission from silicon single crystal slabs has been
investigated with the electron beam of MAMI. The re-
sults of the line width measurements in backward geom-
etry shows that the PXR line widths in Si crystals are
superior to the Darwin-Prins widths only for the (111)
and (333) reections. Line broadening eects due to mul-
tiple scattering spoil the predicted outstanding resolution
for higher-order reections. With crystals from low-Z ma-
terials, like diamond or LiH, the small-angle scattering is
reduced and much narrower lines may be expected as for
silicon single crystals.
The radiation from silicon single crystals, emitted close
to the electron beam direction, has been studied to eluci-
date the discussion of kinematical versus dynamical pro-
193
duction of PXR. For thin crystals pronounced interference
structures in forward direction have been observed when
about a Bragg condition was fullled. This interference is
corroborated by the narrow FDPXR lines observed with a
thick Si crystal for which essentially only the exit interface
of the target crystal contributes to the observed intensity.
Our experimental results show that PXR production is a
dynamical rather than a kinematical process.
We thank A. Steinhof for substantial contributions in the re-
design and reconstruction of the undulator interferometer and
N. Clawiter, S. Dambach, Th. Doerk, M. El-Ghazaly, F. Hagen-
buck, G. Kube, A. Rueda and D. Schro for their help during
the course of the experiments described in this contribution.
This work has been supported by Deutsche Forschungsgemein-
schaft DFG under contract BA 1336/1-4.
References
1. H. Backe, K.H. Brenzinger, F. Buskirk, S. Dambach, Th.
Doerk, N. Eftekhari, H. Euteneuer, F. Gorgen, C. Her-
berg, F. Hagenbuck, K. Johann, K.H. Kaiser, O. Kettig,
G. Knies, G. Kube, W. Lauth, B. Limburg, J. Lind, H.
Schope, G. Stephan, Th. Walcher, Th. Tonn, R. Zahn,
in R.L. Johnson, H. Schmidt-Bocking, B.F. Sonntag (Edi-
tors), X-Ray and Inner-Shell Processes: 17th International
Conference, AIP Conf. Proc. 389 (AIP Press, Woodbury,
New York, 1997).
2. F. Hagenbuck, H. Backe, N. Clawiter, H. Euteneuer, F.
Gorgen, P. Holl, K. Johann, K.-H. Kaiser, J. Kemmer,
Th. Kerschner, O. Kettig, H. Koch, G. Kube, W. Lauth,
H. Matthay, M. Schutrumpf, R. Stotter, L. Struder, Th.
Walcher, A. Wilms, C.v. Zanthier, M. Zemter, IEEE Trans.
Nucl. Sci. 48, 843 (2001).
3. M. El-Ghazaly, H. Backe, W. Lauth, G. Kube, P. Kunz,
A. Sharafutdinov, T. Weber, these proceedings.
4. G. Kube, H. Backe, H. Euteneuer, A. Grendel, F. Hagen-
buck, H. Hartmann, K.H. Kaiser, W. Lauth, H. Schope,
65, 056501 (2002).
5. H. Backe, W. Lauth, H. Mannweiler, H. Rochholz, K.
Aulenbacher, R. Barday, H. Euteneuer, K.-H. Kaiser,
G. Kube, F. Schwellnus, V. Tioukine, in Proceedings
of NATO Advanced Radiation Sources and Applications,
Nor-Hamberd, Yerevan, Armenia, 2004, edited by H.
Wiedemann, NATO Sci. Ser. II, Vol. 199 (Springer,
Dortrecht, 2006) p. 267.
6. H. Backe, W. Lauth, A. Rueda, M. El-Ghazaly, P. Kunz,
A. Picard, A. Scharafutdinov, A. Sossalla, T. Weber in
Topics in Heavy Ion Physics - Proceedings of the Memorial
Symp. for G. So, Frankfurt, Germany, 2005, edited by
W. Greiner, J. Reinhardt, (EP Systema Bt., Budapest,
2005) p. 1.
7. A.V. Korol, A.V. Solovyov, W. Greiner, Int. Jour. Mod.
Phys. E-Nucl. Phys. 13, 867 (2004).
8. G. Materik, C.J. Sparks, K. Fischer (Editors), Resonant
Anomalous X-ray Scattering (North Holland, Amsterdam,
London, New York, Tokyo, 1994).
9. J. Stohr, NEXAFS Spectroscopy, Springer Ser. Surface Sci.,
Vol. 25 (Springer, New York, 1992).
194 The European Physical Journal A
10. B. Lengeler, in [8], p. 35.
11. U. Bonse, M. Hart, Appl. Phys. Lett. 7, 238 (1965).
12. W.K. Warburton, K.F. Ludwig, Phys. Rev. B 33, 8424
(1986).
13. M. Deutsch, M. Hart, Phys. Rev. B 30, 643 (1984).
14. A. Freund, in Anomomalous Scattering, edited by R.
Ramaseshan, S.C. Abrahams (Munksgaard Copenhagen,
1975) p. 69.
15. N. Kato, S. Tanemura, Phys. Rev. Lett. 19, 22 (1967).
16. R.L Blake, J.C. Davis, D.E. Graessle, T.H. Burbine, E.M.
Gullikson, in [8], p. 79.
17. F. Polack, D. Joyeux, J. Svatos, D. Phalippou, Rev. Sci.
Instrum. 66, 2180 (1995).
18. S. Dambach, H. Backe, Th. Doerk, N. Eftekhari, H. Eu-
teneuer, F. Gorgen, F. Hagenbuck, K.H. Kaiser, O. Kettig,
G. Kube, W. Lauth, H. Schope, A. Steinhof, Th. Tonn, Th.
Walcher, Phys. Rev. Lett. 80, 5473 (1998).
19. M. Itou, T. Harada, T. Kita, Appl. Optics 28, 146 (1989).
20. R. Soui, E.M. Gullikson, Appl. Opt. 36, 5499 (1997).
21. N. Kerr Del Grande, Physica Scripta 41, 110 (1990).
22. J. Stohr, J. Magn. Magn. Mater. 200, 470 (1999).
23. G. Schutz, W. Wagner, W. Wilhelm, P. Kienle, R. Zeller,
R. Frahm, G. Materlik, Phys. Rev. Lett. 58, 737 (1987).
24. J.B. Kortright, Sang-Koog Kim, Phys. Rev. B 62, 12216
(2000).
25. http://www.andor-tech.com/germany/products/oem.
cfm.
26. H. Soltau, P. Holl, J. Kemmer, S. Krisch, C.v. Zanthier, D.
Hau, R. Richter, H. Brauninger, R. Hartmann, G. Hart-
ner, N. Krause, N. Meidinger, E. Pfeermann, C. Reppin,
G. Schwaab, L. Struder, J. Trumper, E. Kendziorra, J.
Kramer, Nucl. Instrum. Methods A 377, 340 (1996).
27. M.L. Cherry, G. Hartmann, D. Muller, T.A. Prince, Phys.
Rev. D 10, 3594 (1974).
28. M.J. Moran, B.A. Dahling, P.J. Ebert, M.A. Piestrup, B.L.
Bergman, J.O. Kephart, Phys. Rev. Lett. 57, 1223 (1986).
29. P. Goedtkind, J.-M. Salome, X. Atru, P. Dhez, M.
Jablonka, N. Maene, F. Poortmans, L. Wartski, Nucl. In-
strum. Methods B 56/57, 1060 (1991).
30. M.A. Piestrup, D.G. Boyers, C.I. Pincus, Qiang Li, G.D.
Hallewell, M.J. Moran, D.M. Skopik, R.M. Silzer, X.K.
Maruyama, D.D. Snyder, G.B. Rothbart, Phys. Rev. A
45, 1183 (1992).
31. V.V. Kaplin, V.N. Zabaev, E.I. Rozum, S.R. Uglov, S.A.
Vorobiev, Phys. Lett. A 174, 165 (1993).
32. H. Backe, S. Gampert, A. Grendel, H.J. Hartmann, W.
Lauth, Ch. Weinheimer, R. Zahn, F.R. Buskirk, H. Eu-
teneuer, K.H. Kaiser, G. Stephan, Th. Walcher, Z. Phys.
A 349, 87 (1994).
33. O. Kettig, H. Backe, N. Clawiter, S. Dambach, Th. Do-
erk, N. Elbai, H. Euteneuer, F. Hagenbuck, P. Holl, H. Ja-
cobs, K.H. Kaiser, J. Kemmer, Th. Kerschner, G. Kube, H.
Koch, W. Lauth, H. Mannweiler, H. Matthay, H. Schope,
D. Schro, M. Schuttrumpf, R. Stotter, L. Struder, Th.
Walcher, A. Wilms, C.v. Zanthier, M. Zemter, in R.W.
Dunford, D.S. Gemmell, E.P. Kanter, B. Krassig, S.H.
Southworth, L. Young (Editors) X-ray and Inner-Shell
Processes: 18th International Conference, AIP Conf. Proc.
506 (AIP Press, Melville, New York, 2000).
34. O. Kettig, Dissertation, Institut fur Kernphysik, Univer-
sitat Mainz, 2000 (in German).
35. H. Backe, N. Clawiter, S. Dambach, H. Euteneuer, F. Ha-
genbuck, K.-H. Kaiser, O. Kettig, G. Kube, W. Lauth, Th.
Walcher, in Proceedings of the International Conference
on Fundamenttal and Applied Aspects of Modern Physics,
Luderitz, Namibia, 2000, edited by S.H. Connell, R. Tegen
(World Scientic, New Jersey, London, Singapore, Hong
Kong, 2001) p. 123.
36. O. Kettig et al., to be published.
37. B.L. Henke, E.M. Gullikson, J.C. Davis, Atom. Data and
Nucl. Data Tabl. 54, 181 (1993).
38. E. Storm, H.I. Israel, Atom. Data and Nucl. Data Tabl. A
7, 565 (1970).
39. U. Bonse, G. Materlik, Z. Phys. B 24, 189 (1976).
40. U. Bonse, I. Hartmann-Lotsch, Nucl. Instrum. Methods
222, 185 (1984).
41. V.G. Baryshevsky, Dokl. Akad. Nauk BSSR 15, 306
(1971).
42. V.G. Baryshevsky, I.D. Feranchuk, Zh. Exper. Teor. Fiz.
61, 944 (1971); (Sov. Phys. JETP 34, 50 (1972); adden-
dum, ibid 64, 760 (1973)).
43. G.M. Garibian, C. Yang, Zh. Eksp. Teor. Fiz. 61, 930
(1971), (Sov. Phys. JETP 34, 495 (1972)).
44. G.M. Garibian, C. Yang, Zh. Eksp. Teor. Fiz. 63, 1198
(1972), (Sov. Phys. JETP 36, 631 (1973)).
45. M. Ter-Mikaelian, High-Energy Electromagnetic Processes
in condensed Media (Wiley-Interscience, New York, Lon-
don, Sydney, Toronto, 1972).
46. K.-H. Brenzinger, C. Herberg, B. Limburg, H. Backe, S.
Dambach, H. Euteneuer, F. Hagenbuck, H. Hartmann,
K. Johann, K.H. Kaiser, O. Kettig, G. Knies, G. Kube,
W. Lauth, H. Schope, Th. Walcher, Z. Phys. A 358, 107
(1997).
47. J. Freudenberger, H. Genz, V.V. Morokhovskyi, A.
Richter, J.P.F. Sellschop, Phys. Rev. Lett. 84, 270 (2000).
48. H. Backe, G. Kube, W. Lauth, Proceedings of NATO Ad-
vanced Research Workshop on Electron-Photon Interac-
tion in Dense Media, Nor-Hamberd, Yerevan, Armenia,
2001, edited by H. Wiedemann, NATO Sci. Ser. II, Vol. 49
(Kluwer Academic Publishers, Dordrecht, Boston, London,
2001) p. 153.
49. 5th International Symposium on Radiation from Relativis-
tic Electrons in Periodic Structures (RREPS-01), Lake
Aya, Russia, September, 2001, Topical Issue of Nucl. In-
strum. Methods Phys. Res. B 201, 1 (2003).
50. X. Artru, P. Rullhusen, Nucl. Instrum. Methods Phys. Res.
B 145, 1 (1998); addendum, ibid 173, 16 (2001).
51. V. Baryshevsky, I. Feranchuk, J. Phys. (Paris) 44, 913
(1983).
52. I. Feranchuk, A. Ivashin, J. Phys. (Paris) 46, 1981 (1985).
53. H. Nitta, Phys. Lett. A 158, 270 (1991).
54. A. Caticha, Phys. Rev. B 45, 9541 (1992).
55. A.V. Shchagin, V.I. Pristupa, N.A. Khizhnyak, Phys. Lett.
A 148, 485 (1990).
56. J. Freudenberger, V.B. Gavrikov, M. Galemann, H. Genz,
L. Groening, V.L. Morokhovskii, V.V. Morokhovskii, U.
Nething, A. Richter, J.P.F. Sellschop, N.F. Shulga, Phys.
Rev. Lett. 74, 2487 (1995).
57. O.M. Lugoskaya, S.A. Stepanov, Sov. Phys. Crystallogr.
36, 478 (1991) and http://sergey.gmca.aps.anl.gov/
cgi/X0h.html.
58. J. Freudenberger, H. Genz, V.V. Morokhovskii, A. Richter,
V.L. Morokhovskii, U. Nething, R. Zahn, J.P.F. Sellschop,
Appl. Phys. Lett. 70, 267 (1997).
59. K.-H. Brenzinger, B. Limburg, H. Backe, S. Dambach,
H. Euteneuer, F. Hagenbuck, C. Herberg, K.H. Kaiser,
 W. Lauth et al.: Coherent X-rays at MAMI
O. Kettig, G. Kube, W. Lauth, H. Schope, Th. Walcher,
Phys. Rev. Lett. 79, 2462 (1997).
60. H. Backe, C. Ay, N. Clawiter, Th. Doerk, M. El-Ghazaly,
K.-H. Kayser, O. Kettig, G. Kube, F. Hagenbuck, W.
Lauth, A. Rueda, A. Scharafutdinov, D. Schro, T. Weber
in: W. Greiner, A. Solovyov, S. Misicu (Editors) Proceed-
ings Symp. Channeling - Bent Crystals - Radiation Pro-
cesses, Frankfurt (Germany) 2003 (EP Systema, Debre-
cen, 2003) p. 41.
61. H. Nitta, J. Phys. Soc. Jpn. 69, 3462 (2000).
62. V.G. Baryshevsky, Nucl. Instrum. Methods Phys. Res. B
122, 13 (1997).
63. A. Kubankin, N. Nasonov, V. Sergienko, I. Vnukov, Nucl.
Instrum. Methods Phys. Res. B 201, 97 (2003).
195
64. N. Nasonov, V. Sergienko, N. Noskov, Nucl. Instrum.
Methods Phys. Res. B 201, 67 (2003).
65. N. Aleinik, A.N. Baldin, E.A. Bogomazova, I.E. Vnukov,
B.N. Kalinin, A.S. Kubankin, N.N. Nasonov, G.A. Nau-
menko, A.P. Potylitsyn, A.F. Sharafutdinov, JETP Lett.
80, 393 (2004) (Pisma Zh. Eksp. Teor. Fiz. 80, 447
(2004)).
66. H. Backe, A. Rueda, W. Lauth, N. Clawiter, M. El-
Ghazaly, P. Kunz, T. Weber, Nucl. Instrum. Methods
Phys. Res. B 234, 138 (2005).
67. A. Caticha, Phys. Rev. A 40, 4322 (1989).
68. C.D. Back, D. Weller, J. Heidmann, D. Mauri, D. Guarisco,
E.L. Garwin, H.C. Siegmann, Phys. Rev. Lett. 81, 3251
(1998).
Eur. Phys. J. A 28, s01, 197 208 (2006)
DOI: 10.1140/epja/i2006-09-021-6 EPJ A direct
electronic only

X-ray phase contrast imaging at MAMI

M. El-Ghazalya , H. Backe, W. Lauth, G. Kubeb , P. Kunz, A. Sharafutdinov, and T. Weber

Institut fur Kernphysik der Universitat Mainz, D-55099 Mainz, Germany

/
Published online: 6 June 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. Experiments have been performed to explore the potential of the low emittance 855 MeV electron
beam of the Mainz Microtron MAMI for imaging with coherent X-rays. Transition radiation from a micro-
focused electron beam traversing a foil stack served as X-ray source with good transverse coherence.
Refraction contrast radiographs of low absorbing materials, in particular polymer strings with diameters
between 30 and 450 m, were taken with a polychromatic transition radiation X-ray source with a spectral
distribution in the energy range between 8 and about 40 keV. The electron beam spot size had standard
deviation h = (8.6 0.1) m in the horizontal and v = (7.5 0.1) m in the vertical direction. X-ray
lms were used as detectors. The source-to-detector distance amounted to 11.4 m. The objects were placed
in a distance of up to 6 m from the X-ray lm. Holograms of strings were taken with a beam spot size
v = (0.50 0.05) m in vertical direction, and a monochromatic X-ray beam of 6 keV energy. A good
longitudinal coherence has been obtained by the (111) reection of a at silicon single crystal in Bragg
geometry. It has been demonstrated that a direct exposure CCD chip with a pixel size of 13 13 m 2
provides a highly e cient on-line detector. Contrast images can easily be generated with a complete
elimination of all parasitic background. The on-line capability allows a minimization of the beam spot
size by observing the smallest visible interference fringe spacings or the number of visible fringes. It has
been demonstrated that X-ray lms are also very useful detectors. The main advantage in comparison
with the direct exposure CCD chip is the resolution. For the Structurix D3 (Agfa) X-ray lm the standard
deviation of the resolution was measured to be f = (1.2 0.4) m, which is about a factor of 6 better
than for the direct exposure CCD chip. With the small eective X-ray spot size in vertical direction of
v = (1.20.3) m and a geometrical magnication of up to 7.4 high-quality holograms of tiny transparent
strings were taken in which the holographic information is contained in up to 18 interference fringes.

PACS. 87.59.Bh X-ray radiography 52.59.Px Hard X-ray sources 41.50.+h X-ray beams and X-ray
optics 07.85.Fv X- and gamma-ray sources, mirrors, gratings, and detectors 07.85.Nc X-ray and
gamma-ray spectrometers

1 Introduction
The contrast in conventional absorption X-ray imaging is
based on the dierence in the absorption of the materials
constituting the sample. Thin samples of light elements,
such as soft tissues and organic materials with Z 8, show
a weak absorption contrast even at low X-ray energies, i.e.,
the big deciency is that the conventional absorption ra-
diography cannot distinguish between materials with simi-
lar attenuation coe cients. For low-Z materials, however,
a high contrast could be obtained if the phase shift of
the X-rays introduced by the object could be exploited
instead of the intensity of the transmitted wave. The en-
hancement of the contrast is attributed to the fact that, in
a
Former PhD Scholarship Holder in the Long Term Mission
System from the Arabic Republic of Egypt.
b
Present address: Deutsches Elektronen-Synchrotron
DESY, Notkestra e 85, D-22603 Hamburg, Germany.
particular for low-Z materials, the phase shift for X-rays
is higher than the absorption of the incident X-rays. Also,
for the radiography based on the phase shift mechanism,
the absorbed dose is considerably lower in comparison to
the conventional absorption radiography, see, e.g., refs. [1,
2,3].
X-ray phase contrast imaging can be carried out with
various methods, for an overview see the recent ref. [4]. In
particular, it has been pointed out by Wilkins et al. [5]
that a very simple experimental setup with a polychro-
matic X-ray source of good transverse coherence, i.e. a
small micro-focused spot, is already su cient. Informa-
tion can be supplied by such a method on the sample
morphology, i.e. its boundaries, interfaces and location of
small features, see e.g. ref. [6,7,8]. If, in addition, the X-
ray source emits monochromatic X-rays, holograms can
be taken. The experimental setup is similar to that of
Gabor in-line holography [9]. In principle, such a setup
198 The European Physical Journal A
is rather simple but a highly transverse and longitudi-
nal coherent X-ray source of good intensity and also high
spatial resolution detectors are required. Such sources are
available at third generation synchrotron radiation sources
like ESRF, APS, and SPRING8, and hard X-ray phase
contrast imaging, in-line holography and microtomogra-
phy have been accomplished at these facilities, see, e.g.,
refs. [10,11,12].
The work presented here exploits the potential of the
low-emittance 855 MeV electron beam of the race track
microtron MAMI to produce X-rays with very good trans-
verse coherence. Our approach is based on transition radi-
ation (TR) production in the X-ray region with a micro-
focused electron beam. In sect. 2 some features of the com-
plex refraction index will be recalled with particular view
on phase contrast methods. In sect. 3 the results of phase
contrast imaging with a polymonochromatic X-ray beam
from a TR foil stack with good transverse coherence will
be presented. Section 4 deals with our approach toward
a hard X-ray in-line holography using monochromatic X-
rays. The paper closes with a conclusion.
2 Absorption versus phase shift
When a parallel beam of X-rays penetrates matter, it suf-
fers an attenuation and a phase shift. These macroscopic
quantities are described by the complex refraction index
of X-rays [13]
n() = 1 () + i() . (1)
The real part [n()] = 1 () describes the refrac-
tion of the wave of angular frequency in a material, the
quantity () gives the deviation of the refractive index of
a material from unity (refraction index of vacuum). It is
called the refractive index decrement. The imaginary part
[n()] = () species the attenuation of the X-rays in
matter. It is called the absorption index.
The transmission of an electromagnetic wave through
a piece of matter of thickness d is illustrated schematically
in g. 1. The undisturbed wave propagation in x -direction
is described by the expression
Av = A0 ei(kv dt) ; kv = . (2)
c
The amplitude of the outgoing wave behind the object is
Am = A0 ei(km dt) = A0 ei c d e c d ei(kv dt) , (3) The phase shift of the outgoing wave relative to the wave
where km = n/c is the wave number in the medium.
Equation (3) contains a phase factor exp(i(d)) with
(d) = m (d) v (d) = (/c) d the phase dierence
between the wave in matter with phase m , and in vacuum
with phase v . In addition, the wave suers an amplitude
attenuation |Am |/|Av | = exp[(/c) d].
The ratio / is drastically larger for a low-Z material
in comparison with a high-Z material at photon energies
in the order of 20 40 keV. For example, at a photon energy
!
Av d 3KDVH6KLIW
$WWHQXDWLRQ
Am
x
Fig. 1. Transmission of an electromagnetic wave through a
piece of matter of thickness d and complex refraction index
n = 1+i. The transmitted wave is attenuated by the factor
exp[(/c)d] and has suered a phase shift m (d)v (d) =
(/c) d with respect to the unperturbed vacuum wave.
h = 30 keV, / for polycarbonate (C14 H14 O3 ) is ap-
proximately 40 times larger than for nickel (Z  28). This
clearly demonstrates that for low-absorbing materials the
phase shift dominates in comparison with the attenuation.
For a polycarbonate foil of 10 m thickness and a photon
energy h = 12 keV, for which the complex refraction in-
dex parameters are = 1.826 106 and = 1.573 109 ,
the phase shift is = 1.11 rad while the intensity attenu-
ation is only 1 exp[(2/c) d] = 1.91 103 . These
considerations lead to the important conclusion that a
high contrast combined with a low absorbed dose could
be achieved by using the phase shift mechanism to pro-
duce a radiograph [5]. In the next section the question
will be addressed how the phase shift can be exploited for
radiography.
3 Refraction contrast radiography
3.1 Basics
In an ideal experiment a point source emanates a
monochromatic wave and illuminates the sample. The X-
ray wavefront impinging on a sample will be deformed at
the passage through the medium when its thickness or re-
fractive index is inhomogeneous. In the framework of the
eikonal approximation the wave vectors of the X-rays are
normal to the equi-phase surfaces. In this picture of ray
optics, i.e. for 0, the deviation from the initial direc-
tion is due to refraction.
For the sake of simplicity, in g. 2 a one-dimensional
object such as a string of radius R and a homogeneous
refraction index n2 = 1 2 + i2 is considered, which is
embedded in a medium of refraction index n1 = 1 1 +
i1 . It is illuminated with a nearly parallel X-ray beam.
in vacuum is given by
# ! z "2
4 o
(z0 ) = (2 1 )R 1 , (4)
R
with zo the vertical coordinate at the object. The angular
deviation of the normal to the incoming wavefront is, in
the eikonal approximation,
 
 (z0 )  2(2 1 ) zo
= = 1 . (5)
2  zo  R [1 ( zRo )2 ] 2
 M. El-Ghazaly et al.: X-ray phase contrast imaging at MAMI 199

]
Object Intensity
,PD[
( ; [ , ] ), 0 ,PLQ

Detector
S

Source

[VR [RG

Fig. 2. Formation of a refraction contrast radiograph accord-

ing to geometrical ray optics. Refracted X-rays slightly deviate
from the initial propagation direction at the interfaces in accor-
dance with Snells law of refraction. Since the refraction index
for X-rays is slightly smaller than unity (about 10 6 ), X-rays
are refracted in opposite manner to visible light, i.e. they are
focused by a concave and defocused by a convex object. For
tangentially incidence the X-rays encounter maximum devia-
tion resulting in the formation of a contrast which enhances the
visibility of the interfaces. The source-to-object distance is xso ,
the object-to-detector distance xod and the source-to-detector
distance xsd = xso + xod .

The phase gradient diverges at zo = R. The rays deviate

by a large angle from the original propagation direction
even though (2 1 ) is very small as in the case of X-rays,
which leads to a loss of intensity at boundaries or an edge
contrast. This explains why the radiograph looks like a di-
rect image of contours of the details which constitute the
sample. More generally, any rapid variation of the refrac-
tion index or the thickness of the sample may be imaged
by the edge contrast which appears in the radiograph even
when a polychromatic X-ray beam is used.
In fact, the wave refracted by the sample interferes
with the unperturbed wave. The diracted wave and
the unscattered wave form an interference pattern at
the detector which is called a hologram. It is recorded
by an image detector of high spatial resolution. For the
polymer string with circular cross-section stretched along
the horizontal yo -axis, the normalized electric wave eld
E(zd , )/E0 (zd , ) at the detector plane can be calcu-
lated by means of the Fresnel-Kirchho integral. As shown
in [14,15] the result is
# malized intensity distributions g(zs ) of the beam spot and
E(zd , ) xsd
=1+
E0 (zd , ) ixso xod
+R
#  
4  z2
exp i + R 1 o2 1 (7)
R
R It can be seen from g. 3 (b) that a few interference

xsd xso 2
exp i (zo zd ) dzo .
xso xod xsd
The quantities xso , xod , and xsd are dened in g. 2.
The normalized intensity distribution perpendicular to
the string direction, the zd -direction at the detector
()
plane, In (zd , 0 ) = |E(zd , 0 )|2 /|E0 (zd , 0 )|2 is shown in
g. 3 (a) for a monochromatic X-ray source with wave-
length 0 . In a real experiment the spectral distribution
of the X-rays and the nite beam spot size deteriorate
Fig. 3. Calculated interference patterns for a polymer string
with a diameter of 30 m, complex refraction index =
7.24 10 6 and = 2.42 10 8 at an X-ray energy of 6 keV,
source-to-object distance xso = 10.45 m, object-to-detector
distance xod = 3.15 m. (a) Normalized intensity distribu-
tion |E(zd , 0 )|2 /|E0 (zd , 0 )|2 derived from eq. (6) for a point
source and monochromatic X-rays with 0 = 2.067 A, corre-
sponding to an energy of 6 keV; (b) for a point source but a
Gaussian spectral distribution around 0 with standard devi-
ation of = 0.6 A; (c) for a Gaussian intensity distribution
of the X-ray source spot with standard deviation z = 6 m
and monochromatic X-rays. Convolutions according to eq. (7).
Ideal detector resolution is assumed.
the visibility of the interference fringes as demonstrated in
g. 3 (b) and (c). Source spot size and spectral distribution
have been taken into account as convolutions with the nor-
the spectral distribution f () of the X-rays according to
  
xod
In (zd ) = In()
zd zs , 0 g(zs )f ()dzs d .
xso
fringes remain visible, which resemble the structure of the
string, even for polychromatic X-rays with a small longi-
(6)
tudinal coherence length LL = 2 /(2) , and/or a
beam spot size z = 6 m (rms) which corresponds to a
transverse coherence length LT = xsd /(2z ) = 75 m.
Radiography based on these conditions will be called in
the following refraction contrast radiography. For X-ray
in-line holography, both a very good transverse coherence
and a good longitudinal coherence of the X-ray beam are
required since the holographic information is imprinted in
the interference pattern as shown in g. 3 (a).
200
30
Photons [1/(e 10 BW sr)]
20
-3
-
10
0
0 10 20 30 40
[keV]
e-
Quadrupole Foil
doublet stack
855 MeV TR chamber
Electron xso
beam
Fig. 4. Schematic diagram showing the experimental setup for
refraction contrast radiography. The inset shows the calculated
TR spectrum as function of the photon energy for which mul-
tiple scattering, electron beam divergence (0.6 mrad) and self
absorption were taken into account. The foil stack consists of
30 polyimide foils with a thickness of 25 m each, and spac-
ings between the foils of 75 m. It was optimized for a photon
energy of 33 keV.
3.2 Experimental
The principle of the refraction contrast radiography will be
explained by means of g. 4. The 855 MeV electron beam,
with a Lorentz factor = 1673, produces in a transition
radiation foil stack a polychromatic X-ray beam which
propagates in the forward direction in a cone with a typ-
ical apex angle of 2/  1.2 mrad. The X-ray emission
spectrum is shown in the inset of g. 4. The polychro-
matic X-rays leave the vacuum system through a poly-
imide exit window of 120 m thickness which is located
at a distance of 5.88 m from the foil stack. The beam
line is shielded by a concrete wall of 1 m thickness and
3.5 m height to reduce the background in the experimen-
tal area. The background originates from electrons which
emitted a bremsstrahlung photon in the TR foil stack and
left the beam line behind the bending magnet, as well as
the background from the beam dump itself. The objects
to be imaged are mounted in air at dierent distances
from the target xso , and from the X-ray lm xod , with
5.88 m < xso < 13 m and 0 m < xod < 7.12 m, respec-
tively. The source-to-detector distance was xsd = 11.38 m.
The Mamoray MR5 II PQ X-ray lm produced by
Agfa1 was used as position-sensitive detector. It is based
on silver bromide with an emulsion thickness of df =
12 m [16]. The exposed X-ray lms were processed man-
ually. The X-ray lms were digitized with a Nikon lm
scanner Super CoolScan 4000 ED [17] which has a spatial
resolution of 4000 dpi 2 corresponding to a pixel size of
(6.35 6.35) m2 .
The primary quantity which is measured by an X-ray
lm is the photographic density Dp [18,16]. It is dened
with the basis-10 logarithm as Dp = log(i0 /i) with i0
1
Agfa-Gevaert N.V., B2640 Mortsel Belgium.
2 3
Dots per inch.
The European Physical Journal A
the optical light intensity impinging on the lm and i
the intensity measured by the detector of the densitome-
ter. From this primary quantity the so-called fog Df =
log(i0 /i0f ) of an unexposed part of the lm must be sub-
tracted to obtain the density D = Dp Df = log(i0f /i).
The latter must be related to the exposure b(r) of the lm,
Objects
i.e. the energy per unit area dE(r)/dA deposited by the
50 X-ray photons at a certain location r of the lm.
to beam X-ray In a simple theoretical model [18] the photographic
dump film density can be described by
X-rays D(r) = Dsat (1 exp(b(r)/b0 )) . (8)
xod
The saturation density Dsat and b0 are characteristic
5m
quantities of the X-ray lm. From eq. (8) the relative ex-
posure is obtained as
 
b(r) Dsat
= ln . (9)
b0 Dsat D(r)
Since we are interested in normalized exposure ratios
(b(r)/b0 )/(b/b0 ) with b the value without the object which
can be replaced by a mean value on some position outside
the domain of interest, the unknown quantity b0 cancels.
The still unknown saturation density Dsat must, in princi-
ple, be determined. However, since the digitization devices
to our disposal had only a depths of 8 bits the main re-
striction in the dynamical range is expected to originate
from the digitization procedure and not from the dynam-
ical range of the X-ray lm.
In view of the low digitization depth, the procedure we
adapted to obtain the contrast Cref of a string as dened
by eq. (10) in the next subsection was the following. At
rst, domains on the X-ray lm were selected in which the
photographic density was assumed to be in the linear re-
gion. In this case eq. (8) reduces to D(r) = Dsat b(r)/b0
and in the exposure ratio also the saturation density can-
cels. Thereafter, we determined the contrast Cref at var-
ious positions of the string for which the exposure varied
due to the intensity prole of the X-ray beam spot and se-
lected the maximum value as the experimental contrast.
3.3 Results
An extensive study of the contrast generation as a func-
tion of the object-to-detector distance xod has been per-
formed for polyamide strings of dierent diameters. Fig-
ure 5 (a) shows a typical example for a polyamide string 3
with a diameter of about 270 m. The calculated absorp-
tion contrast for such a polyamide string does not exceed
about 1%. Therefore, no absorption contrast can be ob-
served with the traditional contact radiography, i.e. for
xod = 0, in accord with our measurements. By moving
the object away from the detector, the imaging regime is
changed from absorption radiography to phase contrast
radiography and phase shift is the mechanism to produce
the contrast. The contrast appears at the borders of the
Supplied by Goodfellow.
 M. El-Ghazaly et al.: X-ray phase contrast imaging at MAMI
Fig. 5. Refraction-enhanced radiograph of a polyamide string
with a diameter of 270 m at an object-to-detector distance
xod = 5.5 m and a source-to-detector distance xsd = 11.38 m.
The electron beam current was 6 nA, the exposure time
amounted to 60 s. X-ray source sizes were h = (8.6 0.1) m
and v = (7.5 0.1) m in horizontal and vertical direction,
respectively. (a) Radiograph, (b) intensity prole for which
100 vertical pixels were added together to improve the statis-
tics. (c) Normalized intensity prole according to geometrical
optics with the following parameters: lm and scanner reso-
lution t =(10.0 0.4) m, and the wave optical contribu-
tion w = xsd xod /(2xso ) = 2.3 m with = 0.633 A. account that with increasing xod the edge spread increases
(d) Same as (c) on the basis of wave optics.
polyamide string where the density gradient reaches its
maximum value. An edge contrast can be dened as
Imax Imin
Cref = (10)
Imax + Imin
with Imax and Imin dened in g. 2. As can be seen from
g. 5 (b) the contrast amounts to Cref = 17.8%. The con-
trast Cref as a function of the object-to-detector distance
xod is shown in g. 6 as error bars for all measurements.
3.4 Discussion
The most interesting feature of the radiograph shown in
g. 5 is that an edge enhancement or phase contrast can be
observed with a polychromatic X-ray beam. This fact has
been discussed in a number of papers also in connection
with the interplay between refraction and diraction [5,
19]. The general features of refraction contrast imaging
will be discussed by means of g. 6. It can be stated that
the distance xod between object and detector must be at
least as large that the wave optical spread of the diracted
X-rays becomes comparable with the detector resolution.
201
Fig. 6. Contrast Cref for a polyamide string of 270 m diam-
eter as a function of the object-to-detector distance xod . The
source-to-detector distance xso = 11.38 m was kept constant.
Error bars are measurements, crossed circles calculations on
the basis of the wave optical model with a beam spot size
v = 7.5 m and a total X-ray lm resolution and scanner
resolution t = (10.0 0.4) m. Stars designate calculations
according to geometrical optics.
plane increases and this worsens the contrast at larger xod
distances. The maximum of the contrast is a function of
the beam spot size and of the lm resolution. But contrast
is not the only gure of merit. It must also be taken into
and the resolution deteriorates. The latter might be unde-
sirable in case that resolution is of importance and nearby
features must be resolved.
Next, the question will be addressed whether the mea-
sured edge enhancement structures can be understood
quantitatively in the framework of the wave optical and
geometrical models. As has already been pointed out
above, the refraction contrast is in a strict sense a wave op-
tical phenomenon, however, with some care it can also be
explained in the framework of geometrical optics. Such an
approach might be a good approximation at experimen-
tal conditions in which interference patterns are smeared
out, i.e., if the object is illuminated with polychromatic
X-rays, or if the projected source size or the detector res-
olution are too large, respectively too bad. Both models
have two free parameters which are the standard devia-
tion of the beam spot size and the resolution of the X-ray
lm including the lm scanner.
The wave optical calculations were performed with the
eective X-ray spectrum which takes into account the
transition radiation spectrum and the absorption char-
acteristics of air and the X-ray lm. The eective X-
ray spectrum was approximated with 22 discrete values
in the energy range between 8 and 30 keV. For the ge-
ometrical model it was su cient to approximate the X-
ray spectrum by a delta-function at the mean photon
Otherwise all interference fringes are blurred and the con- energy h = 19.6 keV, corresponding to a wavelength
trast is low. With increasing object-to-detector distance = 0.633 A, since model calculations showed, via the op-
xod the contrast increases about linearly. However, at the tical parameters, a rather weak energy dependence of the
same time the projected X-ray spot size on the detector sharp edge structure. As can be seen, both the geometrical
202 The European Physical Journal A
Fig. 7. A refraction contrast radiograph of a part of green
leaf of Rumex crispus. The radiograph was recorded by the
MAMORAY MR5 II PQ (Agfa) X-ray lm. The object-to-
detector distance was xod = 5.5 m at a source-to-object dis-
tance xso = 5.88 m. With these parameters the magnica-
tion was 1.94 times. The electron beam energy was 855 MeV,
the electron beam spot size had standard deviations of h =
(8.6 0.1) m and v = (7.5 0.1) m in the horizontal and
vertical direction, respectively. The TR foil stack described in
g. 4 was used. The electron beam current was 6 nA, the ex-
posure time 40 s.
and the wave optical model describe the general features
of the measurement quite well. The rather good results
for the contrast ratio Cref of the geometrical model, as
shown for the example in g. 5 (c), were obtained after a
convolution with a Gaussian of standard deviation
 2
xod xod xsd
= s2 + d2 + p2 + . (11)
xso 2xso
This parameter takes into account projected source
size s , X-ray lm resolution d , pixel resolution of
 lm scanner p , and an additional term w =
the
xso xod /(2xsd ). This term accounts for the dirac-
tion which is absent in the geometrical model. It was
estimated from the second exponential in eq. (6). The
argument is that for a xed point zo at the object the
typical spread in the detector plane is given by a region
|zo xsd /xso zd | < w . For |zo xsd /xso zd |  w the
exponential oscillates rapidly and the mean value in zd
approaches zero. The real parameter w may dier from
the assumed one. However, the good agreement may be
a consequence of the rather poor total lm resolution
$ are observed in diraction experiments in which the refer-
2 2
t = d + p = 10 m which is much larger as the wave
optical contribution w = 2.3 m.
As an example of the visualization of low-Z objects by
the refraction contrast, in g. 7 the image of a green leaf
is shown. In the part labelled with (a) where the leaf is
thinner than 1 mm, the visibility of a bundle of vascular
tissue (veins) could be resolved with high contrast. In the
middle part labelled by (b), the object is about 3 mm thick
and contains a bundle of vascular tissue (veins). However,
the identication of an individual vein is di cult since im-
ages from the dierent veins in the radiograph are over-
lapping. Such three-dimensional structures may be disen-
tangled by a holographic method some principles of which
are sketched in the next section.
4 Toward hard X-ray in-line holography
In the preceding section it has been shown that the transi-
tion radiation (TR) X-ray source is well suited for refrac-
tion contrast imaging. This chapter deals with the inves-
tigation of the possibility of X-ray phase contrast imag-
ing and hard X-ray in-line holography with monochro-
matic X-rays at MAMI. The good emittance of MAMI
allows the preparation of a micro-focus which is a pre-
requisite of the required transverse coherence of the TR
X-ray source. The longitudinal coherence can be achieved
by a single-crystal monochromator. The basics of in-line
holography, the experimental setup, the preparation of the
micro-focused electron beam and the results obtained so
far will be described in the following.
4.1 Basics
A wave emanating from a point source may illuminate
an object from which it is scattered. The wave amplitude
E(r) = E0 (r) + Escat (r) can be split into the reference
wave E0 (r) and a scattered wave Escat (r) = a(r) E0 (r).
The amplitude ratio can be written as E(r)/E0 (r) =
1 + a(r). The scattering amplitude a(r) contains the re-
quired information on the object. On a detector screen,
such as an X-ray lm or a CCD detector, the squared ab-
solute values of the amplitudes |E(r)|2 and |E0 (r)|2 are
measured from which the contrast image |E(r)|2 |E0 (r)|2
can be obtained. By division through the reference wave
|E0 (r)|2 the normalized contrast ratio,
|E(r)|2 |E0 (r)|2
Inorm (r) = = 2 [a(r)] + |a(r)|2 ,
|E0 (r)|2
(12)
can be determined.
The appearance of 2 [a(r)] = a(r) + a (r) on the
right-hand side of eq. (12) shows that the hologram con-
tains also information on the real part of the scattering
amplitude rather than only its absolute value squared
|a(r)|2 which may be referred to as classical dirac-
tion pattern of the complementary transmission func-
tion of the object [15]. Such classical diraction patterns
ence wave is absent, e.g., at diraction on a slit which
is the complementary to an opaque object as, e.g., an
opaque wire. While the classical diraction pattern is
rather smooth, see g. 8 (a), the holographic diraction
pattern oscillates rapidly, see g. 8 (b). These oscillations
have a rather small amplitude and can hardly be seen in a
measurement of the hologram such is shown in g. 8 (c).
Much more pronounced oscillations are observed for trans-
parent objects as polymer strings, see g. 8 (d) and (e)
which are maintained in the sum of the classical and the
 M. El-Ghazaly et al.: X-ray phase contrast imaging at MAMI 203

Photons [1/(e 10 BW sr)]

20

-3
-
10 Si (111) crystal
Polyimide NH9
monochromator
foil stack

0
H 10 20 30 40 50
[keV] 10 4


H to
beam dump
600 MeV e- beam
Quadrupole
Transition radiation 38.5 6 keV
doublet
Micro focus

Cooled slow scan

Objects CCD detector
P

Fig. 9. Schematic experimental setup for X-ray in-line holog-

Fig. 8. Analysis of the calculated normalized contrast image
in distinct patterns for a totally opaque tungsten wire, left
column (a), (b) and (c), and for an approximately transparent
polymer string, right column (d), (e) and (f). Both wires have
the same diameter of 25 m. The X-ray photon energy is 6 keV
( = 2.067 A), the complex refraction index parameters are
W = 8.5 10 5 and W = 1.1 10 5 and P = 7.3 10 6 and
P = 2.55 10 8 for tungsten and polymer, respectively, at
this energy. The source-to-object distance is xso = 1.92 m and
the object-to-detector distance xod = 11.68 m. Panels (a) and
(d) show the classical diraction pattern |a(zd )|2 which is the
diraction pattern of the complementary object, (b) and (e)
raphy at MAMI. Shown are the TR foil stack, the single-crystal
monochromator at a distance of 7.8 m from the target, and a
CCD detector or an X-ray lm at a distance of 5.8 m from the
monochromator. The objects to be imaged can be positioned at
distances of 1.88, 4.3, 7.47, 10.78, 12.71 and 13.6 m from the X-
ray source. All components are housed in a connected vacuum
system to avoid self-absorption of the X-rays. The inset shows
the calculated TR energy spectrum as function of the photon
energy for which multiple scattering, electron beam divergence
(0.8 mrad) and self-absorption were taken into account. The
TR foil stack consists of 25 polyimide foils with a thickness
of 12.5 m which are spaced out by aluminium foils of 100 m
thickness, the latter with centric holes of 2 mm diameter for
the passage of the electron beam.
to Nuclear Physics Beam Dump
-
Experiments e - Beam
P
X1 Hall
e-- Beam
Crystal
Monochromator
Polychromatic
show the holographic diraction pattern 2[a(zd )] which come
about by the interference between the wave front disturbed by
the object and the reference wave emanating from the source,
(c) and (f) show the normalized contrast images.
holographic diraction pattern, see g. 8 (f). These oscil-
lations contain information on the distance between the
object and the detector or the source, and via the refrac-
tive index decrement and the absorption also on the
bulk of the string. In addition, the hologram contains via
the transverse coherence length also information on the
beam spot size (ref. [20]).
4.2 Experimental
The observation of interference patterns as shown, for in-
stance, in g. 8 requires both, a good transverse and a
good longitudinal coherence which can be achieved with a
microfocused and monochromatic X-ray beam. These re-
quirements led to an experimental arrangement at MAMI
Focusing Transition Radiation
Quadrupoles
RTM3
Radiator Chamber
Monochromatic
CCD detector
Transition Radiation
Fig. 10. Floor plan of the experimental area at MAMI. The
electron beam is fed into the X1-beam line just behind the third
stage of the race track microtron RTM3. During the course of
the experiments the entrance to the X1 hall below is closed by
a concrete door.
which is schematically depicted in g. 9. The oor plan at
MAMI is shown in g. 10. A at single crystal in Bragg
geometry is used as monochromator. The objects to be
imaged can be placed between the TR radiator and the
monochromator crystal close to the TR source resulting in
a magnication of the object of up to a factor of 7.4 or, al-
ternatively, between monochromator and X-ray detector.
The magnication may be of importance to compensate
204 The European Physical Journal A
for a moderate detector resolution if, e.g., CCD-chips in a
direct exposure mode are used, see below.
For the preparation of a micro-focused electron beam,
a low beam emittance in horizontal and vertical directions
is of particular importance. The emittance of the MAMI
electron beam in horizonal direction is bigger than the
emittance in vertical direction because the electrons emit
synchrotron radiation in the bending magnets of race-
track microtron 3. The horizontal emittance grows rapidly
above an electron beam energy of 400 MeV, while the ver-
tical emittance still decreases. As a compromise, a beam
energy of 600 MeV was chosen for which the emittances
are h = 2.3 m mrad and v = 0.52 m mrad in the hori-
zontal and vertical directions, respectively.
The polyimide foil stack to produce transition radia-
tion is optimized for a high X-ray ux at a photon energy
of 6 keV at the electron beam energy of 600 MeV. The cal-
culated photon energy spectrum is shown in the inset of
g. 9. The at silicon single crystal with its surface par-
allel to the (111) crystal plane acts as a mirror for the
TR photons. However, the mirror is energy dispersive in
the horizontal direction. The deviation of the photon
energy, dened by the equation h = hB (1 + ), from
the nominal Bragg energy
2 h2 + k 2 + l2 hc
hB = (13)
a0 2 sin B
is approximately given by the expression [21]
(0 ) x The standard deviation of the resolution was measured to
= , (14)
2 sin2 B tan B
where x = B is the deviation from the nominal
Bragg angle B . The integers h, k, l are the Miller indices,
a0 = 5.4309 A the lattice constant, and (0 ) the real
part of the dielectric susceptibility 0 . The Bragg angle
for hB = 6 keV amounts for the (111) reection to B =
19.25 .
In in-line holography a scattered wave from the object
interferes with an unscattered wave from the source. For
an assumed transverse coherence length of LT = 250 m,
at a distance of 13.6 m a x = 18.4 rad results. The
angular spread which originates from the beam spot size
and the pixel resolution is less than 1 rad and can be ne-
glected. The angle x corresponds, according to eq. (14),
to a relative energy shift of 5.3 105 or h = 0.32 eV.
This means that two waves with slightly dierent ener-
gies must interfere which is only possible if the longitu-
dinal coherence length LL is long enough. The width of
the reecting power ratio of the monochromator crystal
is = 1.4 104 corresponding to h = 0.84 eV [21], changing and the sector with the best contrast was se-
and a longitudinal coherence length LL = 0.52 / =
0.5/ = 0.74 m results. This value is su ciently large
for all objects investigated in this work which had thick-
nesses in the sub-mm range, since at a refractive index
decrement of = 1 106 the optical path dierence is
less than 0.01 m.
Hard X-ray holography requires, like refraction con-
trast radiography, a two-dimensional resolving detector
with a large dynamic range and linear relationship be-
tween the incident radiation intensity and the response of
the detector. Such conditions can be fullled by a charge-
coupled device (CCD) or an X-ray lm. For the current ex-
periments the CCD system ANDOR DO-434 BN CCD [22]
was used. It contains a back-illuminated CCD low-noise
sensor from Marconi CCD47-10 [23] with 1024 1024 pix-
els of size 13 13 m2 . The chip has a good quantum
e ciency over a wide spectral range. For X-rays of 6 keV
energy it amounts to still about 45%. These features oer
the opportunity to use the CCD chip in the direct ex-
posure mode in which the signal is generated by direct
energy deposition of X-rays in the sensitive layer of ap-
proximately 10 m thickness.
Direct-exposure CCD camera chips have, compared
with X-ray lms, the big advantage that they have a good
linearity over a wide dynamical range, a good signal-to-
noise ratio, and that they are on-line capable. The latter
fact is very important since contrast or normalized con-
trast images can easily be generated in which all parasitic
background, originating not from the object, can be elim-
inated. The disadvantage of a moderate spatial resolution
in comparison to an X-ray lm can be alleviated by a ge-
ometrical magnication. In reality, however, the spatial
resolution is larger than the pixel size because of so-called
split events in which the deposited energy is shared by
neighboring pixels.
The Structurix D3 X-ray lm from Agfa is a useful
detector as well. The main advantage in comparison with
the direct exposure CCD chip is its very good resolution.
be f = (1.2 0.4) m, which is about a factor of 6 better
than for the direct-exposure CCD chip. The main disad-
vantage of the X-ray lm is the missing on-line capability
with the consequence that the generation of normalized
contrast images is rather involved.
The procedure to obtain the intensity information from
the photographic density is similar to that already de-
scribed in sect. 3.2. The X-ray lm was digitized with a
lm scanner (Nikon Coolscan LS 4000 [17]) and with an
optical microscope equipped with a high-resolution 8-bit
CCD camera (F-View XS [24]). From this system limita-
tions are expected because the dynamical range cannot
be better than the digitization depth of the ADC (1:256),
while the X-ray lm has a dynamical range which is more
than a factor of 10 better (3.5 decades corresponding to
1:3160). Since, in addition, the illumination time was se-
lected automatically by the scanner after the part of in-
terest of the picture and the optical magnication were
selected, the holograms were digitized at various positions
of the string. Along the imaged strings the exposure is
lected for further analysis.
4.3 Measurements and discussion
4.3.1 Investigation of the transverse coherence in horizontal
direction
To study the coherence in the horizontal and vertical di-
rections, radiographs of two polymer strings of the same
 M. El-Ghazaly et al.: X-ray phase contrast imaging at MAMI
Fig. 11. Two radiographs of a polymer string of diameter
30 m. In radiograph (a) the string is mounted vertically, in
(b) horizontally. The X-ray source spot size was h = (1.7
0.1) m in the horizontal and v = (3.90.4) m in the vertical
direction. The electron beam spot size was checked with the
wire scanner before and after the imaging in order exclude a
possible shift of the beam spot. The source-to-object distance
was xso = 4.3 m, and the object-to-detector distance xod =
9.61 m. Electron beam current 700 nA, exposure time 1.8 s, 50
frames added up.
Fig. 12. A background-corrected hologram (contrast image)
of a polymer string with (150 20) m diameter, supplied by
Goodfellow. Source-to-object distance xso = 1.88 m, source-to-
detector distance xsd = 13.91 m, corresponding to a magni-
cation of 7.4 times, tilt angle of the object 46 , X-ray source
spot size h = (5.9 0.1) m, and v = (2.6 0.1) m, electron observed. However, in the hologram inhomogeneities are
beam current 600 nA, exposure time 8.1 s, 50 frames added up.
thickness of 30 m were taken which were mounted hor-
izontally and vertically. The radiographs are shown in
g. 11. Although the beam size in the horizontal direction
h = (1.7 0.1) m was smaller than v = (3.9 0.4) m
in the vertical direction, no interference patterns were
observed for the vertically mounted polymer string. The
maximum contrast, Cref = (Imax Imin )/(Imax + Imin )
for the horizontally mounted string was 64%, while for
the vertically mounted one it was only 11%. The only rea-
sonable explanation for this observation is that the trans-
verse coherence in the horizontal direction is deteriorated
by the monochromator crystal. Obviously, in the energy
dispersive direction (horizontally) an additional angular
divergence is introduced by the crystal. Since the reason of
this eect could not be found, all experiments with strings
described in this work were performed with horizontally
mounted strings.
Despite the moderate transverse coherence in horizon-
tal direction holograms could be taken which show also
interesting features in the horizontal direction. Figure 12
shows a contrast image of a polymer string with a diameter
205
Fig. 13. Fringe visibility as a function of the X-ray source spot
size. Shown are holograms of a polymer string with a diameter
of 30 m for an X-ray source spot size as measured with a
wire scanner of (a) h = (5.9 0.1) m, v = (2.6 0.1) m,
and (b) h = (19.1 0.7) m, v = (0.50 0.05) m. From
the smallest discernible fringe spacings rmax = 50 m (a) and
25 m (b), with eq. (15) standard deviations v = 2.4 m (a),
and 1.2 m (b) result. Source-to-object distance xso = 1.88 m,
object-to-detector distance xod = 12.03 m, corresponding to
a geometrical magnication of 7.4 times. The angle between
string and beam direction amounted to 46 . The electron beam
current was 500 nA, exposure time 8.1 s per frame, 100 frames
added up.
of (15020) m. According to inspection under an optical
microscope the string has a nearly ideal cylindrical shape.
No deformations or impurities inside the string could be
clearly visible which may be air bubbles or impurity inclu-
sions with a dierent density than the string material. The
background correction assures that these inhomogeneities
do not originate from dust particles on the monochroma-
tor crystal or the detector.
4.3.2 Optimization of the beam spot size
The most important prerequisite for taking high-quality
holograms is the minimization of the beam spot size. In
the rst step, the spot size was measured with a tungsten
wire of (4.00.4) m diameter which was scanned through
the electron beam. In particular, the electrical current of
the quadrupole doublet in g. 9 was varied until the scan
yielded the smallest spot size. In the next step holograms
of polymer strings were taken with the CCD camera and
the spot size was estimated from the smallest discernible
fringe visibility estimated according to [25]
xso
= 0.31 rmax . (15)
xod
As already mentioned, a CCD chip allows fast on-line
imaging, however, the resolution in the direct-exposure
206 The European Physical Journal A

Fig. 14. (a) Hologram of a polyamide string with diameter of

(150 20) m. Source-to-object distance xso = 1.88 m, object-
to-detector distance xod = 11.73 m, corresponding magnica-
tion 7.24 times, X-ray source spot size h = (19.1 0.7) m
and v = (0.50 0.05) m as measured with the tungsten wire.
The X-ray lm was digitized with an optical microscope with
a magnication of 4 in order to maintain a good resolution.
Therefore, only part of the hologram was in the eld of view.
(b) Intensity prole with 200 rows added up. Notice that also
this string was stretched horizontally.

mode is limited by the pixel size of 13 m. To achieve a

geometrical magnication, the object was placed at close
distance to the X-ray source. In g. 9 the possible po-
sitions of the objects are marked. An example of the
spot size measurement by the fringe method is shown in
g. 13 which yielded after optimization a standard devia-
tion v = 1.2 m.
To exclude a possible inuence of the moderate reso-
lution of the CCD detector, high-quality holograms were
taken with the high-resolution Structurix D3 X-ray lm
from Agfa. Figure 14 (a) shows a part of a hologram
for a polyamide (Nylon) string. A large number of about
18 interference fringes can be seen, as demonstrated in
g. 14 (b). In this radiograph the main deterioration in the
fringe visibility results from the X-ray spot size. The min-
imum discernible distance between two adjacent fringes is
about 25 m and the estimated X-ray source size is again
about v = 1.2 m.
In comparison with the wire scanner measurement
which yielded a spot size v = (0.50 0.05) m, the
measured values with the fringe method either deter-
mined with the direct exposure CCD, or with the high-
resolution X-ray lm, deviate signicantly. This deviation
can be explained by the longitudinal depth of the foil
stack which amounts to 2.8 mm. For the measured ver-
tical emittance v = 0.52 m mrad at the electron beam
energy of 600 MeV, and a micro-focused electron beam
spot size v = (0.50 0.05) m, the corresponding di-
vergence is 1.04 mrad. When the focus is exactly in the
middle of the foil stack, the beam spread within the foil
stack amounts to a standard deviation of 1.5 m, in ac-
cord with the observation with the direct-exposure CCD
chip and the X-ray lm 4 .
4.3.3 Analysis of holograms for polyamide strings
There are two possibilities to analyze holograms of strings.
In the rst one, calculations on the basis of the Fresnel-
4
The errors of the fringe method may be in the order of
20%.
Fig. 15. (a) Radiograph of a polyamide string of (15020) m
diameter. Source-to-object distance xso = 4.3 m, object-to-
detector distance xod = 9.31 m, corresponding geometrical
magnication 3.17 times, X-ray source size h = (19.10.7) m
and v = (0.50 0.05) m. The X-ray lm was digitized with
an optical microscope with a magnication of 4. (b) Intensity
prole, 200 rows added up. (c) Calculated intensity prole of
the radiograph. The spatial resolution of the X-ray source spot
size, v = (0.500.05) m, of the lm, f = (1.20.4) m, and
the optical resolution, sc,4 = (4.10.1) m, were incorporated
in the calculations.
Kirchho integrals can be performed in which assumptions
about the density and morphology of the string are incor-
porated. The right solution can be found by a systemati-
cal trial and error method. An example is shown in g. 15.
Although a good overall agreement between measurement
and calculation could be achieved with the assumption
of a homogeneous density distribution within the string,
in detail signicant dierences in the interference pattern
close to the boundaries can be recognized. These dier-
ences may indicate a density gradient at the periphery of
the string. Renements of the string model are required
to further investigate the origin of these dierences.
The second possibility is based on reconstruction algo-
rithms to nd the phase prole produced by the transpar-
ent object. One of these is the modied Gerchberg-Saxton
algorithm [26]. It is an iterative method with which the
phase information can be found from two holograms taken
 M. El-Ghazaly et al.: X-ray phase contrast imaging at MAMI 207

ing cases may be of particular interest for the reconstruc-

tion of the phase prole. However, this issue went beyond
the scope of this explorative experimental work and is sub-
ject of ongoing investigations.

4.4 Conclusions

Phase contrast radiography has been accomplished with

an external 855 MeV electron beam using broad-band
transition radiation X-rays with a mean photon energy
Fig. 16. Holograms of a polyamide string of 30 m diameter
taken at dierent positions: (a) xod = 0.9 m, (c) xod = 2.83 m,
(d) xod = 9.31 m and (g) xod = 11.73 m, the latter at an angle
of 45 . The corresponding intensity proles are shown on the
right-side in panels (b), (d), (f), and (h), 200 rows were added
up. The X-ray lm was digitized with an optical microscope,
(a) and (b) with a magnication by a factor of 10, (c) and (d)
with a magnication by a factor of 4.
at dierent distances between object and detector. There-
fore, holograms of a polyamide string with a diameter of
30 m were taken at dierent object-to-detector distances.
The holograms are shown in g. 16.
The selected distances cover dierent imaging regimes.
For the contact regime the contrast would be (1
exp[(4D)/]) = 4.26% and the absorption contrast
of the polyamide string with a diameter of D = 30 m
can be neglected. In the near eld region, at an object-to-
detector distance xod = 0.9 m, g. 16 (a) and (b), the in-
terference pattern produced by both edges have only very
little overlap.
The reason is that the size of the rst Fres-
nel zone xod = 13.6 m is smaller than the diameter of
the string. However, at the largest distance xod = 11.73 m,
g.
16 (g) and (h), one obtains for the rst Fresnel zone
xod = 49.2 m and the interference pattern from both
edges do overlap. The resemblance between the original
object and the radiograph is more or less lost. Both limit-
h 20 keV and a micro-focus with standard deviations
of typically h = 8.6 m and v = 7.5 m in the horizontal
and vertical direction, respectively. In-line holograms of
polymer strings were taken with a low-emittance 600 MeV
electron beam using narrow-band transition radiation X-
rays with a photon energy of h = 6 keV and a micro-focus
with a standard deviation of typically v = 1.2 m. High-
quality holograms were obtained with high-resolution X-
ray lms and a direct-exposure cooled CCD camera chip.
The advantage of the former is the very good spatial res-
olution, that of the latter its on-line capability.
An X-ray beam spot with micro-dimensions can be
prepared directly with the micro-focused external elec-
tron beam via transition radiation production in a foil
stack. Objects to be investigated can be placed in close
distance to the small X-ray beam spot. This has the ad-
vantage that a large geometrical magnication of up to a
factor of 10 can easily be achieved in our relatively small
experimental area. The disadvantage of the transition ra-
diation X-ray source is its contamination with high-energy
bremsstrahlung photons.
Typical electron beam charges required to capture a
single image are about 0.3 C for phase contrast radio-
graphs with broad-band polychromatic X-rays and an X-
ray lm as detector, and some nC for a cooled CCD chip.
In-line holograms with narrow-band X-rays require about
500 C for a high-resolution X-ray lm, and 5 10 C for
the CCD detector.
We thank F. Hagenbuck, and H.-K. Kaiser for signicant con-
tributions in the early stage of the experiment, and Mrs. C.
Koch-Brandt, Institut fur Biochemie, Universitat Mainz, for
making her lm scanner device available to us. This work has
been supported by Deutsche Forschungsgemeinschaft DFG un-
der contract BA 1336/1-4.
References
1. F. Arfelli, M. Assante, V. Bonvicini, A. Bravin, G. Can-
tatore, E. Castelli, L. Dalla Palmaz, M. Di Michiel, R.
Longox, A. Olivox, S. Panix, D. Pontoni, P. Poropat, M
Prestx, A Rashevskyx, G. Trombay, A. Vacchix, E. Val-
lazza, F. Zanconati, Phys. Med. Biol. 43, 2845 (1998).
2. C.J. Kotre, I.P. Birch, Phys. Med. Biol. 44, 2853 (1999).
3. L.D. Turner, B.B. Dhal, J.P. Hayes, A.P. Mancuso, K.A.
Nugent, D. Paterson, R.E. Scholten, C.Q. Tran, A.G.
Peele, Opt. Expr. 12, 2960 (2004).
208 The European Physical Journal A
4. F. Pfeier, T. Weitkamp, O. Bunk, Ch. David, Nature
Physics advance online publication www.nature.com/
naturephysics, published online: 26 March 2006;
doi:10.1038/nphys265.
5. S.W. Wilkins, T.E. Gureyev, D. Gao, A. Pogany, A.W.
Stevenson, Nature (London) 384, 335 (1996).
6. Xizeng Wu, Hong Liu, Med. Phys. 30, 2169 (2003).
7. T. Takeda, A. Momose, E. Ueno, Y. Itai, J. Synchrotron
Rad. 5, 1133 (1998).
8. R.A. Lewis, Phys. Med. Biol. 49, 3573 (2004).
9. D. Gabor, Nature 161, 777 (1948).
10. P. Spanne, C. Raven, I. Snigireva, A. Snigirev, Phys. Med.
Biol. 44, 741 (1999).
11. P. Cloetens, R. Barrett, J. Baruchel, J. Guigay, M.
Schlenker, J. Phys. D 29, 133 (1996).
12. Z.W. Hu, B. Lai, Y.S. Chu, Z. Cai, D.C. Mancini, B.R.
Thomas, A.A. Chernov, Phys. Rev. Lett. 87, 148101
(2001).
13. R.W. James, The optical Principles of the Diraction of
X-rays (Cornell University Press, 1965).
14. V. Kohn, I. Snigireva, A. Snigirev, Opt. Commun. 198,
293 (2001).
15. Mahmoud El Ghazaly, X-ray Phase Contrast Imaging at
the Mainz Microtron MAMI, Dissertation, Institut fur
Kernphysik, Universitat Mainz, 2005.
16. B.L. Henke, J.Y. Uejio, G.F. Stone, C.H. Dittmore, F.G.
Fujiwara, J. Opt. Soc. Am. B. 11, 1540 (1986).
17. http://www.filmscanner.info/NikonSuperCoolscan-
4000ED.html.
18. Georg Joos, Erwin Schopper, Grundriss der Photogra-
phie und ihrer Anwendungen besonders in der Atomphysik
(Akademische Verlagsgesellschaft M. B. H., Frankfurt am
Main, 1958).
19. Y. Hwu, H.H. Hsieh, M.J. Lu, W.L. Tsai, H.M. Lin, W.C.
Goh, B. Lai, J.H. Je, C.K. Kim, D.Y. Noh, H.S. Youn, G.
Tromba, G. Margaritondo, J. Appl. Phys. 86, 4613 (1999).
20. O. Chubar, A. Snigirev, S. Kuznetsov, T. Weitkamp, V.
Kohn, Proceedings DIPAC 2001, ESRF, Grenoble, France.
21. A. Caticha, Phys. Rev. A 40, 4322 (1989).
22. http://www.andor-tech.com/germany/products/oem.
cfm
23. http://www.data.it/support/data sheets/e2vtech/
47-10back.pdf
24. http://www.olympus.pl/pliki/mikroskopy/dokumenty/
LM cameras ENG.pdf.
25. C. Raven, Microimaging and Tomography with High En-
ergy Coherent Synchrotron X-Rays (Shaker Verlag, 1998).
26. R.W. Gerchberg, W.O. Saxton, Optik 35, 237 (1972).
 Author index

Backe H. El-Ghazaly M. Maas F.E.: Parity-violating electron scattering at the

Backe H. Lauth W.
Beck R.: Experiments with photons at MAMI 173
Boeglin W.U.: Few-nucleon systems at MAMI and beyond
19
Cardman L.S.: Physics at the Thomas Jeerson National
Accelerator Facility 7
MAMI facility in Mainz 107
Mecking B.A.: Twenty years of physics at MAMI What
did it mean? 209
Merkel H.: Experimental tests of Chiral Perturbation The-
ory 129
Milner R.G.: The beauty of the electromagnetic probe 1

dHose N.: Virtual Compton Scattering at MAMI 117
El-Ghazaly M., Backe H., Lauth W., Kube G., Kunz P.,
Sharafutdinov A. and Weber T.: X-ray phase contrast
imaging at MAMI 197
Ostrick M.: Electromagnetic form factors of the nucleon
81
Rohe D. (A1 and A3 Collaboration): Experiments with
polarized 3 He at MAMI 29

Hammer H.-W.: Nucleon form factors in dispersion theory Scherer S.: Chiral perturbation theory 59
49 Schmieden H.: Photo- and electro-excitation of the
Hillert W.: The Bonn Electron Stretcher Accelerator -resonance at MAMI 91
ELSA: Past and future 139 Schwamb M.: Few-nucleon systems (theory) 39
Sharafutdinov A. El-Ghazaly M.
Jankowiak A.: The Mainz Microtron MAMI Past and Sharafutdinov A. Lauth W.
future 149
Thomas A.: The Gerasimov-Drell-Hearn sum rule at
Kettig O. Lauth W. MAMI 161
Kowalski S.: Parity violation in electron scattering 101
Kube G. El-Ghazaly M. Vanderhaeghen M.: Two-photon physics 71
Kunz P. El-Ghazaly M.
Kunz P. Lauth W. Weber T. El-Ghazaly M.
Weber T. Lauth W.
Lauth W., Backe H., Kettig O., Kunz P., Sharafutdinov
A. and Weber T.: Coherent X-rays at MAMI 185
Lauth W. El-Ghazaly M.
Eur. Phys. J. A 28, s01, 209 219 (2006)
DOI: 10.1140/epja/i2006-09-020-7 EPJ A direct
electronic only

Twenty years of physics at MAMI What did it mean?

B.A. Meckinga
Thomas Jeerson National Accelerator Facility, 12000 Jeerson Avenue, Newport News, VA 23606, USA

/
Published online: 7 June 2006 
c Societa Italiana di Fisica / Springer-Verlag 2006

Abstract. The development over the last twenty years of the physics program and the experimental facilities
at the Mainz Microtron MAMI will be reviewed. Ground-breaking contributions have been made to the
development of experimental techniques and to our understanding of the structure of nucleons and nuclei.

PACS. 29.17.+w Electrostatic, collective, and linear accelerators 25.20.-x Photonuclear reactions
25.30.Bf Elastic electron scattering 25.30.Dh Inelastic electron scattering to specic states

1 Introduction
The goal of nuclear physics is to study the properties of
nuclei, and to understand these properties on the basis of
the fundamental theory of the constituents making up the
nucleus. Quantum-chromodynamics (QCD) has emerged
as the leading candidate for the theory of hadronic inter-
actions. Presently, QCD cannot be solved in the strong-
coupling regime due to the lack of appropriate pertur-
bative solution. The best hope in the near future is to
use large-scale numerical calculations to approximate the
space-time continuum by a discrete lattice in the frame-
work of Lattice QCD (LQCD). Due to limitations in com-
pute power, the lattice spacing is still fairly coarse, and
the masses of the quarks used are much larger than their
actual values. These limitations require signicant extrap-
olations which need to be constrained theoretically.
For zero-mass quarks, QCD can be solved in a rigor-
ous way via Chiral Perturbation Theory (PT). Again, an
extrapolation is required, this time from zero-mass quarks
up to the actual values. An interesting recent development
is the use of functional forms derived from PT to extrap-
olate LQCD results.
Before these extrapolations can be trusted, it is very
important to verify the predictive power of PT via ex-
perimental tests, in particular via pion photoproduction
close to threshold.
On the experimental side, valuable information on the
properties of bound quark systems is still lacking. In par-
ticular, the knowledge of the spatial distribution of the
charges and the currents inside the nucleon is not satis-
factory, especially for the neutron. Our knowledge of the
excited states of the nucleon is still insu cient, e.g. what
are the degrees-of-freedom governing the mass spectrum,
a
e-mail: mecking@jlab.org
and what are the dierences in quark wave functions be-
tween the ground state and the excited states.
Many of the issues mentioned above can be investi-
gated using electron scattering. Electrons interact only
with the charged constituents of the object under investi-
gation. The interaction is described by Quantum-Electro-
dynamics (QED), and is su ciently small to be handled
with perturbative methods.
In the past, the usefulness of the electromagnetic probe
was limited by the technical features of the available elec-
tron accelerators. In particular, coincidence experiments
and the operation of large acceptance detectors were ham-
pered by the low duty-cycle of the electron beams. Dra-
matic progress in accelerator and detector technology has
made it possible to overcome these limitations and to
study electromagnetic processes with an accuracy that is
no longer limited by technical problems.
The Institut fur Kernphysik (Institute for Nuclear
Physics) at the University of Mainz in Germany has been
at the forefront of this development. For more than 20
years, the Institute has developed novel electron accel-
erators (the MAMI series of microtrons) and the corre-
sponding experimental equipment, and has used those de-
vices for ground-breaking research into the electromag-
netic structure of nucleons and nuclei.
On the occasion of the retirement of six key people
(Hartmuth Arenhovel, Hartmut Backe, Dieter Drechsel,
Jorg Friedrich, Karl-Heinz Kaiser, and Thomas Walcher)
from the Institute, a symposium was held in October 2005
to review 20 years of Physics at MAMI and to commem-
orate their contributions.
This paper will attempt to review the major contri-
butions MAMI and its user community have made to the
eld, to identify the particular circumstances that made
these contributions possible, and to speculate on their last-
ing impact.
210 The European Physical Journal A

Fig. 1. Layout of the MAMI accelerators and the experimental areas. MAMI A is located in the RTM2 area, MAMI B in
RTM3, the new MAMI C in the area labeled HDSM.

Table 1. MAMI microtron development. eective way to build an accelerator capable of delivering
a high quality electron beam [1]. The layout of the accel-
Year Activity erator and its experimental areas is shown in g. 1. The
1975 Proposal for a Race-Track Microtron microtron design relies on sending the beam repeatedly
(design by H. Herminghaus et al.) through the same room-temperature accelerating struc-
1979 14 MeV beam from MAMI A1 ture with moderate energy gain per turn. Recirculation is
1982 Preliminary Sonderforschungsbereich achieved by two homogeneous 180 end-magnets. The size
(SFB) established of these end-magnets for the last microtron stage, MAMI
1983 183 MeV beam from MAMI A2 B, is evident from g. 2. The perpendicular entry and exit
1983 - 87 MAMI A operation with a total of 18,700 h of the electron orbits at the end-magnets results in simple
1983 - 90 Development of the 855 MeV MAMI B and robust beam optics. Due to the continuous-wave na-
1984 SFB 201 established ture of the radio-frequency power and the constant mag-
1990 First 855 MeV beam from MAMI B netic eld, the quality of the beam is very high: an energy
(rst experiment by A2 Collaboration) spread of E/E = 1.5 105 and an emittance of  = 8
1990 - 2005 MAMI B operation with a total of 82,843 h 109 m is achieved routinely. A laser-driven polarized gun
1999 Sonderforschungsbereich 443 established produces electron beams with 80% polarization. Parity vi-
2000 Approval of 1.5 GeV HDSM olation experiments are possible since helicity-correlated
(Harmonic Double-Sided Microtron,
changes in the beam parameters are very small: energy
design by K.-H. Kaiser et al.)
variations of E/E 108 and position variations of x
2001 - 03 Installation of the four HDSM magnets
2006 Commissioning and begin of physics
100 nm have been achieved. Particularly impressive is the
high operational stability: overnight and during weekends,
the MAMI microtrons are routinely operated by students.
2 MAMI microtron development It has been the conventional wisdom in the accelerator
community that the maximum energy of a microtron is
The history of the MAMI microtron development is sum- limited to about one GeV since the construction of the
marized in table 1. The pioneering development of MAMI end-magnets which increase rapidly in size with increasing
A and B at Mainz has established the microtron as a cost- energy becomes technically and nancially impractical.
 B.A. Mecking: Twenty years of physics at MAMI What did it mean? 211

Table 2. Three-spectrometer system parameters.

Spectrometer A B C
conguration QSDD D QSDD
pmax [MeV/c] 665 810 490
[msr] 28 5.6 28
min 18 7 18
p/p [%] 20 15 25

Fig. 2. The MAMI accelerator team standing in front of one

of the MAMI B 180 end-magnets. The common accelerating
section is located between the magnets on the right-hand side,
the separated return paths are on the left-hand side.

9.0 MV / turn max gain

LINAC I (4.90GHz)

n 1500MeV
Extractio

43 recirculations
In
je
ct
io

Fig. 4. Photograph of the three high-resolution magnetic spec-

n
85
5M

10 m trometers in the A1 area. Spectrometer A (red) is on the left-

eV

hand side, B (blue) in the center, and C (green) is on the

right-hand side.

B max=1.539 T LINAC II (2.45GHz)

tiatives , abbreviated SFB), a funding scheme used by
9.3 MV / turn max gain
the German funding agency Deutsche Forschungsgemein-
schaft (DFG) to support new initiatives for a limited pe-
Fig. 3. Layout of the MAMI C microtron. riod of time.

The design and construction of the 1.5 GeV Harmonic
Double-Sided Microtron (HDSM, design by K.-H. Kaiser
et al.) [2] is poised to shatter that boundary. Building
on the experience with the previous microtrons, the chal-
lenging HDSM design relies on two parallel accelerating
sections joined by four inhomogeneous 90 end-magnets
with a weight of 250 metric tons each (see g. 3 for a lay-
out). The strong vertical defocusing at the entrance and
exit of the magnets is compensated by a radial gradient
eld. Meeting the microtron coherence condition within
the conned space of the existing experimental area forces
the fundamental accelerating frequency to be twice the
frequency of the MAMI B microtron. Phase stability con-
siderations require to leave one of the two accelerating
sections at the present MAMI B frequency. The installa-
tion of the four HDSM magnets has been completed, and
commissioning is expected to start soon.
As shown in table 1, important milestones parallel
to the technical developments were the establishment of
the Sonderforschungsbereiche ( Special Research Ini-
3 Experimental equipment at MAMI
The broad physics program at MAMI requires an equally
broad range of experimental equipment, from high-resolu-
tion magnetic spectrometers to large acceptance detectors.
Most of the electron scattering instrumentation has been
provided by the Institute, a large fraction of the equip-
ment for the tagged photon experiments has been con-
tributed by the user community. For electron scattering
experiments, the three-spectrometer system oers an un-
precedented combination of momentum resolution, solid
angles, and momentum range. The parameters are given
in table 2, a photograph is shown in g. 4.
For experiments with real photons, the Glasgow-Mainz
bremsstrahlung tagging system, located in the A2 area,
provides photons of known energy and ux. Circularly po-
larized photons can be obtained from the bremsstrahlung
of polarized electrons, linearly polarized photons from an
oriented crystal radiator. The detection equipment is fo-
cused on charged and neutral particle detection in a large
212 The European Physical Journal A
Fig. 5. The DAPHNE large acceptance detector.
Fig. 6. The Crystal Ball detector (left) and TAPS (right).
solid angle and energy range. The Saclay-built DAPHNE
detector (Detecteur a grande Acceptance pour la Physique
photo-Nucleaire Experimentale) uses a combination of
proportional wire chambers and layers of scintillation
counters and absorbers for charged particle and photon
detection. A three-dimensional drawing is shown in g. 5.
The TAPS detector (original abbreviation for Two-
Arm Photon Spectrometer) can be arranged in dierent
congurations. Its 528 BaF2 crystals give good energy res-
olution for photon detection. Charged particles can be
identied via the ratio of fast and slow scintillation light.
The newest addition to the experimental arsenal is the
Crystal Ball detector which has seen prior service at high-
energy facilities like SPEAR, DORIS, and the BNL AGS.
Its central detector consists of 672 NaI crystals, again opti-
mized for photon detection. The electronic readout system
has been modernized and equipped with 80 MHz ash-
ADCs. In the rst experiment, the Crystal Ball will be
used in combination with TAPS as a forward detector (see
g. 6 for a picture of the setup) to measure the + mag-
netic moment via the angular distribution of the decay
photons in the + + transition (from the high-mass
tail of the + to its low-mass tail).
Of particular importance for studying the spin degrees-
of-freedom of the nucleon has been the addition of the
Bonn frozen-spin polarized H and D target (see g. 7 for
a picture). Its low magnetic holding eld and open geome-
try make it an ideal match for large acceptance detectors.
Fig. 7. The Bonn frozen-spin polarized target during a re-
polarization cycle. The polarizing magnet is on the left-hand
side, the pumping unit on the right-hand side.
In combination with the DAPHNE detector it has been
used for studying the GDH sum rule on hydrogen and
deuterium.
4 Selected experiments
The following sections will give examples for experiments
that are characteristic for the MAMI physics program and
that have had a large impact on the eld.
4.1 Real Compton Scattering and the Polarizability of
the Proton
The electric and magnetic polarizabilities of the nucleon
are static quantities that characterize the response of the
system to external electric and magnetic elds. Since the
highest elds that one can produce in the laboratory are
much too weak to have a measurable inuence, the best
approach is to derive the polarizabilities from the energy
and angular dependence of real photon scattering (RCS)
at low photon energies. The experiments are challenging
since the cross sections are very small and, above pion pro-
duction threshold, there is a large background of photons
from 0 decays.
The experimental results obtained with the TAPS de-
tector and the bremsstrahlung tagging system [3] require
signicant theoretical corrections and interpretation (see
g. 8 for an example).
The cross section is dominated by scattering o the
charge and the magnetic moment of the proton. In addi-
tion, the incident and outgoing photons can couple to an
exchanged pion. Finally, the polarizabilities enter linearly
only in the low-energy expansion. In practice, higher-order
terms need to be incorporated. A dispersion relation anal-
ysis [4] of the entire body of Compton scattering data
shows that the proton is a very sti objects, i.e. it does
not deform much under the inuence of external static
elds.
 B.A. Mecking: Twenty years of physics at MAMI What did it mean? 213

Dispersion relation
N contribution

asymptotic + N

asymptotic contribution

MAMI data DR analysis

JLab
LEX analysis

Fig. 8. Photon energy dependence of the Compton scatter- Fig. 9. Q2 -dependence of the generalized polarizabilities
ing cross section at a scattering angle of 135 . The dispersion E (Q2 ) and M (Q2 ).
relation analysis is the thin black line marked DR.

4.2 Generalized nucleon polarizabilities and virtual

Compton scattering
The concept of the nucleon polarizabilities can be ex-
tended to virtual photons. This leads to six general polar-
izabilities which are functions of the momentum transfer,
Q, and can be accessed in the ep e p() reaction. The
dierential cross section is dominated by the Bethe-Heitler
process and by Born terms which can be calculated from
QED and the known proton form factors. These contri-
butions have to be subtracted from the measured cross
sections.
At MAMI, the process was measured by using two
high-resolution magnetic spectrometers to detect the scat-
tered electron and the recoiling proton, and using the
missing-mass technique to identify the (undetected) pho-
ton in the nal state [5]. Fig. 10. Beam asymmetry for 0 photoproduction as a func-
From the MAMI data (see g. 9), the generalized tion of the 0 production angle.
polarizabilities E (Q2 ) and M (Q2 ) have been deter-
mined using a dispersion theoretical analysis [4]. The
data are well described by taking the asymptotic and they have very strong photon energy dependence. Also,
the N -contribution into account; this demonstrates the the reaction close to threshold typically involves low-range
importance of the pion cloud contribution to the nucleon particles which are di cult to detect.
polarizability. The experimental setup for the 0 -photoproduction
part of the program [3] used the tagged photon beam and
4.3 Near-threshold 0 production the TAPS photon spectrometer. Using an oriented crystal
as a radiator, linearly polarized photons were produced
The physics goal of this program is to test the predictions to get the desired polarization information from the beam
of PT which is an exact representation of QCD for the asymmetry. The TAPS crystals were arranged into 7 inde-
limiting case m 0. To compare to the real world, an pendent arrays covering the entire 0 angular range. As an
extrapolation to physical pion mass required. This intro- example for the results, g. 10 shows the beam asymmetry
duces low-energy constants into PT which need to be de- as a function of the 0 c.m.s. angle.
termined from experiment. Once the low-energy constants The experimental results show that PT has consider-
are known, the predictive power of PT can be tested. able predictive power. This nding, in combination with
The experimental information required are precise 0 other tests of PT, has increased dramatically the con-
and + photo- and electro-production dierential cross dence in the validity of PT. The most important appli-
sections close to threshold where PT is expected to be cation in other areas is the use of PT for extrapolating
valid. In addition, polarization data are necessary to sep- LQCD calculations.
arate the contributing multipoles. The experimental chal- The compute power required for LQCD calcula-
lenges are considerable: the cross sections are small, and tions depends on the pion mass as m7 . Therefore, the
214 The European Physical Journal A

Fig. 12. GDH integral as a function of the upper limit max .

Fig. 11. LQCD results for the proton magnetic moment. The
data points give the results of the calculations for dierent val- Table 3. GDH Results from ELSA and MAMI.
ues of m 2 . The physical values of p and m 2 are marked by Source of Information k [MeV] IGDH [b]
the red cross. The short blue arrow indicates the typical PT
extrapolation range; the dotted blue arrow indicates the typical MAMI 200 800 226 5 12
LQCD extrapolation range. The best theoretical extrapolation PRL 87 (2001) 022003
is given by the green solid line. ELSA 800 2900 27.5 2.0 1.2
PRL 93 (2004) 032003
Low-energy extrapolation 140 200 27.5 3
calculations are performed far away from the physical pion MAID (2003)
mass. The generally accepted technique to extrapolate to High-energy extrapolation 2900 14
PLB 450 (1999) 439
the physical pion mass is to use the functional form for
the m -dependence given by PT. As an example, g. 11 Experimental sum 212 5 12
shows the result of a LQCD calculation [6] for the mag- Theoretical GDH integral 205
netic moment of the proton as a function of m 2 . With
the chiral extrapolation, the LQCD result gets close to the
true value.
experimentally) and above 2900 MeV were estimated the-
oretically and added to the experimental sum.
For the MAMI GDH experiment [7], circularly polar-
4.4 Experimental test of the GDH sum rule ized tagged photons from the bremsstrahlung of polarized
electrons were hitting a longitudinally polarized frozen-
The GDH sum rule for the nucleon was derived in 1966 by spin target. The DAPHNE detector in combination with
Gerasimov and, independently, by Drell and Hearn. The a forward scintillation counter and shower detector was
sum rule is based on fundamental assumptions: Lorentz used to measure the total cross section via identifying and
and gauge invariance, unitarity, and no-subtraction dis- adding up the individual reaction channels. In addition to
persion relations. The sum rule links the weighted integral testing the GDH sum rule, this experimental technique
over 3/2 (k)1/2 (k) to the anomalous magnetic moment also allows the measurement of the energy and angular
of the nucleon. 3/2 (k) and 1/2 (k) are the total hadronic dependence of the helicity-separated dierential cross sec-
photoproduction cross sections for total helicity 3/2 and tions for the individual reaction channels that contribute
1/2, respectively, as a function of the photon energy k. The to the total cross section.
integral over k extends from pion production threshold to From the MAMI and ELSA data, the GDH integral
innity, the weight factor is 1/k. has been calculated. It is shown as a function of the upper
This fundamental sum rule had never been tested due integration limit in g. 12. Table 3 gives the partial inte-
to the lack of appropriate experimental facilities. The test gral for dierent energy ranges; in the table, the statistical
of the sum rule requires a circularly polarized photon error is given rst, the second entry is the systematic error.
beam and a longitudinally polarized target, both with high Also listed are the contributions from photon energies be-
polarization. In addition, a detector to measure the total low and above the range that was covered experimentally.
cross section reliably needs to be available. Note that, because of the 1/k-weighting, the contribution
A major eort has been launched in Europe to test from threshold to 200 MeV is about the same as the con-
the validity of the GDH sum rule. For the proton, the low- tribution from 800 to 2900 MeV. Within the errors, the
energy part of the sum (200 800 MeV) has been measured nal experimental value agrees well with the theoretical
at MAMI, the contribution between 800 and 2900 MeV prediction, thus verifying the GDH sum rule.
has been investigated at the ELSA accelerator in Bonn. First preliminary data from a polarized deuteron tar-
Contributions below 200 MeV (that are di cult to access get are available. Extracting the GDH sum for the neutron
 B.A. Mecking: Twenty years of physics at MAMI
from these experimental data is a conceptual challenge
since the GDH sum for the deuteron is dominated by the
D pn process close to the 2.2 MeV threshold. Only
with major theoretical support will there be a chance
to extract the GDH integral for the neutron from the
deuteron data.
It is instructive to look at the impact the GDH ef-
fort has had on the community beyond just validating the
GDH sum rule. To address this important physics ques-
tion, a coherent eort of several European groups was re-
quired, an experience that will likely lead to more col-
laboration in the future. On the physics side, thinking in
terms of the sum rule created an awareness that there are
no isolated physics problems: all energy regions are in-
terrelated, sometimes in ways that are not immediately
obvious. An important consequence of the technical ad-
vances required for the GDH sum rule tests is that the
measurement of helicity observables has now become rou-
tine. These observables contain valuable information, e.g.
on the excitation of the nucleon resonances. Beyond the
total cross section, detailed information on the energy and
angular dependence for the dierent decay channels will
be required to make the best use of the information.
4.5 Nucleon electromagnetic form factors
In elastic electron scattering, the response of a compos-
ite system is given by its electromagnetic form factors.
These describe the probability that the system will stay in-
tact after absorbing a virtual photon characterized by the
four-momentum transfer Q. Form factors provide the ideal
meeting ground for experiment and theory. Electric (due
to the charges) and magnetic (due to the magnetization)
form factors can be separated by measuring the angular
dependence of the scattering cross section at constant Q2
(Rosenbluth separation). An intuitive picture of the nu-
cleon is provided by the spatial distributions of charge
and magnetization which can be obtained from the form
factors via a Fourier transform.
Separating the electric, Ge , and magnetic, Gm , form
factors using the Rosenbluth separation technique is dif-
cult when Ge /Gm  1 which is always the case for the
neutron. The solution is to use polarization transfer to
measure the Ge /Gm ratio. In practice, there are two pos-
sibilities: using polarized electrons on a polarized target,
or using polarized electrons on an unpolarized target and
analyzing the recoil nucleon polarization.
From the measured proton and neutron form factors
(and assuming charge symmetry), the u and d-quark con-
tributions to the nucleon form factor can be determined.
Information on the s-quark contribution can be obtained
by adding the results of parity violating scattering o the
proton (parity violation is due to interference between the
exchange of a virtual photon and a Z 0 ).
To derive neutron form factors from either deuteron
or 3 He data requires careful consideration of nuclear ef-
fects. Observables and kinematical conditions need to be
selected to minimize nuclear corrections. Systematic cal-
What did it mean? 215
Fig. 13. Neutron polarimeter part of the Ge /Gm setup.
culations for all observables, including polarization, have
been carried out by Arenhovel and collaborators [8].
For example, the calculations show that the asymme-
try AVeD in the quasi-free e-n scattering of polarized elec-
trons o a polarized deuterium target (with the deuteron
spin perpendicular to the direction of the momentum
transfer Q) is linearly dependent on Gne ; there is very lit-
tle background. The remaining sensitivity to the reaction
dynamics (nal-state interactions, isobar congurations,
meson-exchange currents, and relativistic eects) and to
the N N -potential can be quantied, and corrections can
be applied to the measured Gne value.
MAMI has made signicant contributions to all nu-
cleon form factors [9]. For the neutron, the ratio Ge /Gm
has been measured in quasi-free e-n kinematics with two
dierent polarization techniques: 1) using polarized elec-
trons o an unpolarized D target and analyzing the recoil
neutron polarization, and 2) using polarized electrons on
a polarized 3 He target. The neutron magnetic form factor,
Gnm , has been determined via the quasi-free eD e n(p)
reaction, and the strange quark contribution has been
measured via parity-violating electron scattering o the
proton [10].
A particularly interesting experimental technique was
developed for measuring the ratio Gne /Gnm in the quasi-free
D(e, e n)p scattering of polarized electrons o an unpolar-
ized D target and analyzing the longitudinal and trans-
verse polarization of the recoiling neutron. The setup of
the neutron polarimeter is shown in g. 13. The polarime-
ter is only sensitive to the transverse polarization compo-
nents. To access the longitudinal polarization component,
a magnet (which also duplicates as a sweeping magnet for
charged particles) is used to precess the longitudinal com-
ponent into the transverse plane. The experimental trick is
now to precess the dominant longitudinal component just
enough to compensate the small transverse component. By
varying the magnetic eld and thus the neutron precession
angle, a setting can be found where the transverse neutron
polarization disappears. Ge /Gm can then be directly cal-
culated from the precession angle. The great advantage of
this null measurement which was used at MAMI for the
rst time is that it is completely insensitive to the precise
216 The European Physical Journal A
Fig. 14. Neutron electric form factor, Gne , as a function of Q
2
2
in units of (GeV/c) . The red curve is the best t to the world
nucleon form factor data [11].
Fig. 15. Neutron (top) and proton (bottom) radial charge
distributions from a Fourier analysis of the world nucleon form
factor data [11]. The green solid lines show the pion cloud
contribution. Note that for the neutron r 2 (r) is plotted which
emphasizes the contributions at large radii.
values of the beam polarization and the analyzing power
of the neutron polarimeter.
The MAMI measurements o D and 3 He, in combi-
nation with measurements from other laboratories, have
now led to a consistent set of experimental data for the
neutron electric form factor (see g. 14). No generally ac-
cepted theoretical interpretation exists.
Friedrich and Walcher [11] arrived at an intuitive inter-
pretation by describing the nucleon as the sum of a bare
Fig. 16. Ratio of electric quadrupole, E2, to magnetic dipole,
M 1, strength in the N (1232) transition as a function
of Q2 .
nucleon and a polarisation part. By tting all proton and
the neutron form factor data, a picture emerges where the
neutron spends a considerable part of its time as a system
of a proton (located close to the center) and a (located
on average about 1.5 fm away from the center); see g. 15
for the charge distributions of neutron and proton. Sim-
ilarly, the proton can be viewed as spending part of its
time as a neutron- + combination.
4.6 Excitation of the (1232)-resonance
The physics motivation for studying the electromagnetic
excitation of the nucleon resonances is to understand QCD
in the strong coupling regime. The mass spectrum and the
quantum numbers of the nucleon excited states need to be
understood in terms of the relevant degrees of freedom and
the wave function and the interaction of the constituents.
The electromagnetic amplitudes for the N N tran-
sition are sensitive to the dierence between the quark
wave functions of the N and N . The spatial resolution
of the probe can be tuned by varying the momentum
transfer. The electric and magnetic parts of the transi-
tion can be separated using partial-wave analysis (PWA)
techniques.
The determination of the electromagnetic transition
form factors for the N N transition requires a large,
high-quality data set covering a broad kinematical range
in momentum transfer, excitation energy, decay modes
(, , , , ), and decay angles. Polarization information
is especially useful since it is sensitive to the interference
between overlapping resonances, or to the interference be-
tween a resonance and the background.
 B.A. Mecking: Twenty years of physics at MAMI
Fig. 17. Setup to use quasi-monochromatic transition radiation for K-edge imaging. The sample containing the Mo foil is
located in front of the pn-CCD detector.
For higher N resonances, the extraction of the transi-
tion form factor becomes quite involved: in the theoretical
description, Born terms, unitarity, and channel coupling
need to be taken into account. A full PWA is presently not
possible due to lack of data; especially polarization data is
missing. To compensate for the lack of experimental data,
the analysis is often constrained by assuming the energy
dependence of the resonance excitation.
A particularly interesting example is the N (1232)
transition. A full partial wave analysis is possible since the
(1232) is an isolated resonance, and the Watson theorem
constrains the phases of the helicity amplitudes. The spin
1/2 3/2 transition is dominated by spin-ip M 1; how-
ever, non-zero E2 and C2 multipoles are possible. This
would be a signature for a non-spherical charge distribu-
tion in the (1232) or could be caused by the virtual
photon coupling to a pion cloud.
At MAMI, a major experimental eort [3,12] was laun-
ched to determine the E2/M 1 ratio for the N (1232)
transition starting at the real photon point and extend-
ing the measurements to virtual photons. For real pho-
tons, the experiments used the tagging system to pro-
duce linearly polarized photons and TAPS to detect the
0 decay. For electroproduction, the three-spectro- avoids making the X-ray beam monochromatic with a slit
meter setup was used to measure the ep e p( 0 ) pro-
cess. Tilting the proton spectrometer provided access to
out-of-plane observables.
The Q2 -dependence of the E2/M 1 ratio is shown in
g. 16. The results show that the E2/M 1 ratio is small
(around 2%) and negative. This nding is at variance with
all models that consider constituent quarks, only. Models
that explicitly include the pion cloud can explain the data.
4.7 Production of low-energy radiation
In the context of the applied physics program [13], tech-
niques were developed that make use of the high-quality
MAMI electron beam for the production of high brilliance
X-rays. In addition to clarifying fundamental aspects
of radiation production, there are potential practical
What did it mean? 217
applications, e.g. in the material sciences, in biology, and
in medicine.
The most powerful X-ray sources in the energy regime
of interest (K-edge of oxygen at 0.53 keV to the K-edge
of iodine at 33.16 keV) are dedicated synchrotron radia-
tion facilities. Modern electron accelerators with their low
emittance electron beams may oer an attractive alter-
native. The interaction of these beams with matter have
the potential of producing high brilliance X-rays with a
tunable time structure. Processes of interest are transi-
tion radiation, parametric X-rays, undulator radiation,
the Smith-Purcell eect, and channeling radiation.
An illustrative example is the use of the 855 MeV
MAMI beam hitting a stack of 30 polyimide foils to pro-
duce hard X-rays via transition radiation [14]. Using a
highly oriented pyrolytic graphite crystal, an X-ray beam
of about 20 keV was prepared; its two-dimensional spa-
tial distribution was measured in a pn-CCD detector (see
g. 17 for details). Due to the crystal monochromator,
the X-ray beam had a correlation between position and
energy. By synchronously changing the electron beam di-
rection and the crystal position, the X-ray energy spec-
trum could be swept as a function of time. This technique
system, and thus does not reduce the ux. Using the X-
ray beam for K-edge imaging as a demonstration project,
a 2.5 m Mo foil hidden in a 100 times thicker copper foil
could be detected. Once perfected, this technique may be
used to image the human lung using xenon (mixed with
oxygen) as an absorber.
5 Experiment Theory interplay at MAMI
As already pointed out in the discussion of the experimen-
tal program, there is a very close and eective collabora-
tion between the theorists working in the Institute and
the experimentalists using MAMI. This tight coupling in-
creases the impact of the MAMI experimental program in
two important ways.
218 The European Physical Journal A

Fig. 18. Photos of the retirees. Top row from left to right: Hartmuth Arenhovel, Thomas Walcher, and Karl-Heinz Kaiser;
bottom row: Dieter Drechsel, Jorg Friedrich, and Hartmut Backe.

First, it provides the necessary corrections and facil-
itates the physics interpretation for quantities for which
the measurement strategy is clear and unambiguous. Typ-
ical examples are the calculations of the corrections to the
measured Gne values, or the corrections to the GDH in-
tegral to account for the unmeasured angular and energy
range.
Second, theoretical support starting in the early phase
of the experiment makes it possible to attack the deter-
mination of quantities for which due to the existence of
strong competing channels the measurement and anal-
ysis strategy is not obvious. Typical examples include the
extraction of the nucleon polarizability from p p, the
pion polarizability from p + n, and the magnetic
moment of the + (1232) from p p 0 .
The tight coupling between theorists and experimen-
talists at MAMI is unusual and unique. It clearly has been
a major contributor to the success of MAMI. The ingre-
dients necessary for the tight coupling are not easy to
identify, and cannot easily be transfered to other facilities.
Very likely, the early history of the MAMI project and the
personal inclination of the people involved on both sides
have played an important role.
6 MAMI funding and operation
Securing steady funding for operating a major electromag-
netic nuclear physics facility within a university environ-
ment has been a real challenge. Contributions from the
University of Mainz, the state of Rhineland-Palatinate,
and from the Federal Government are required to keep the
operation of MAMI nancially viable. A large fraction of
the federal funding is provided by the DFG in the frame-
work of the SFB which is meant to support new ventures
for a limited period of time, but not meant to support
steady-state operation. The success of MAMI in this di -
cult funding environment is testimony to the skills of the
people in charge of the Institute.
The MAMI physics program has been comprehensive
and more characteristic for a national facility than for
a university-based accelerator. No physics problem that
was worth attacking has been left out, even when it re-
quired additions to the experimental equipment and im-
provements to the accelerator (e.g.: the parity violation
program).
International collaborators have played an important
role at MAMI, contributing both ideas and experimental
equipment. They obviously felt welcome at the Institute;
the natural hospitality and curiosity of the population of
 B.A. Mecking: Twenty years of physics at MAMI
Mainz and the surrounding areas have likely had a bene-
cial eect, too.
MAMI has been an ideal training ground for students
who could experience all stages of an experiment, from
planning to analyzing and publishing the results. The stu-
dents were also trained in developing and using sophis-
ticated experimental equipment, a good preparation for
those who went on to pursue careers outside of nuclear
physics. In addition to educating students at MAMI, a
generation of young researchers has been trained who have
now gone out and successfully competed for faculty posi-
tions.
7 The legacy of 20 year of MAMI physics
What this 20-year period at MAMI will be remembered
for will depend on the range of interests of the person
asking the question.
The development of the single-sided microtrons,
MAMI A and B, has already changed the textbooks on
accelerators; the design and construction of the double-
sided MAMI C microtron will complete the microtron de-
velopment.
The quality of the experimental data will not be sur-
passed for a long time. Note that very often the accuracy
of the nal answer is limited by the accuracy of the theo-
retical corrections. Better accuracy can only be achieved
by an improvement of both the experimental results and
the theoretical interpretation.
Twenty years of MAMI physics have demonstrated
that a tight coupling between theory and experiment can
be mutually benecial. This coupling, which may not be
easy to reproduce at other facilities, has been a major
contributor to the success of MAMI.
Finally, 20 years of MAMI physics have demonstrated
that it is possible although with a lot of eort to
operate a major facility within the framework of a German
university.
What did it mean? 219
8 Summary
The development of the physics program and the experi-
mental facilities at the Mainz Microtron MAMI over the
last twenty years has been reviewed. MAMI and its user
community have been working at the forefront of electro-
magnetic nuclear physics. Novel electron accelerators and
experimental equipment were developed and have been
used for ground-breaking research into the structure of
nucleons and nuclei.
The six people, Hartmuth Arenhovel, Hartmut Backe,
Dieter Drechsel, Jorg Friedrich, Karl-Heinz Kaiser, and
Thomas Walcher (g. 18) who have retired (or are about
to retire) have played key roles in this development. They
have every right to be proud of what has been accom-
plished. I would like to take this opportunity to wish them
Happy Retirement .
References
1. A. Jankowiak, these proceedings.
2. A. Jankowiak et al., prepared for the 8th European Particle
Accelerator Conference (EPAC 2002), Paris, France, 3-7
June 2002.
3. R. Beck, these proceedings.
4. M. Vanderhaeghen, these proceedings.
5. N. dHose, these proceedings.
6. R. Young (Jeerson Lab), private communication.
7. A. Thomas, these proceedings.
8. H. Arenhovel et al., Z. Phys. A 331, 123 (1988).
9. M. Ostrick, these proceedings.
10. F. Maas, these proceedings.
11. J. Friedrich, T. Walcher, Eur. Phys. J. A 17, 607 (2003),
arXiv:hep-ph/0303054.
12. H. Schmieden, these proceedings.
13. W. Lauth, these proceedings.
14. F. Hagenbuck et al., IEEE Trans. Nucl. Sci. 48, 843 (2001).