Documentos de Académico
Documentos de Profesional
Documentos de Cultura
10*, is about three times as large as the mea- longer observation times. The soliton size 14. F. Schreck ef ai. Phys. Rev. Lett. 87. 080403 (2001).
sured atom number in tlie soliton, it is likely could then be measured in situ, as well as its 15. See, for instance. R. Grimm. M, Weidemler, Y. B.
Ovchinnikov, Adv. At. Moi Opt. Phys. 42, 95 (2000).
that during the 50 ms phase where a is changed lifetime. The study of coherence properties of
16. S. Inouye et ai. Nature 392, S1 (1998).
from 0 to negative values, one or several col- solitons and of binary collisions between soli- 17. J, L. Roberts, N, R. Claussen, S. L Cornish. C. E.
lapses occur until the critical number for a tons is an immediate extension of the present Wieman. Phys. Rev. iett. 85, 728 (2000).
stable BEC is reached. Indeed, the collapse time work. 18. M. D, Barrett, J. A. Sauer, M, S. Chapman, Phys. Rev.
constant is predicted to be much less than 50 ms Lett. 87, 010404 (2001).
19. In the vertical direction, a residual astigmatism of the
for our experimental conditions (2/). During References a n d N o t e s imaging system gives a resolution limit of 16 |i.m.
the transfer into the 1D waveguide, the BEC is 1. See, for example, the recent special issue: Chaos 10,
20. Over this distance, the change in magnetic field due
471 (2000).
transformed into a soliton and the noncon- to the curvature is 0.1 G, and therefore the change in
2. P. A. Ruprecht, M. ]. Holland, K. Burnett, M. Edwards,
densed cloud is clearly observed at guiding Phys. Rev. A 5 1 , 4704 (1995).
the scattering length is negligible (Fig. 2).
21. Y. Kagan, A. E, Muryshev, G. V. Shlyapnikov, Phys.
times up to 6 ms a.s a broader background 3. C C. Bradley, C, A. Sackett, R. C. Hulet, Phys. Rev.
Rev. Lett. 8 1 , 933 (1998).
distribution. Nonadiabatic projection of the iett 78, 985 (1997),
22. L. D, Carr, Y. Castin, in preparation,
4. J, L Roberts et ai., Phys. Rev. Lett. 86, 4211 (2001).
BEC from the confining onto the expulsive 5. V. E. Zakharov, A. B, Shabat, Sov. Phys. JETP 34, 62 23. We are grateful to K. Corwin. M. Olshanii. C.
potential is expected to play a negligible role (1972). Shlyapnikov, C. Williams, V. Venturi. and B, Esry for
important contributions t o this work and to J.
here, according to numerical simulations (22). 6. V. M. Prez-Garcia, H. Michinel, H. Herrero, P/iys. Rev.
Dalibard and C. Cohen-Tannoudji for useful discus-
A 57, 3837 (1998).
At longer times, the noncondensed atoms sions. Supported by the DAAD (F.S.), the European
7. L D. Carr, M. A. Leung, W, P. Reinhardt.7. Phys. B 33.
spread apart and become undetectable. Thus. 3983 (2000),
Union network CT 2000-00165 CAUAC (CF.), and
the NSF MPS-DRF 0104447 (L.D.C.). This work was
during the propagation phase the soliton de- 8. S. Burger et ai. Phys. Rev. Lett. 83, 5198 (1999).
supported by CNRS, Collge de France and Rgion
couples itself from the noncondensed fraction, 9. J. Denschlag et ai. Science 287, 97 (2000).
Ile de France. Laboratoire Kastler Brossel is Unit
10. E. Tiesinga, B. J. Verhaar, H. T. S. Stoof, Phys. Rev. A
resulting in a nearly pure soliton. 47, 4114 (1993),
de recherche de l'Ecole Normale Suprieure et de
l'Universit Pierre et Marie Curie, associated w i t h
Finally, removal of the expulsive axial 11. V. Venturi, C. Williams, personal communication.
CNRS,
12. M.-O. Mewes, C. Ferrari, F. Schreck, A. Sinatra, C.
potential will allow us to extend the stability Salomon, Phys. Rev. A 6 1 , 011403R (2000).
domain toward lower values of N\a\ and 13. F. Schreck et ai. Phys. Rev. A 64, 011402R (2001). 19 February 2002; accepted 5 April 2002
capping agent consisting of a combination of still unknown (//), quantum confinement in Si diffusing Si NCs dispersed in nonaqueous
hydrogen and alkoxide, using solvents heated has led to efficient PL (12, 3), and radiative organic solvents were measured at a Pt elec-
and pressurized above their critical points to transitions have been observed (2) in a variety trode {6). Discrete steps associated U'ith
temperatures between 350 and 500C (10). of Si nanostructures, including thin wires {14), single-electron charging and a large central
These NCs are efficient emitters of visible light, dots {10), and porous Si (2,15). For the exper- gap between the onset of oxidation and
with quantum yields between 5 and 20"/;) and iments reported here, octanol-, octene- and oc- reduction, characteristic of the energy differ-
size-tunable color that can range from blue to tanethiol-capped Si NCs were prepared in su- ence between the highest occupied and low-
red. different from H- or oxide-tenninated sam- percritical cydohexane according to the meth- est unoccupied molecular orbitals (the HO-
ples in the same size range reported earlier. This ods described in (W). The Si NCs are readily MO-LUMO gap), were observed. Typical
efficiency is rather remarkable, given that the soluble in a variety of organic solvents, includ- electrochemical responses for different solu-
indirect band gap of bulk Si makes it a poor ing those used here: TV^A^'-dimethylformamide tions of Si NCs are given in Fig. 1. with NC
c;mdidate for a light-emitting material. Al- (DMF) and acetonitrile (MeCN). sizes of (I) 2.77 0.37. (ii) 2.96 0.91, and
though the precise origin of the light emission is The eleetrochemical properties of freely (iii) 1.74 0.67 nm. For example, in Fig. I A,
there are as many as five well-resolved dif-
ferential pulse voltammetry (DPV) peaks be-
tween 0 and -2.1 V. These regularly spaced
peaks are reversible and highly reproducible.
The observed response was stable on repeti-
-008 0 tive potential cycling over long time periods,
0.56 O
with no evidence of fouling or film formation
-1.01 0.45
-1 42 0.41
on the Pt electrode surface.
>1.69 0.27 I '(B^A) This DPV response is analogous to that
-2,0 0 31
J.18 Oa
observed by Murray and co-workers (3) for
thiol-capped metal clusters, who introduced
the term quantized double-layer charging to
differentiate this collective electrochemical
response from Coulomb blockade phenome-
na observed for single-charge injection to
isolated individual dots in scanning tunneling
microscopy (STM) experiments, also called
"addition spectra" (17). Unlike the electrical
response measured by STM, the electrochem-
ical response is limited by quantum dot dif-
fusion to the electrode surface. Nonetheless.
the charging energy (JL, required for electron
or hole addition is the same for STM and
DPV measurements. For example, we recent-
ly observed an eleetrochemical Coulomb
staircase based on two nanometer-sized elec-
trodes connected in series through a solution
containing a redox couple (/iV).
The small size of the NCs causes electron
addition to be quantized. The charging ener-
gy required to add the first eleetron |JL, equals
the electron quasiparticle energy _.,, which
depend.s on the size-dependent shift in the
LUMO e,.|" and the electron "self-energy"
i,,p which results from its image charge in
the surrounding dielectric media: e^, =
,.," + S^|. Calculated values of S^, typically
range from 0.2 to .5 eV for Si nanocrystals
in the size range studied here (/9). A second
electron must also overcome the electron-
electron Coulomb interaction./^,, _,, within the
particle to add to a negatively charged NC,
0 -1 -2 -3 -2 Al 2*^* = ^-2 - M-i = *^Pi.rf- ^^ Si dots, the
calculated ,,, values are almost entirely de-
E(vs.Ag QRE) E (vs. Ag QRE)
tennined by Coulomb energies, because the
Fig. 1. Cyclic voltammograms (right panels) and differential pulse voltammograms (left panels) for six lowest-energy conduction levels are very
several batches of Si nanoparticles in 0.1 M tetrahexylammonium perchlorate (THAP) DMF close in energy. A classical estimation of
solution. /, current; E current peak potential, Af , potential difference between two succeeding J^y e2 * ^/2Cf^c- reveals that the sub-attofard
peaks; 1^^^^, ECL photocurrent from the photomultiplier tube; QRE, quasi-reference electrode. Tbe (aF) capacitances for NCs give rise to A, j ' " '
NC'5 size and dispersion were (A) 2.77 0.37, (B) 2.96 0.91, and (C) 1.74 0.67 nm. Cyclic > > A:jrevenat room temperature, leading to
voltammetric ECL voltage curves are plotted in (B) and (C). The dotted curves in (A) represent the
response of tbe blank supporting electrolyte solution. The dotted curves in (B) and (C) are the discrete charging events in the DPV scans (3,
responses for different initial DPV scan potentials. 8, 18, 20). If C^^. is independent of the
number of electrons injected, consecutive Light emission by charge injection, or 2A). which may indicate that the eiectrogener-
charge injection should occur in regular po- ECL, into freely diffusing NCs occurred un- ated oxidized forms are more stable. Light emis-
tential steps, AK - elC^^ (i. 8). Although der repetitive electrode potential cycling (as sion was not observed when the applied elec-
the DPV peaks in Fig. 1 were not separated in Fig. I, B and C) or pulsing (Fig. 2) be- trode potential was not sufficient to generate
by exactly the same AF, the average AK ^ tweeti NC oxidation and reduction. In ECL both the negatively and positively charged spe-
0.4 V corresponds to a capacitance of approx- experiments, electron-transfer annihilation of cies. The ECL spectrum (Fig. 3A) obtained
imately 0.4 aF per cluster (Fig. lA); this is eleetrogenerated anion and cation radicals re- from the annihilation (Hq. I) in MeCN. where
comparable to the value reported for organic- sults in the production of excited states {20, the applied electrode potential wa.s pulsed be-
capped noble metal clusters (4). The A^^ 21) tween the oxidation and reduction potentials
compares favorably with values reported for (double potential step) in 100-ms steps (20, 21),
(I)
organic-capped Au nanocrystals and with re- showed a maximum at 640 nm.
cent calculations by Franceschetti and Zunger (2) Higher intensity light emission x)m the Si
{9) for Si quantum dots with A; ,''* on the In this case, R" and R* refer to negatively and NC solution was observed when eoreactants
order of 0.4 to 0.6 eV. With increasing NC positively charged Si NCs eiectrogenerated at were added, which help overcome either the
charge. At' decreased measurably, perhaps the Pt electrode, which then react in solution to limited potential window of a solvent or poor
because of multielectron effects or NC size give the excited state R*. ECL was not observed radical anion or cation stability {20). For exam-
dispersity, which can smear the observed re- through an annihilation-type mechanism 1 for ple, by adding excess C,O^^ to the NC solution
sponses shown in Fig. 1, A and B (4). The thioglycerol-capped CdS quantum dots {9) be- (Fig. 2B), ECL only requires hole injection and
electrochemical behavior seen in these fig- cause of the instability of eiectrogenerated reac- can be obtained by simply oxidizing the NCs. In
ures was very sensitive to the NC size vari- tants. In order for ECL to occtir through mech- this case, the oxidation of oxalate produces a
ation (0.37 nm in Fig. lA and 0.91 nm in anism 1, the NCs must be chemically stable and strong reducing agent, CO^"". which can inject
Fig. IB). maintain their charged states long enough to an electron into the LUMO of an oxidized Si
The DPV responses in Fig. 1 are from dif- transfer charge upon colliding with oppositely NC to produee an excited state that then emits
ferently sized NCs. Common features include charged NCs in solution. The relative ECL in- light (Figs. 2B and 3B) (20. 22).
the appearance of a large central gap (\L^ - \L , tensity was greater in the potential region where
> 1.3 V; the subscript " - I " refers to the hole anionic NCs are eiectrogenerated (Figs. lC and R^'+CO, R*+CO, (3)
chemical potential) and the general absence of
DPV peaks in the positive potential region. This
last feature was not a limitation of the available
electrochemical window, as seen in Fig. lA,
where the DPV response in the absence of Si
NCs is given. Electron injection occurred as
discrete charging events; however, NC oxida-
tion {hole injection) was not generally quantized
although calculations suggest that it should be
(19), and the DPV response was characterized
by a continuous increase in current with poten-
tial indicative of multiple charge transfers. Nev-
i
[
Similarly. ECL was observed in the potential that depends sensitively on excitation energy, as 4A). As the electrode potential is made more
region for NC reduction through the addition shown in Fig. 4B, in which the excitation wave- positive, holes are injected into the partiele.
of excess SjO^^" to solution. Reduction of length from top to bottom was between 360 and Concurrently, oxalate is oxidized and under-
persulfate produces a strong oxidant, SO^" 520 nm at 20-nm intervals. The shift in peak goes a following ehemical reaction generat-
which can then react with the negatively position with excitation energy is unlikely to be ing the powerful reducing agent C02~. This
charged NCs by injecting a hole into the caused by polydispersity, because the NCs were intermediate injects an electron across the
HOMO, producing an excited state (Figs. 2C size-selected by column chromatography, Fur- partiele surface (Eq. 3) and makes emission
and 3C) {23, 24) themiore. the PL energy is considerably greater possible through surfaee electron-hole re-
than the absorption edge, indicating that the combination. The other two ECL processes
R-+SO.,-->R* + S O / " (4) lowest-lying energy levels do not result in (Eqs. 1 and 4) are assumed to occur through
A higher concentration of Si NCs was used strongly radiative transitions. Qxialitatively, the a simitar mechanism. The observed ECL
for the persulfate measurements, which may ECL and DPV data are as expeeted for Si NCs. emission insensitivity to core size and eap-
account for the higher ECL intensity and Quantitatively, however, the energetic dif- ping agent supports the surfaee state conjec-
lower decay rate apparent in Figs. 2C and 3C ference between the PL and ECL of approx- ture (/ /) and this proposed mechanism. Thus,
in comparison to Fig. 2, A and B, and Fig. 3. imately 0.8 eV is more difficult to understand the difference in light emission through PL
A and B. The solvent used may also play a and suggests that the emitting states are dif- and ECL from the same Si NCs most likely
role. After the ECL experiments, the solu- ferent. Previously, for Si NCs terminated results from the greater signifieanee of sur-
tions showed the same PL as the original with hydrogen atoms, the indirect band gap face states for charge injeetion as opposed to
solution, so no bulk degradation of the Si was reported to shift from the bulk value of photoinjection. Similar long-wavelength
NCs occurred. 1.1 eV to 2.1 eV for nanocrystals about 2 emission is found with porous Si produced by
nm diameter, and the direct transition ap- anodic etehing in the presence of organic
The ECL spectra in the above three cases all
peared to blue shift by 0.4 eV from its 3.4 eV surface modifiers (28).
show a maximum wavelength of 640 nm, which
bulk value over the same size range. In that
is substantially red-shifted from thai in the PL These results reveal that NCs of the ele-
case, violet PL (~365 rmi) was the most
spectra (Fig. 4B). The orange ECL emission mental semiconductor Si are more ehemieally
intense emission and was attributed to direct
was not sensitive to NC size or the capping robust than the eompound semiconductor
electron-hole recombination, whereas other
agent used. On the other hand, the Si NC PL is NCs studied to date. The Si NCs have the
less intense PL peaks (580 nm) were as-
size-dependent (10). A few important observa- ability to store charge in solution, which can
signed to surface state and phonon-assisted
tions and conclusions can be drawn from the subsequently lead to light emission upon
recombination (//). Undoubtedly, ECL de-
ECL data. First of all, the electrochemieal "tum- eleetron and/or hole transfer. This quality
pends more sensitively on surface ehemistry
on voltage" (that is, the potential gap in Fig. I, provides eiectrochemically sensitive opto-
and the presence of surface states. PL mainly
B and C) for radiative electron-hole annihilation electronic properties that may find future use
occurs through excitation and emission with-
betu'een positive and negatively charged NCs in new sensor technologies.
in the NC core, though the electron and hole
e.xceeds the optical transition energy. This ob-
wave functions can interact strongly with the
servation is consistent with electron and hole
NC surface. Despite a few notable exceptions References and N o t e s
injection into separate NCs requiring greater 1. A. P. Alivisatos, Science 2 7 1 , 933 (1996).
{25, 26), charge injection in a Si nanocrystal
energy than optical excitation. Second, the turn- 2. S, S. Iyer, Y. H. Xie, Science 260, 40 (1993).
is generally assumed to occur via its surfaee 3. S. Chen et ai., Science 280, 2098 (1998).
on voltage for ECL significantly exceeds [x, -
states, given the large surface area and the 4. S. Chen. R. W. Murray, S. W. Feldberg,;. Phys. Chem.
^L_^. Although the potential diffrence jji; - |JL_ , B 102, 9898 (1998).
possible presenee of many dangling bonds. If
enables eleetron and hole injection, and electron 5. P. Hoyer, H. Weiler, Chem. Phys, Lett. 224, 75 (1994}.
we consider the Si NC/oxalate coreaetant sys-
transfer between charged NCs in solution is 6 .,J. Phys. Chem. 99, 14096 (1995).
tem (Eq. 3) as an example, the Si cores have 7. S. Chen, L. A, Truax, J. M. Sommers, Chem. Mater. 12,
possible, the carrier energies are not large
band gaps greater than the energy separation 3864 (2000).
enough to produce optically radiative electron- 8. A. E. Hanna, M. T. Tuominen, M. Tinkham, Phyy Rev.
of the surface states. The core band gaps
hole recombination. This observation is consis- Lett. 68, 3228 (1992).
depend on the NC size, whereas the energy 9. S. K. Haram, B. M. Quinn, A. J, Bard,/ Am. Chem. Soc.
tent with previous observations of Si NCs. Ex-
separation of the surface states is only slight- 123, 8860 (2001).
citation energies for effieient PL significantly 10, J- D. Holmes ef al., J, Am. Chem. Soc. 123. 3743
ly affected by the particle size {I!, 27) (Fig.
exceeded the absorption edge, with PL intensity (2001).
n. J. p. Wilcoxon, C. A. Samara, P. N. Provencio. Phys.
Rev. 8 Condens. Matter Mater. Phys. 60, 2704 (1999).
12. M. Nirmal, L Brus, Ace. Chem. Res. 32, 407 (1999).
13. D. Kovalev et ai., Phys. Rev. Lett. 8 1 , 2803 (1998).
14. J. D. Holmes, K. P. Johnston, R, C. Doty, B, A, Korgel,
Science 287, 1471 (2000).
15. P- McCord, S. L. Yau, A. J. Bard. Science 257, 68
(1992).
16. General experiments for eiectrochemistry (such as cy-
clic voltammetry and differential pulse voltammetry)
ECL and ECL of Si nanopartJcles were done as follows; A
cylindrical Pyrex vial 1.2 cm in diameter was used as an
electrochemical cell, where a 1- or 2-mm Pt disk, Pt coil,
and silver wire served as working, counter, and refer-
ence electrodes, respectively. The ECL signal was re-
corded on a charge-cou pled device camera. ECL could
also be measured by a photomultiplier tube and record-
21. L R. Faulkner, A. J. Bard, in Electroanatyticat Chemistry. 24. A. W. Knight, G. M. Greenway, Anatyst 119, 879 28. H. C. Choi, J. M. Buriak, Chem. Mater. 12, 2151
A. J, Bard, Ed. (Dekker. New York, 1977), voL 20, pp. (1994), (2000).
1-95. 25. M. Shim, P. Cuyot-Sionnest, Nature 407, 981 (2000). 29. The support of this research by grants from NSF and
22. T. C, Richards, A. J. Bard, Anal. Chem. 67, 3140 26. C. Wang, M. Shim, P. Guyot-Sionnest, Science 291, the Robert A, Welch Foundation is grateftjlly ac-
(1995). knowledged. Thanks to P. Barbara, F. R. Fan, and M.
2390 (2001).
Buda for helpful discussions.
23. H. S. White, A. J. Bard, y. Am. Chem. Soc. 104, 5891 27. T. Makimura, Y. Kunii, N, Ono, K. Murakami, Appl.
(1982). Surf. Sd. 127-129, 388 (1998), 26 December 2001; accepted 3 April 2002