Está en la página 1de 5

Energy landscapes of interacting randomly and periodically charged rigid

polyampholytes
Y. Rabin*+, ++, S. Panyukov++, m and M. Wilhelm+++, mm
+

Yukawa Institute for Theoretical Physics, Kyoto University, Kyoto 606-01, Japan
Department of Physics, Bar-Ilan University, Ramat-Gan 52900, Israel
+++
Department of Materials and Interfaces, Weizmann Institute of Science, Rehovot 76100, Israel
m
On leave from: Theoretical Department, Lebedev Physics Institute, Russian Academy of Sciences, Moscow 117924, Russia
mm
On leave from: Max-Planck-Institut fur Polymerforschung, D-55021 Mainz, FRG
++

We study the interaction between randomly and irreversibly charged rigid objects. We consider arbitrary parallel displacements of two rod-like polyampholytes and of two plates and calculate the statistical properties of the resulting energy landscape, such as the distribution of energy minima and the depth, width and density of typical energy wells, as a function of the
separation between the objects and of the Debye screening length. The possibility that stickslip phenomena may arise during relative motion of the objects is discussed. Other cases considered in this work are that of random distribution on one rod
and a periodic one on the other, and that of local periodic order on both rods.

1. Introduction

The statistical physics of interacting manybody systems deals with a thermodynamically large number of elementary units (atoms, spins, etc.), each of which is
described by a small number of parameters. This leads to a
class of models in which complex physical behavior arises
due to cooperative behavior of a large number of such
simple units. There is, however, a different class of interacting complex systems, of the type encountered in biology, in which the number of parameters necessary to specify a single unit (e. g., a protein) is very large. Due to the
inherent complexity of the individual units, even a small
number of such interacting objects can exhibit very complex behavior.
In this work we investigate a generic model of the latter
type, i. e., that of two interacting rigid objects on which a
random distribution of charges is irreversibly placed. Such
objects are extremely complex in the sense that complete
specification of the charge distributions requires precise
information about the location of each of the charges, the
number of which can be arbitrary large, depending on the
size of the objects. We allow for charges of both signs and
the resulting combination of long-range attractive and
repulsive forces leads to the appearance of complex interaction energy landscapes. We would like to stress that our
model differs from that of usual polyampholytes (i. e.,
polymers carrying charges of both signs) since we assume
that the charged objects are rigid and therefore both the
charge distribution and the conformation are quenched.
Recall that all polyampholyte models to date deal with
flexible polymers [1] and consider the thermal fluctuations
and the associated entropic elasticity of the polymer
chains.
In Sec. 2 we study the statistical properties of the interaction energy landscape of two parallel rod-like polyampholytes, on which positive and negative charges were randomly and irreversibly placed. The detailed derivation of
these results was presented in [2]. New results are derived
Fax: +972-3-535 3298
E-mail: yr@rabinws.ph.bin.ac.il

544

for the case of a random distribution of charge on one rod,


interacting with a periodic charge distribution on the other,
and for the case where charge distributions on both rods
are locally periodic, up to some finite correlation length. In
Sec. 3 we present the analogous results for two parallel
randomly charged plates and discuss the possibility of
stickslip motion when force is applied to one of the plates.
In Sec. 4 we discuss our results and point out the limits of
validity of our approach.
2. Parallel rods
2.1. Random charge distributions

Consider an ensemble of pairs of parallel rods of lengths


L and L9, respectively, on which positive and negative
charges with local charge densities qi x and qi x (the
index i takes the values 1 and 2; corresponding to the two
rods) are randomly and irreversibly placed. The frozen distribution of total charge on each rod is given by qi x
qi x qi x (i 1; 2). Note that although different realizations of the charge distribution will have, in general, a
different selfenergy, we will disregard the question of
how such charge distributions can be prepared and assume
that all possible realizations are equally probable.
The assumption that the charges are randomly distributed on the rods with average charge density q q q
leads to a Gaussian distribution of the quenched variations
of density dqi x qi x q, with the second moment
dqi xdqj x9 gdij dx x9

where g 3 q q is the number density of charges and


the bar denotes averaging over the ensemble of all possible
realizations of quenched disorder
R on the rods, with probability Wdq V exp2g1 dq2 dx.
The charges on the two rods interact via a screened Couq q
lomb potential
Vx x9 expj x x9 2 h2 = x x92 h2 ,
where j1 is the screening length and h is the distance
between the rods. When the two rods undergo parallel rela-

Acta Polym. 1998, 49, 544 548 i WILEY-VCH Verlag GmbH, D-69451 Weinheim 1998

0323-7648/98/1010-0544$17.50+.50/0

tive displacement u with respect to some arbitrary origin of


coordinates,
R R the interaction energy is given by
Eu
dx1 dx2 q1 x1 q2 x2 Vx1 x2 u, where the
integration is taken over all the points in the two rods. We
assume that L s L9 and consider only relative displacements for which the two rods overlap. In the following we
will not consider any effects associated with the presence
of counterions (other than Debye screening, which can
arise due to the presence of added salt), i. e., counterion
condensation, electrical double layers, etc. [3]. While such
effects may be of importance in reallife situations, they
are irrelevant to the main theme of this work: the interaction between randomly frozen charge distributions on rigid
rods and plates.
The energy Eu associated with a particular relative
shift of the two rods differs, in
R general, from the average
interaction energy E q2 L dx Vx, by an amount
dEu Eu E. In Fig. 1 we plot a typical energy landscape dEu that results from the relative shift u of two
rods. This plot was generated by a computer simulation of
the randomly frozen charge distributions on the rods, with
periodic boundary conditions on the random charge distribution on the longer rod (the period is much larger than
either L or h and should not have a significant effect on our
results).
In order to characterize the energy landscape in a statistical sense, we calculate the average number of energy
minima Nmin EDE with energy in the interval (E,
E DE). This yields
Nmin E

dE dEuhw2 dEu=wu2

extr

where the sum goes over the positions u of all the energy
extrema for which wdEu=wu 0 and the h-function
ensures that only minima are counted (w2 dEu=wu2 A 0).
For two parallel rods, u is a scalar and w2 dEu=wu2 is the
inverse squared radius of curvature of the energy minimum, which is a measure of the stiffness of the potential

Fig. 2. Plot of the number of minima per unit energy per unit
length, Nmin E=L9 , as a function of their energy dE , for different
vertical separations between the rods. The broken line corresponds
to h L and the solid line to h 0:1L (with g 1).

well. Again, the averaging is performed with respect to the


ensemble of randomly charged rods, of lengths L and L9 .
In order to calculate the average in Eq. (2) we express
the h-function as an integral over a d-function and exponentiate each of the two dfunctions in the resulting product. Since the energy dE is quadratic in the charge density
dq and the density distribution function Wdq is Gaussian,
the functional integration over dq can be performed analytically (the details are reported elsewhere [2]). For
unscreened Coulomb interaction between the charges on
the rods, we obtain the following universal form:
Nmin E

L9
p pmin y
g hL

where

p




y2
y2
2
pmin y 3
exp
5 exp

16p2
2p
25p

y erf

y
p
5 p

and
E
y3
g

Fig. 1. Plot of the energy landscape dEu (in units of gL1=2 ) as a


function of the relative displacement u (in arbitrary units of length)
for a randomly chosen realization of disordered charge distributions
on the rods. The crosses corresponds to h 1 and the solid line correspond to h 0:5.

Acta Polym. 1998, 49, 544 548


y

r
h
L

where erf is the error function. In Fig. 2 we plot the number


of energy wells per unit energy and unit length, Nmin =L9 , as
a function of dE, for several values of L=h. Although this
distribution is always peaked at a negative value of dE
there is a finite probability of observing a potential well
with a positive value of dE: This reflects the possibility of
having local energy minima located inside broad energy
maxima (see Fig. 1). Note that the probability of observing
an energy minimum with energy much lower than dEmin

Energy landscapes of interacting randomly and periodically charged rigid polyampholytes

545

decays exponentially with dE dEmin 2 and therefore


such minima are statistically insignificant. The distribution
is very broad and its width increases with L as L=h1=2 .
In the presence of screening, the average concentration
of potential wells (total number of minima divided by L9 ) is
s
1
m2 h
c
4
2p
m1 h
Rv
where mk h 3 2 0 dr wk Vr=wr k 2 are the second
moments of the corresponding derivatives of the potential.
For unscreened Coulomb interaction between the rods
(j 0) m1 l 1=h3 and m2 l 1=h5 , and thus c l 1=h (i. e.,
the distance between the energy minima is of order h). We
conclude that the density of energy minima increases with
increasing proximity of the rods and approaches the number
density of charges g at a distance h x 1=g, which corresponds to the limit of validity of our model (at such separations the charge distribution can no longer be considered as
continuous). In the presence of electrostatic screening, the
above qualitative picture remains valid in the range
g1 f h f j1 . At separations exceeding the screening
length, m1 l j1=2 h5=2 exp2jh and m2 l j3=2 h7=2
exp2jh and therefore c l j=h1=2 . The conclusion
that screening increases the number of energy minima
(although their depth is exponentially decreased compared
to the unscreened case) is quite unexpected.
The average depth of an energy minimum is
s
gm1 h
pL
dEmin
5
4
m2 h
and its characteristic width is


m2 h 1=2
wx
m3 h

Since this average width is of the same order as the average


distance between the typical minima, c1 , we conclude
that the characteristic minima form a densely packed random lattice. At distances smaller than the screening length
the average spacing of this lattice is of the order of the
spacing between the rods h and at larger distances it is
h=j1=2 . Both the curvature and the width of shallow
energy wells are small and since the probability of observing uncharacteristically deep wells is exponentially small,
the resulting energy landscape does not havep
a fractal character (i. e., is smooth). Note that jdEmin j V L, and since
for sufficiently large systems the energy of attraction can
easily exceed the thermal energy kB T, the frozen randomness of the charge distribution on microscopic scales may
result in pinning, i. e., the system will be trapped in one
of the typical energy minima. Such effects can be observed
by keeping the rods at a fixed distance and monitoring their
stickslip response to externally applied tangential forces.
This phenomenon will be discussed in more detail in the
following when we consider the interaction of charged
plates.
546

Rabin, Panyukov, Wilhelm

Another important result is the dependence of the stiffness of the average potential well, K w2 dEu=wu2 , on
the distance h between rods,
Kh Kh0 m2 h=m2 h0 1=2

Equation (7) shows that the distribution of the radii of curvature of the energy minima is invariant with respect to the
change of the distance between rods. Thus, when the two
rods approach each other, existing minima become progressively stiffer and deeper, just as in the case of rods with
charges that form a regular periodic lattice. At the same
time, new shallow minima appear and grow in depth as h is
reduced. The appearance of new minima is a characteristic
signature of random charge distributions on the rods and
has no analog in the periodic case, i. e., when the frozen
charge distributions on both rods are periodic, with the
same period k. In the latter case, the number of minima is
independent of the distance between the rods, and their
depth increases monotonically with decreasing separation
h.
Our main analytical results for the unscreened Coulomb
case can be rederived by a simple scaling estimate. Let us
replace the microscopic charge distribution by one that is
coarse-grained over a length scale h (the distance between
the rods). Each electrostatic blob contains gh charges of
both signs and thus the total charge per blob is of order
Qh lgh1=2 . Since there are L=h such blobs and the
average interaction energy between the rods vanishes
(assuming that there is no net charge on the rods), dE can
written as the sum of interaction energies of L=h neighboring blobs on the two rods. Each one of the contributions is
a random function, which takes the values lQ2h =h, and
therefore the total interaction energy is also a random
quantity with zero average and characteristic deviation
dE x l L=h1=2 Q2h =h l gL=h1=2 , in agreement with
the exact result, Eq. (5).

2.2. Randomperiodic case

We study the interaction energy landscape for two parallel rods: rod 2 is of length L9 with a randomly frozen density distribution q2 and rod 1 is of length L (L s L9 ) with a
frozen periodic density distribution
dq1 x qeiqx q eiqx ;

q 2p=d

where d is the period of the distribution.


In the following we will assume that both rods are very
long and neglect end effects. Introducing
the Fourier transR
ikxx9u
Vk
and
forms
Vx

x9

dke
R
dq2 x dk9eik9x dqk9 , the interaction energy can be
written as Eu
Eu Vq dqq qeiqu Vq dqq q eiqu

The calculation of the average number of minima is performed using the method described in [2] and the resulting
expression for Nmin is very similar to that in Eq. (19) in the
above reference,
Acta Polym. 1998, 49, 544 548

Zv

Nmin E L

3 Z
Y
dkj
dKK
6
2p
j1

"
2

exp ik1 E ik3 K gjqj L

2.3. Locally periodic case

#
Djl kj kl

10

jl

where the matrix of the second moments D is defined as


0
1
m0
0 m1
D @ 0
m1
0 A
11
m2
m1 0

What happens if we introduce local periodic order with


period d on both rods, with a finite correlation length rc ?
Clearly, when rc s h,L the interaction energy landscape is
not sensitive to the local periodic order and is identical to
that obtained for random charge distributions on both rods.
However, new results are obtained in the range rc S h,
which will be considered below.
Consider the charge density profile
dq1 x qeiqxvx q eiqxvx ;
where
jvx vx9j2 d2

with
mi q2i jVq j2

12

This gives an average concentration of energy wells


q
13
c
2p
which depends only on the wavelength of the periodic distribution of charge and is independent of the distance
between the rods. Thus, the number of energy minima for
the randomperiodic case is the same as in the case of identical periodic charge distributions on both rods. However,
while in the latter case the depth of energy minima scales
like L, in the present case the average well depth is given
by
p
p=2 p
dEmin
14
q gLjVq j
2
For screened Coulomb potentials,
p
Z
j x2 h2
iqx e
Vq dxe p
x 2 h2

15

dEmin is a function both the wavelength 2p=q and the distance between the rods, h. Obviously, this expression
vanishes in the limit jh S 1 (strong screening). In the
opposite limit, jh s 1 (no screening), we have to distinguish between the cases qh S 1,
s
2p qh
e
Vq x
qh
and qh s 1, for which
Vq x 2 lnqh
Since, in the case of random charge distributions on
both
p
rods, the depth of a typical minimum is dEmin V g L=h,
we conclude that for g h s 1this depth is larger in the randomperiodic case than in the randomrandom case.
Another difference between the two cases is that the number of such typical energy minima is independent of the
distance between the rods in the randomperiodic case and
decreases with the separation in the randomrandom case.
Acta Polym. 1998, 49, 544 548

q 2 p=d

jx x9j
rc

This gives the charge density correlator


gx x9 dq1 xdq1 x9 2 jqj2 e2p


2p x x9
cos
d

jxx9j=rc

and we obtain the same form of Nmin E as in the random


random case, where we replace the expressions for the
moments by
2 jqj4 rc 2k 2
q Vq
mk x
p2
and set g 1. As in the randomperiodic case, the concentration of wells is given by
c 1=d
and the average well depth is

p
dEmin L q2 rc LjVq j

3. Parallel plates

Similarly to the case of two parallel rods, the study of


the statistical properties of the interaction energy landscape produced by the relative displacement u (u is now a
vector in two dimensions) of two parallel plates of areas A
and A9 (A s A9 ) separated by a distance h shows that the
typical landscape consists of closely packed minima and
maxima of similar depth and width [2]. In the unscreened
Coulomb case the density of these minima is c x h2
(their width and the spacing between them are of order h).
At distances exceeding the screening length we obtain
c x j=h. At separations smaller than the screening length,
the depth of a typical energy minimum is dEmin x gA1=2 (g
is the average number of charges per unit area), independent of the distance between the plates. At distances
exceeding the screening length, the depth decays exponentially with the separation between the plates.
In order to illustrate how this energy landscape may lead
to stickslip behavior, we consider the following gedanken
experiment (the geometry is similar to that of the surface
force apparatus, operated in the shear mode [4]): A spring is
attached to one of the plates (say plate 2, with area A9 ) and
the other plate (plate 1, with area A) is moved by a distance u

Energy landscapes of interacting randomly and periodically charged rigid polyampholytes

547

parallel to it. If the distribution of charge on the plates is uniform, there will be no restoring forces associated with this
displacement and consequently plate 2 will not move. The
presence of frozen random charge distributions on the plates
leads to the appearance of a large number of energy minima,
which correspond to different relative parallel displacements of the plates. Since, for macroscopic plates, the typical energy associated with these minima scales as A1=2 , for
sufficiently large plates it may become much larger that the
thermal energy kB T and the plates will get stuck in one of
the minimal energy configurations. Plate 2 will move
together with plate 1 (stick phase) until the force on it
exceeds some critical value fcrit , at which point it will recoil
back (slip phase). This value can be estimated from the characteristics of the typical energy well and for h f j1 we
obtain fcrit x gA1=2 =h. At larger distances, screening comes
into play and this force decays exponentially with separation between the plates (no stick phase is expected at large
separations). Conversely, the stickslip phenomenon can
be suppressed by the addition of salt, even when the separation between the plates is fixed.
When the critical force is exceeded, plate 2 will recoil
back to a position in which the spring force becomes sufficiently small. The process will repeat itself as long as we
continue to drag plate 1, and stick-slip motion of plate 2
will be observed. Note that each time the critical spring
force is exceeded, plate 2 recoils to a different new equilibrium state, which corresponds to an energy minimum that
differs in general from the initial one (such minima are
densely distributed in the space of relative displacements
of the two plates) and, therefore, a statistical spread of
equilibrium positions of the lower plate will result.
4. Discussion

In this work we studied the energy landscape generated


by relative translation of two random frozen charge distributions interacting through long-range electrostatic forces.
Two simple geometries were considered, for which the statistical properties of this energy landscape can be calculated analytically, namely, two parallel rigid rods and two
parallel rigid plates. When the total charge on each of the
objects vanishes and the Coulomb interaction between
them is unscreened (i. e., when there is no characteristic
length scale associated with the interaction potential), the
interaction between the objects becomes significant and
energy minima appear when the separation between these
objects becomes of the order of their dimensions (when the
two objects are of different sizes, this happens when the
separation is of the order of the linear dimensions of the
larger of the two). When the separation is further
decreased, these energy wells deepen and new wells
appear. We calculated the corresponding energy distributions, the depth and the width of the characteristic energy
wells and the spacing between them and found that both
the width and the spacings are of the order of the separation
between the objects. This indicates that the typical energy
wells are densely packed in the space generated by the
relative translation of the objects.
548

Rabin, Panyukov, Wilhelm

An entirely different picture arises if we consider the


interaction between a randomly charged rod and one on
which a periodically varying charged distribution is irreversibly placed. In this case, the density of typical energy
minima depends only on the wavelength of the periodic
distribution and not on the separation between the rods.
Yet another case considered in this work is that of local
periodic order on both rods. As expected, the energy landscape reduces to that of the randomrandom case at
separations that exceed the charge correlation length. At
smaller separations between the rods, the landscape is qualitatively similar to that of the randomperiodic case.
We would like to comment on the limitations of our
model. The theory is applicable to separations down to the
average distance between the charges (h F g1 for rods
and h F g1=2 for plates), at which point our assumption
of continuous charge distributions on the objects breaks
down. There is also an upper cutoff of the distance between
the objects: as shown elsewhere [2], the method of averaging over frozen disorder used in this work becomes inaccurate for separations that exceed the linear dimension of
the longer rod, h F L9 . The resulting corrections are of limited physical interest since in this limit the average number
of energy minima is of order unity and their depth tends to
zero. A more restrictive condition follows from the fact
that we did not explicitly introduce the constraint of fixed
total charge on each of the rods (we assumed that it is fixed
and, consequently, did not average over the distribution of
total charge on the rods). It can be shown that accounting
for the fixed total charge constraint leads to corrections of
order h=L (L is the length of the shorter rod) to our results.
Finally, in this work we considered only parallel rods and
plates. Although analytical results can be obtained for
other geometries as well, this increases the number of free
parameters and obscures the simple physical picture presented here.
Acknowledgements

We thank I. Freund, I. Kanter, Y. Kantor, J. Klein and


A. Tkachenko for helpful comments. SP acknowledges the
hospitality of the Department of Physics, Bar-Ilan University, and MW acknowledges the hospitality of the Department of Materials and Interfaces, Weizmann Institute of
Science, and the support of the Minerva Foundation. YR
acknowledges financial support through a grant from the
Israel Science Foundation and expresses his gratitude to
the Yukawa Institute for Theoretical Physics, Kyoto University, where this work was completed.
References
[1] Y. Kantor, M. Kardar, Phys. Rev. Lett. 1991, 14, 421.
[2] S. Panyukov, Y. Rabin, Phys. Rev. E 1997, 56, 7053.
[3] J. Israelachvili, Intermolecular and Surface Forces, 2nd ed., Academic Press, London 1992.
[4] J. Klein, D. Perhia, S. Warburg, Nature 1991, 352, 143.

Received December 25, 1997


Final version June 10, 1998

Acta Polym. 1998, 49, 544 548

También podría gustarte