PHYSICAL REVIE% A VOLUME 34, NUMBER 4 OCTOBER 1986

Time-dependent electrolyte friction on charged spherical polyions

H. Ruiz-Estrada, A. Vizcarra-Rendon, and M. Medina-Noyola

Apartado Postal 14-740, 07000 Mexico Distrito Federa/, Mexico

R. Klein
Fakultat fur Physik, Uniuersitat Konstanz, D-7750 Konstans, Federal Republic of Germany
(Received 12 June 1986}
The time-dependent electrolyte friction on a single-charged spherical polyion is calculated analyti-
cally within the Debye-Huckel approximation for the bulk interionic correlations of the supporting
electrolyte. Its time integral is the generalization of Schurr s previous result for the electrolyte fric-
tion for arbitrary charges and mobilities of the ions of the supporting ionic solution (with an arbi-
trary number of species}. The relationship between these results and analogous results available in
the theory of ionic conductance and in the theory of self-diffusion of interacting polyions is estab-
hshed, and the finite-size effects of the electrolyte ions are calculated.

I. INTRODUCTION the electrolyte friction for an arbitrary number of ionic

In the recent years there has been a renewed interest in
'thc diffusive properties of charged iilacroparticlcs ili
suspension. ' In particular, their tracer diffusion proper-
ties have been measured to study the effect of the interac-
tion of the tracer with other macroparticles and with the
small ions of the supporting electrolyte solution. 3 In an
attempt to establish a theoretical interpretation of these
measurements, various existing theories have been em-
ployed. For example, the theoretical analysis of the fric-
tion on the tracer produced by its interaction with the
electrolyte has made use of the theories developed, at the
Debye-Hiickel level, by Booth~ and by Schurr. s Although
the latter ignores important hydrodynamic effects taken
into account by the former, it is not yet clear how impor-
tant such effects are in particular experimental situations.
In the meanwhile, additional theoretical work on the
properties of the electrolyte friction is still needed, which
generalizes the existing theories in other important direc-
tions (such as for regimes other'than the Debye-Hiickel
limit), and which places the existing theories in a more
general context. In recent work, Medina-Noyola and
Vizcarra-Rend6n developed a theory which explicitly
derives the Langevin equation for a single-chargtxl tracer
implicitly assumed in Schurr s work. As in this work, hy-
drodynamic interactions between the ions in the solution
were ignored by Medina-Noyola and Vizcarra-Rendon.
Their work provides, however, the framework in which
some of the mentioned generalizations can be realized.
The purpose of this paper is to report the specific con-
clusions that can be obtained from the formal expression
for the time-dependent friction k(t) derived in Ref. 6.
Analytic expressions are provided here for such a quantity
and its Laplace transform k(s) within the Debye-Huckel
approximation for the interionic correlation functions of
the supporting electrolyte. The zero-frequency limit of
k(s), p', provides a generalization of Schurr's results for
species, arbitrary charge, and mobility. For a pointhke
tracer we notice that this result does not exactly reproduce
Onsager's limiting law7 for ionic conductance, yielding
values for P' about 15% smaller. We also notice that in
this limit, k(t) has a close resemblance with an analogous
result for the Brownian one-component plasma derived by
Hess and IGein within a mode-mode coupling approxi-
mation. Finally, in order to assess the effects of a finite
size of the electrolyte ions, we compare the Debye-Hiickel
limit for k(t =0) and k(s =0) with the corresponding re-
sults obtained from the mean spherical approximation'
for the interionic correlations between the electrolyte ions
modeled by charged hard spheres. We find that the
Debye-Huckel limit clearly underestimates the electrolyte
effects at high ionic strengths when the polyion is not
very large compared with the electrolyte ions.
This paper is organized as follows. Starting from the
formal expression for k (t) derived in Ref. 6, in Sec. II we
express k(t) as an integral involving the bulk interionic
correlation functions9 between the small ions of the sup-
porting solution. These functions are modeled by those of
the restricted primitive model (a neutral mixture of
charged hard spheres of equal radius a, in a uniform
background of dielectric constant e), and calculated
within the mean spherical approximation (Sec. III). Be-
fore showing numerical results to illustrate the effect of
finite-size electrolyte ions, we establish in Sec. IV the
Debye-Huckel limit of k(t) (corresponding to the limit
a, ~0 in the mean spherical approximation), for which
explicit analytic results are obtained. The additional limit
in which the size of the tracer also vanishes is considered
in Sec. V, where it is compared with the result of Hess
and Klein for the self-friction in the Brownian one-
component plasma. The effect of a finite size of the small
ions on the time-dependent friction k (t) of a polyion are
illustrated in Sec. VI by the comparison of the initial
value and of the time integral of k(t), as obtained from

&986 The American Physical Society

34 TIME-DEPENDENT ELECTROLUX~ m FRICTION ON CHARGED. ..

the mean spherical approximation, with the corresponding The matrix products implicit in Eq. (5) involve sum over
Debye-Huckel limits. In Sce. VII we establish the rela- common discrete indices and unrestricted integration over
tionship between our results for g in which the tracer be- common continuous variables. In Eq. (4) e is the dielec-
comes identical to one of the electrolyte iona, with the tric constant of the solvent, q;, Do, and n; are the charge,
corresponding classical limiting law derived by Onsager free-diffusion constant, and bulk concentration of electro-
The most sahent results of this work are finally summa- lyte ions of species i (i =1,2, .. ., v), whose equilibrium ra-
rized in Sec. VIII. dial concentration around the polyion is n;~(r}. In Eq. (7)
et'J '(r, r') is the interionie direct correlation function,
II. SUMMARY OF PREVIOUS RESULTS which in the Deby+Hiickel approximation for pointlike
small iona (a, =0) is given by
In their paper Medina-Noyola and Vizcarra-Rendbns
derived a Langevin equation for a charged spherical c'J '(r, r') = &/(ektt T r —
qq— r ~ ~
) .
Brownian particle diffusing in an ionic solution. Such an
equation reads To proceed further, we now approximate Eij(r, r') by its
bulk value, i.e., we write n, +(r) = n; and
dV(t) P V( et'J '(r, r')=ei( r — r' ) in Eq. (7), so that Fi(r, r', t) can
be written as a function of r r' —
~ ~

such that
~ ~

F"lj (r r't)
t
= 1 e'"" "F"
IJ (k t t)d k
(2 )3
in which V(t) is the velocity of the polyion of mass M,
charge Q, and hard-sphere radius a. Without the last two which defines its Fourier transform Fi(k, t). Then, Eq.
terins on its right-hand-side (rhs), this is the ordinary (4) can be written as
Langevin equation for a neutral Brownian particle, i.e., P
is the Stokes friction coefficient, P=&rtia, and f (t) is
the corresponding fluctuating force, i.e., white noise, with 3 kgT(a')2 0 ki
correlation function given by
(10)
2ktt Tg
(f (t)ftt(t')) = 5~(t t') —
(tz, P=1,2, 3), (2)
z with a'=a +a, . More specific results for k (t) require ad-
ditional approximations for c;J(r), which will now be in-
where T is the temperature and ks is Boltzmann's con- troduced.
stant. The last timn in Eq. (1) is a fluctuating force ori-
ginating in the spontaneous deviations from the equilibri-
um spherical symmetry of the electrical double layer sur- III. THE MEAN SPHERICAL
rounding the polyion. P(t) is a Gaussian but colored APPROXIMATION FOR k (t)
noise, with correlation function given by the following
fluctuation dissipation relation:
The function FJ(k, t) can now be calculated using the
(F (t)Ftt(t') ) =, ksT
5,ttk( i
t t' i
) =1,2, 3)
(tt, P— . (3)
mean spherical approximation for the bulk direct correla-
tion functions c J(r) of the restricted primitive model of
the supporting ionic solution, which results in the follow-
6
According to Medina-Noyola and Vizcarra;Rendon,
ing expressions for c,i(r):
the time-dependent electrolyte friction k (t) is given by
2
e&(r)=e (r)+q;qJee(r) .
k(t)= 2 f d r f 13r'gq;qJ.
j l,
In this equation c (r) is the direct correlation function of
XFJ(r, r';t), a fluid of hard spheres of radius a, and number concen-
tration n =g,"n;, and ee(r) is given by
where the volume of integration excludes the region inac-
cessible to the electrolyte ions, r & (a+a, ), with a, be- (eksTr} ', r y—
2a,
~ ~
ce(r) = (12)
ing the radius of the small iona. [Equation (4) is Eq. (71) &Ta,—
(B Br/4a, )(ek— ) ', r &2a,
of Ref. 6 after using Eqs. (64) and (72) of the same refer-
ence, with L,J ~D n, 5,&.] In Eq. (4), FJ(r, r';t) is given by a&here
—
F~J(r, r';t) = [exp( —G2z t )E2z']; J(r, r')
( ~
(5)
B =(x+1 —v'I+2»)/»
with the matrices G22 and E22 defined as with »2=(2a, a) —
= (2a, ) 4m g n;q; /eks T. (13)
—
( G22); 1 (r, r') = Dn; V~EJ(r, r'), — (6)
Using this approximation for c,z(r), along with the elec-
(E22); J(r, r') =E J(r, r')
—
troneutrality condition, g," in;q;=0, Eq (10) can be
=5,J5(r r')/n;~(r) , r') .
eJ'(r— simplifled further to read
3448 H. RUIZ-ESTRADA et al. 34

kMsA(t) 2Q + to the Debye-Huckel hmit, in which analytic results can

3ire(a') be obtained.
(ka') —Dokl[1 —ac s(k&js
x f dk
k
Sin
[1—ace(k)]
8 IV. THE DEBYE-HUCKEL LIMIT OF k(t)

where c «(k) is the Fourier transform of ce(r) and The Debye-Hiickel approximation in Eq. (8) corre-
sponds to the limit a, ~0 of Eq. (12), which leads to
a:—g nial 2
ac e(k) = —st2/k2 . (16)
Unfortunately, the integral in Eq. (14) can only be This expression, when used in Eq. (14}, leads to the fol-
evaluated numerically, and results for k s"(t =0) will be lowing analytic result for the time-dependent friction
reported below. For the moment, let us specialize further k(t):

knH(t) = (I1 —4[(ir'D't)'/'] J

8
——,' I2cosh( 2sta) e—'4[(st D t)' a/(D —
t)' ) e~'4[—
(x D t)'/ +a/(D t)'/ ])I) (17)

where 4(x) is the probability integral. ' Also, the Laplace transform of this function can be evaluated, with the follow-
ing result:
— —
k DH(&)
f 0
e sskDH(t)dt= Q
6&a 2S
[(1 e ~s) (1 e Ma(1+s/BD )
)/(1+s/lt2DO)1/2] (18)

k (t) is a monotonically decreasing function of t, identical if (D0}3/2 in the denominator of Eq. (20) were
whose initial value is given by substituted by (2D ) . This observation deserves addi-
tional comments.
M 2@
+ — The Brownian one-component plasma can be thought
kDH(t 0) Q (1 e 2aa) (19)
668 of as the particular case of the system considered here in
which only one species of small ions are present, and the
At long times, the asymptotic behavior of k (t) is tracer is in fact one of such small ions (which requires of
given by a rigid background to balance the total charge. ) If point-
— like ions are considered, this is the Brownian one-
~2 e Pao~
DH
(20) component plasma, and its self-friction can then be ob-
6&lre(D 0 ) 3/2 t 3/2 tained from Eq. (17}in the hmit a ~0, leading to
which defines the relaxation time kBocp( 4 IQ c
)
~„=(~2D0)-' 3
'
1/2
associated with the diffusive relaxation of the spherical X mstI 1 4[(a-Dt)' —] I e.
double layer of thickness ~ ' around the polyion. We
should point out that for low ionic strengths, say for (22)
'-1000 A, and with D -10
cm /sec, rli maybe as
where a =4rtPnQ /e and P= 1/k&T. Again, this result
long as 1 msec, which is by no means short compared
would be identical to that obtained by Hess and Klein' ex-
with the typical relaxation times of the local structure in-
duced by the interactions between the polyions. ' Under cept that in the expression in brackets in Eq. (22) Hess
and Klein obtain (2D ) instead of D . We could argue
those circumstances, however, the additional friction in-
duced by such interactions may become quantitatively that when the tracer has the same size as the small iona,
very important.
D should in fact be replaced by (2D ), which is the dif-
fusion coefficient that describes the Brownian motion of
V. RELATIONSHIP %'ITH THE SKI.F-FRICTION the relative displacement between the tracer and any of
the diffusing ions around it, in the absence of interactions.
IN THE BRO%'NIAN ONE-COMPONENT PLASMA
This is in fact the correct interpretation of D entering in
It is interesting to notice that a similar asymptotic Eq. (6), which originates in the diffusion equation of the
behavior to that found in Eq. (20) for k(t) has been de- small ions as described from the reference frame of the
rived for the time-dependent self-friction in a system of tracer. However, even this modification of Eq. (22) does
pointlike ions of a single species [the Brownian one- not coincide with the result of Bess and Klein, ' since now
component plasma (BOCP)] by Hess and Klein' from the we would have an extra factor of e '. The reason for
many-body Fokker-Planck equation using a mode-mode this discrepancy arises from the fact that the theory
coupling approximation. In fact the two results would be developed here is based on a description of the fiuctua-
 TIME-DEPENDENT ELECTROLUX IH FRIC'DON ON CHARGED. .. 3449

tions of the cloud around an essentially immobile tracer,

and the (nonlinear) couphng of these fluctuations with the MSA (Oa/a 0. 2)
DH
rapid Brownian motion of the tracer has hem neglected.
The mode-mode coupling approximation employed by
Hess and Klein' is a device to treat precisely this type of
problem with a higher degree of accuracy. The theory
considered here could be in principle be revised so as to
include these effects by an explicit consideration of the
position of the tracer as an additional variable in the non-
contracted description from which the Langevin equation
for the tracer was derived. 6 However, the results derived
here are interesting by themselves because they are suffi-
ciently accurate if the tracer is indeed larger than the
diffusing small ions, and this is the problem that concerns
us here. Furthermore, our results contain already a
variety of extensions that so.sn rather prohibitive with ap- IO

proaches similar to that of Hess and Klein. One such ex-

tension concerns the finite size of the small ions as intro-
duced in Sec. GI. The following section reports some nu-
merical results to illustrate these effects on k (t).
UI. FINITE-SIZE ELECTROLYTE IONS:
Bi&aCT ON k(~)
FIG. 1. Mean spherical approximation {MSA) and Debye-
Huckel limit (DH} {a,/a ~0) for the ionic strength-dependent
factors H{za;a, /a} and 6{~a;a,/a) defined in Eqs. (23} and
g5), respectively. The ratio a, /a was taken to be 0.2 in the
mean spherical approximation.

Let us now summarize the results of including a finite- Pl Q GMSA(&a. a /a ) (25)

size ion in the interionic correlations, as described by the 12eaa'
mean spherical approximation. This amounts to using where
Eqs. (12)— (15) to calculate k(t). In general, we expect
such small-ion effects to be negligible for very large po- . g(&a )'
G Ms~
s~(tta;a, /a)=
lyions at any reasonable salt concentration. We find this ir(1+a, /a )
to be the case when the ratio a /a, is larger than about 30
for salt concentrations such that 0&aa &10. Also, the
sin 1+Q, Q y
X
Debye-Hiickel limit should be obtained at very low salt 1—Qc~ g, KQ, Qg Q
concentrations for any ratio a/a, . An estimate of the
(26)

:
quantitative difference can best be expressed by compar-
in'g the initial value k(t =0) of k (t) and its time intergral,
In the limit a, /a ~0, we obtain
P~, for both cases.
The initial value of k(t), as follows from Eq. (14), can lim GMs+(tta;a, /a)—GD (xa)
a/a 0
be written as
= f 1 —(1+2m )e ~']/~a,
C)2
"(t =0) = &
(xa )H +(za, a, /a ), (23) (27)
36Q
which is Schurr's result. s In Fig. 1 we also compare this
where the concentration-dependent factor H can be read Debye-Huckel limiting expression with the corresponding
from Eq. (14). In the Debye-Huckel limit, a, /a~O, we result for the mean spherical approximation, Eq. (26), as a
have function of ~a, for the ratio (a, /a ) =0.2. As can be seen
from this comparison, there is little difference at very
lim H s (~a;a, /a)=—H H(xa)=(l —e ')l2. (24) small ionic strengths, the peak of P' at ma=i is only
a /a~0
slightly decreased as a result of the finite ratio a, /a, and
In Fig. 1 we compare HDH(~a) as a function of aa for the qualitative behavior around this peak is similar to the
0~~a &10 with H s (aa;a, /a) for the ratio a, /a=0. 2. Debye-Huckel result. More interesting is the fact that as
As can be seen from this figure, the deviations from the the ionic strength increases, the finite size of the small
Debye-Huckel approximation are substantial at all but the ions leads to an increase in P
with respect to its Debye-
smallest concentrations. Hiickel value, leading to a qualitatively different behavior
Although it is at t =0 where such deviation is larger, at high ionic strength. This is due to the fact that as aa
its cumulative effect can be appreciated from an analo- increases it is not only the coulombic interactions between
gous comparison of the results for the time integral of tracer and polyion which produces the friction, but also
k (t), i.e., for the static electrolyte friction P'= —k(0). The the hard-sphere interactions contribute to this effect.
mean spherical result for P' can be written, integrating However, since the prefactor Q /12eaD in Eq. (37) does
Eq. (14), as not depend on (a, /a ), the order of magnitude of P' is not
 H. RUIZ-ESTRADA et al.

changed by these finite-size effects. Another point which that small ion due to ion-ion interactions, yielding, for P,
is important to stress is the fact that P' is quite insensitive pointlike ions in the Debye-Hiickel limit, the following re-
(when plotted versus ~a) to the number of ionic species, sult
and to their dispersity in charge. Furthermore, eve should 2
Kqt
mention that the extension of Eq. (25) for the case in '
which the various ionic species may differ also in their 6'~(i}
free-diffusion constant only amounts to the replacement
where
of D in the RHS of Eq. (25) by

X "jqj' Dj' X "jq' g n;qjP)gj~ (Pj+Pj)

j=l j=1 Dq(i) =kg T {33)
where Dj' is the diffusion coefficient describing the X "jqj
j=l
Brownian motion of the relative distance between an ion
of species i and the tracer. This result cannot be obtained, Equation (32) is to be compared with Onsager's limiting
as Eq. (25), from the direct integration of Eq. (14) since expression for P', which reads
this equation already assumes D =Do How. ever, if we 2
go back to Eq. (6) and replace D by DJ' and integrate Eq. g'; (Onsager) = (1 —d ), (34)
(10) on t before the integration on k indicated there, the 3E'
end result is Eq. (25) with Do replaced by D~. Thus where
another important conclusion is that the ionic strength
y
dependence of Pis quite insensitive to possible differ-
g njqjP( (P+Pj)
ences between the mobilities of the various ionic species j=1
present in the supporting solution. (35)

VII. RELATIONSHIP %'ITH ONSAGER

LIMITING LA% From Eqs. (33) and (40} it is not difficult to see that
In principle, when the tracer is considered to be in fact ktt T
one of the ions of the electrolyte, P is the ion-ion contri- Dq(i) = (36)
bution to the friction on such a tracer, a quantity that has

tance. ""
been calculated in the theories of electrolyte conduc-
We should point out„however, that the
so that our result in Eq. (32) can also be written as

pitq;
r

1 —d;
theory from which the above results were derived neglects pg
(37)
at the outset the coupling betwo:n hydrodynamic and 3E' 2
electrolyte fluctuations, and hence, important aspects of
electrolyte conduction, such as the electrophoretic effect, Thus, we find that our Debye-Hiickel result for g' does
are absent from these results. Furthermore, as mentioned not exactly reproduce Onsager's expression in Eq. (34).
in Sec. V, when the tracer is about the same size as the This is due to the fact that the modification of our theory
involved in Eqs. (29)— (31) is not an exact way of taking
electrolyte ions, not only the fluctuations of the ionic at-
mosphere around an essentially immobile tracer should be
into account the fluctuations on the ionic cloud around
considered, but also a mare accurate description of the ef- the tracer induced by the Brawnian motion of the tracer
fect of the rapid randoin motion of the (small) tracer itself. A quantitative comparison between the results in
should be included. In this respect, we should only modi- Eqs. (34) and (37) is given by
fy our theory to the extent suggested in Sec. V above, i.e., i l2

we shaH approximate Dj' by g,"= P'(Onsager )

I.
2
D =D +D (29)
~g; =q and P=P for
When ~
i =1,2, . . . , v, the factor
where Dj and Dq are the free diffusion coefficients of the [(1+d; )/2]=0. 854, i.e., our Debye-Hiickel result un-
small ions of type j,
and of the tracer, respectively, as derestimates the electrolyte friction by about 15% with
measured from a fixed reference frame. These diffusion respect to Onsager's limiting lm.
coefficients, in their turn, can be approximated by
Einstein's relation
VIII. SUMMARY OF RESULTS
Dj kti T/gj~—
— (30)
In this work me have calculated the time-dependent
electrolyte friction k(t) on a polyion diffusing in an ionic
Dr ks T/P . — (31} solution, based on the results previously derived by
Medina-Noyola and Vizcarra-Rendan. Starting from
With this modification, and with the tracer being con- their formal expression for k {t) in terms of the equilibri-
sidered a small ion of species i, the results of the previous um interionic correlation functions, we first approximated
section can be used to calculate the additional friction on these quantities by their bulk value to render the computa-
 TIM~ DEPENDENT ELECrN. OLY m FRICTION ON CHARGED.
tions feasible. Additional approximations were then re-
quired which consisted in approximating the bulk equili-
brium correlation functions by those of the primitive
model of the supporting ionic solution within the mean
spherical closure (Sec. III). The limit of zero radius of the
small iona in this approximation constitutes the Debye-
Huckel limit, which allows for explicit analytic results for
k (t) and its Laplace transform k(s), as shown in Sec. IV.
In an attempt to establish erne contact with other related
results, we considered in Sec. V the limit of our Debye-
Huckel expression for k(t) in which the radius of the
tracer vanishes. We found that in this limit there is a
close rmmiblance between our results and those derived
by Hess and Klein' for the self-friction in the Brownian
one-component plasma. Returning to the analysis of the
properties of the time. dependent friction on large po-
lyions, we then considered the departures from the
Debye-Hiickel limit when the finite size of the electrolyte
iona is taken into account. We showed numerical results
for k(t =0) and for the time integral of k(t), P'. We
found that departures from the Debye-Huckel hmit were
negligible when the ratio a/a, excedes about 30. The
most important point to stress, however, is that these
differences concern only the ionic strength deliendence of
P', since we found that finite-size effects do not alter the
deINndence of p' on polyion charge, size, and solvent
dielectric constant already exhibited in the Debye-Huckel
limit for P originally derived by Schurr. Similar con-
clusions were arrived at concerning the effect of possible
~%. Hess and R. Klein, Adv. Phys. 32, 173 (1983).
2%. D. Dorier, H. M. Lindsay, and P. M. Chaikin, J. Phys.
(Paris) Colloq, 46, C3-165 (1985).
38. Gorti, L. Planck, and 8. R. %'afe, J. Chem. Phys. Ik, 909
(1984).
sF. Booth, J. Chem. Phys. 22, 1956 (1954).
5J. M. Sch&rr, Chem. Phys. 45, 119 (1980).
~M. Medina-Noyola and A. Vixcarra-aeod6n, Phys. Rev. A 32,
3596 {1985).
..
dispersity in the charge and free-diffusion coefficient of
the various ionic species in the supporting solution.
As a final observation, we compared the limit of our
Debye-Hiickel result for the electrolyte friction in which
the tracer is in reality one of the small iona, with the cor-
responding limiting law derived by Onsager. We found
that our theory does not exactly reproduce such a limiting
law, although the disagreement was not quantitatively
severe (in a simple case our results are 15% smaller than
Onsager's). The inconsistency of our limiting results with
Onsager's law and with Hess and Klein's results for the
time-dependent self-friction in the Brownian one-
component plasma were traced back to the inexact treat-
ment in our theory of the fiuctuations induced on the ion-
ic atmosphere of the tracer by its own rapid Brownian
motion. Of course„ the relevance of these final observa-
tions is smaller the larger the tracer is, compared with the
other diffusing iona, and it was for this regime that our
theory was proposed. Nevertheless, these observations are
interesting from a theoretical point of view.
ACKN0% LEDGMENTS
This work was partially supported by Consejo Nacional
de Ciencia y Tecnologia {CONACyT) (Mexico), by Con-
sejo del Sistema Nacional de la Educacion Tecnologica,
Secretaria de Educacion Publica {COSNET-SEP) {Mexi-
co), and by Bundesministerium fiir Forschung und Tech-
nologie {BMI' I ) (Federal Republic of Germany).
7I . Onsager, Ann. N. Y. Acad. Sci. 46, 241 (1945).
SE. %aisman and J. L Lebovritz, J. Chem. Phys. 56, 3086
(1972).
sD. A. Mcguarrie, Statistical Mechanics (Harper and Row, New
York, 1975).
iol. S. Gradshteyn and I. M. Ryzhik, Table of Integrals Series
and Redacts (Academic, New York, 1985).
S. Harris, Mol. Phys. 26, 953 (1973)
~&H. L. Fricdman, Physica 30, 537 (1964).