Nonlocal statistical field theory of dipolar particles in electrolyte solutions

Yury Budkov1, 2, a)
1)
School of Applied Mathematics, Tikhonov Moscow Institute of Electronics
and Mathematics, National Research University Higher School of Economics,
Tallinskaya st. 34, 123458 Moscow, Russia
arXiv:1903.10054v1 [cond-mat.soft] 24 Mar 2019

2)
G.A. Krestov Institute of Solution Chemistry of the Russian Academy of Sciences,
Akademicheskaya st. 1, 153045 Ivanovo, Russia

1
 We present a nonlocal statistical field theory of a dilute electrolyte solution with small
additive of dipolar particles. We postulate that every dipolar particle is associated
with an arbitrary probability distribution function (PDF) of distance between its
charge centers. Using the standard Hubbard-Stratonovich transformation, we repre-
sent the configuration integral of the system in the functional integral form. We show
that in the limit of a small permanent dipole moment, the functional in integrand
exponent takes the well known form of the Poisson-Boltzmann-Langevin (PBL) func-
tional. In the mean-field approximation we obtain a non-linear integro-differential
equation with respect to the mean-field electrostatic potential, generalizing the PBL
equation for the point-like dipoles obtained first by Abrashkin et al. We apply the
obtained equation in its linearized form to derivation of the expressions for the mean-
field electrostatic potential of the point-like test ion and its solvation free energy in
salt-free solution, as well as in solution with salt ions. For the ’Yukawa’-type PDF
we obtain analytic relations for both the electrostatic potential and the solvation free
energy of the point-like test ion. We obtain a general expression for the bulk elec-
trostatic free energy of the solution within the Random phase approximation (RPA).
For the salt-free solution of the dipolar particles for the Yukawa-type PDF we obtain
an analytic relation for the electrostatic free energy, resulting in two limiting regimes.
Namely, when the Debye screening length attributed to the charged groups of the
dipolar particles is much bigger than the mean-square dipole length, the electrostatic
correlations of the dipolar particles consist in their effective Keesom attraction, so
that the electrostatic free energy is proportional to the square of dipolar particles
concentration. In the opposite case, when the mean-square dipole length is much
bigger than the Debye length, the charge centers of dipolar particles can be consid-
ered as unbonded ions, so that the electrostatic free energy can be described by the
Debye-Hueckel limiting law. Finally, we analyze the limiting laws, following from the
general relation for the electrostatic free energy of solution in presence of both the
ions and the dipolar particles for the case of Yukawa-type PDF.

a)
[email protected]

2
I. INTRODUCTION

Development of the polar fluids statistical theory has a long history starting from the
classical works of L. Onsager1 and J. Kirkwood2 . So far, some progress has been achieved in
statistical mechanics of dipolar fluids allowing us to calculate statistical dielectric permittiv-
ity of molecular fluids, such as water, aliphatic alcohols, etc1–5 with the accuracy appropriate
for experimentalists. There are also a small number of theoretical approaches allowing one
to calculate self-consistently the electrostatic free energy of a polar fluid alongside the di-
electric permittivity. Levin6 formulated a theoretical model that allowed him to obtain an
analytic expression for the free energy of dipolar hard spheres fluid. Using the idea of Debye
charging process7 , he calculated a contribution of a dipole-dipole electrostatic interaction
to the total free energy of the dipolar hard spheres based on the Onsager’s reaction field
theory1 . It should be noted that this approach was earlier applied to the evaluation of the
electrostatic free energy for the dipolar hard spheres in work8 . It is also important to men-
tion the approach, based on the linearized Poisson-Boltzmann (PB) equation application for
the mixture of dipoles with ionic components9 . Within that approach, using the linearized
PB equation with undefined constant dielectric permittivity, the electrostatic potentials of
the point-like charge and point-like dipole were calculated. Further, based on the Debye-
Hueckel approximation, the energies of the electrostatic interaction between the point ion
and surrounding ions and dipoles were calculated. The energies of electrostatic interaction
between the point dipole and the surrounding dipoles and ions were obtained in the same
manner. Further, equating the total interaction energies "ion-dipole" and "dipole-ion", the
authors obtained an equation for dielectric permittivity as a function of dipole moment,
temperature, dipolar particles concentration and ionic strength, that is a generalization of
the classic Onsager’s equation1 . The derived equation allowed the authors to calculate the
electrostatic contribution to the total free energy and analyse the phase behavior of the
electrolyte solution accounting for the formation of ion pairs modelled by dipoles.

There are also a few works, which investigate the mixture of dipolar particles with ions of
salt in the confinement or near the charged electrode within the field-theoretical approach.
In work10 Abrashkin et al. formulated a statistical Poisson-Boltzmann-Langevin (PBL) field
theory taking into account explicitly the molecules of polar fluid. In the mean-field approxi-
mation the authors derived a modified Poisson-Boltzmann (PB) equation taking into account

3
the dipolar particles, making the local dielectric permittivity of solution dependent on the
local electric field. Applying the obtained equation to the description of the electrolyte solu-
tion confined between the oppositely charged flat infinite walls, the authors showed that the
osmotic pressure calculated within their PBL theory deviated significantly from the value
that was obtained in the classical PB theory with a continuous dielectric solvent for rather
small distances (a few nanometers) between the charged walls. It should be noted that a sim-
ilar statistical field theory was earlier obtained by Coalson et al.11 generalizing the classical
PB theory and taking into account explicitly the particles with arbitrary electric multipole
moments. As an application of their theory, the authors made a numerical calculation of the
profiles of the solution local dielectric permittivity between the two flat oppositely charged
walls bounding the mixture of ions and dipolar particles, as well as the surface free energy
as a function of the distance between the walls. Budkov et al.12 generalized the PBL theory,
deriving a modified PB equation with an explicit account of the polarizability tensor of the
solvent molecules in addition to their permanent dipole moment within a field-theoretical
approach. The authors used their modified PB equation to generalize the classical Gouy-
Chapman theory of the electric double layer in the electrolyte solutions for the case when
there is a small amount of polarizable/polar co-solvent molecules in the solution. They in-
vestigated the influence of the permanent dipole moment and polarizability of the co-solvent
molecules on the differential capacitance of the electric double layer. It was shown that at a
sufficiently high surface potential of the electrode, presence of polar or polarizable co-solvent
molecules resulted in an increase in the differential capacitance relative to the capacitance
in solution without co-solvent impurities, though the polarizability effect appeared to be
stronger.

Nevertheless, all the feld-theoretical models of dipolar fluids discussed above treat dipolar
particles as point-like ones, i.e. their contribution to the total potential energy of electrostatic
interactions is taken into account as a term in the total microscopic charge density described
by charge density of point-like dipoles. Such a local dipolar particles model does not allow
us to find out how the finite size of the dipolar particles affects the thermodynamic and
structural properties of the polar fluids. For instance, the electrostatic potential of the point-
like test charge surrounded by point-like dipolar particles in the linear response theory is
simply a vacuum Coulomb potential divided by constant dielectric permittivity5 . Thereby,
the local theory cannot answer the question: How does the electrostatic potential of point-

4
like test charge depend on the distance at at a scale order of the several dipole lengths? In
other words, it is interesting to understand how the electrostatic potential dependence on
the distance from the point-like charge deviates from the Coulomb law at the microscopic
scales, where the picture of continuous dielectric media becomes invalid.
A very good phenomenological theory named as nonlocal electrostatics was formulated
by Kornyshev and coauthors13 . The central idea of the nonlocal electrostatics approach is
the replacement of constant dielectric permeability (or related to it dielectric permittivity)
of medium by the nonlocal dielectric permeability (permittivity) operator and rewriting the
well known electrostatics equations of the continuous media in its terms. Thus, instead of
the well known local Poisson differential equation for the electrostatic potential, the authors
obtained a nonlocal integral equation, allowing, in principle (if the functional form of the
dielectric permittivity operator is known), one to calculate the electrostatic potential at the
microscopic scale. Despite the fact that nonlocal electrostatics allowed the researchers to
calculate the electrostatic potential near the charged surfaces of the metals or dielectrics
immersed in the electrolyte solutions and the solvation free energy of particles with different
electric structures13 , it remains unclear how one can calculate within this approach the
macroscopic thermodynamic functions of the electrolyte solutions with an explicit account
of the dipolar particles.
On the other hand, the microscopic theory of the point-like dipolar particles taking into
account the electrostatic correlations at the many-body level (within the Random phase
approximation) leads to the ultraviolet divergences of the electrostatic free energy. The lat-
ter also indicates the necessity to introduce internal electric structure details of the dipolar
particles into the microscopic theory. Up to date, two methods have been proposed for
obtaining a finite value of the electrostatic free energy of the dipolar fluid within the micro-
scopic statistical theories. The first approach consists in introducing an ultraviolet cut-off
into the integration over the vectors of reciprocal space, so that the cut-off parameter is usu-
ally inversely proportional to the molecular length scale (see, for instance14–17 ). The second
approach starts from the assumption that the charge groups of the dipolar particles are not
point-like ones described by the delta-functional densities, but are smeared over some spatial
region with some form-factor. Thus, introducing the charge form-factor, containing some
phenomenological scale parameter to the microscopic theory allows the authors to avoid the
ultraviolet divergence18 . Note that both approaches give qualitatively similar results for

5
the behavior of electrostatic free energy at a small dipolar particles concentration. Strictly
speaking, the first approach is a particular case of the second one, when the Fourier-image of
the form-factor choice is the Heaviside step function, which, in turn, leads to the ultraviolet
cut-off. We would also like to note that despite the efficiency of these approaches in the-
oretical description of the thermodynamic properties of different dipolar fluids, they could
be considered only as mathematical tricks allowing one to overcome the deficiencies of the
local theory without accounting for the internal electric structure of the dipolar particles.
However, within both approaches the question stated above about the electrostatic potential
behavior at small distances from the test point-like charge remains unanswered.
On the other hand, the recent advances in experimental studies of different organic com-
pounds (such as proteins19,20 , betaines21,22 , zwitterionic liquids23 , etc.) comprising rather
long polar groups (∼ 5 − 10 nm) challenged the statistical mechanics to develop a field-
theoretical approach to modelling the dipolar particles not as the point-like dipoles10,12 (or
even as the dipolar hard spheres24,25 ), but as a set of charge centers with fixed or fluctuat-
ing distances between them. One would expect that such a nonlocal statistical field theory
should be devoid of the above described ultraviolet divergence of the free energy and allow
us to study the electrostatic potential behavior at the distances from the point-like test
charge that are comparable to the characteristic distances between the charge centers of the
dipolar particle. In the present paper, we formulate such a nonlocal statistical field theory
of the ion-dipole mixture.

II. THEORY

A. Configuration integral of ion-dipole mixture in functional integral form

Let us consider an electrolyte solution with N+ point-like cations with charge q+ , N−

point-like anions with charge q− , and Nd dipolar particles with point-like charged groups
with charges ±e (e is the elementary charge) confined in the volume V . Let the numbers
of ions N± satisfy the global electric neutrality condition q+ N+ + q− N− = 0. Let us also
assume that the particles are immersed in a solvent which we will model as a uniform
dielectric background with constant dielectric permittivity ε. We postulate that each dipolar
particle is associated with the probability distribution function (PDF) g(r) of the distance

6
between its centers of charge. Keeping in mind these model assumptions, we start from the
configuration integral of the mixture
Z Z
Q = dΓd dΓs exp [−βHel ] , (1)

where
Nd (+) (−)
drj drj
Z Z Y
(+) (−)
dΓd (·) = g(rj − rj )(·) (2)
j=1
V
(±)
is the measure of integration over the coordinates rj of the charge centers of the dipolar
particles with the above mentioned PDF g(r), satisfying the normalization condition
Z
g(r)dr = 1, (3)

and
N+ N−
1
Z Z Y Z Y
(+) (−)
dΓs (·) = drl drk (·) (4)
V N+ +N− l=1 k=1

is the integration measure over the ionic coordinates; β = 1/kB T is the inverse thermal
energy. The Hamiltonian of the electrostatic interaction can be written as follows
1 1
Z Z
Hel = dr dr′ ρ̂(r)G0 (r − r′ )ρ̂(r′ ) = (ρ̂G0 ρ̂) , (5)
2 2
where G0 (r − r′ ) = 1/(4πεε0|r − r′ |) is the Green function of the Poisson equation and
Nd N+ N−
(+) (−) (+) (−)
X X X
ρ̂(r) = e (δ(r − rj ) − δ(r − rj )) + q+ δ(r − rk ) + q− δ(r − rl ) + ρext (r) (6)
j=1 k=1 l=1

is the total charge density of the system, consisting of the microscopic charge density of the
dipolar particles and ions (first, second, and third terms) and the fixed density of external
charges (fourth term). It should be noted that we omitted the electrostatic self-energy of
the charged particles in (5) for simplicity. Note that in the present study for simplicity
we neglected the contribution of the repulsive interactions between the species. It can be
justified for rather small concentrations of salt and dipolar particles.
Thus, we can rewrite our partition function as follows
 
β
Z Z
Q = dΓd dΓs exp − (ρ̂G0 ρ̂) . (7)
2
Further, using the Hubbard-Stratonovich transformation (see, for instance,26,28 )
  Z  
β Dϕ β −1
exp (ρ̂G0 ρ̂) = exp − (ϕG0 ϕ) + iβ(ρ̂ϕ) , (8)
2 C 2

7
we achieve the following functional representation of the configuration integral
 
Dϕ β
Z
N+ N−
Q= exp − (ϕG0 ϕ) + iβ(ρext ϕ) QN
−1
d [ϕ]Q+ [ϕ]Q− [ϕ]
d
(9)
C 2

with the one-particle partition functions

dr(+) dr(−) (+)

Z
Qd [ϕ] = g(r − r(−) ) exp(iβe(ϕ(r(+) ) − ϕ(r(−) )), (10)
V
dr
Z
Q± [ϕ] = exp(iβq± ϕ(r)), (11)
V
and the following short-hand notations
Z Z Z
−1 ′ −1 ′ ′
(ϕG0 ϕ) = dr dr ϕ(r)G0 (r, r )ϕ(r ), (ρ̂ϕ) = drρ̂(r)ϕ(r),
 
β
Z
−1
C= Dϕ exp − (ϕG0 ϕ) .
2
In the thermodynamic limit

V → ∞, Nd,± → ∞, Nd,± /V → nd,±

we have29
 Nd
1
Z Z
QN (+) (−) (+) (−) (+) (−)


d [ϕ] = 1+ dr dr g(r − r )(exp iβe(ϕ(r ) − ϕ(r )) − 1) ≃
d

V
 Z Z 
(+) (−) (+) (−) (+) (−)


exp nd dr dr g(r − r )(exp iβe(ϕ(r ) − ϕ(r )) − 1) . (12)

Similarly, we obtain
 Z 
N
Q±± [ϕ] ≃ exp n± dr(exp(iβq± ϕ(r)) − 1) . (13)

Therefore, we arrive at the following functional representation of configuration integral

Dϕ
Z
Q= exp [−S[ϕ]] , (14)
C

where the following functional

β
Z Z
S[ϕ] = (ϕG0 ϕ) − iβ(ρext ϕ) − nd dr dr′ g(r − r′ )(exp [iβe(ϕ(r) − ϕ(r′ ))] − 1)
−1
2
Z Z
− n+ dr(exp [iβq+ ϕ(r)] − 1) − n− dr(exp [iβq− ϕ(r)] − 1) (15)

is introduced.

8
 Now let us relate the present theory to the above-mentioned Poisson-Boltzmann-Langevin
(PBL) theory formulated within the Grand canonical ensemble in references10,11 and within
the Canonical ensemble in reference12 . Determining the PDF for the dipolar particles with
a fixed distance between the charge centers |r − r′ | = l in the following form

dn
Z
′
g(r − r ) = δ(r − r′ − l), (16)
4π
R
where integration dn/4π(..) means the averaging over orientations of unit vector n = l/l,
we obtain the following functional

β dn
Z Z
S[ϕ] = (ϕG−1
0 ϕ) − nd dr(exp (iβe(ϕ(r) − ϕ(r − l))) − 1)
2 4π
Z Z
− n+ dr(exp [iβq+ ϕ(r)] − 1) − n− dr(exp [iβq− ϕ(r)] − 1) − iβ(ρext ϕ) (17)

When l → 0, the expansion of the exponential factor into the Taylor series up to the
first order in l and averaging over orientations of vector l lead to the well known PBL
functional10–12
 
β sin βp|∇ϕ(r)|
Z
SP BL [ϕ] = (ϕG0−1 ϕ) − nd dr −1
2 βp|∇ϕ(r)|
Z Z
− n+ dr(exp [iβq+ ϕ(r)] − 1) − n− dr(exp [iβq− ϕ(r)] − 1) − iβ(ρext ϕ), (18)

where p = el is the dipole moment.

B. Mean-field approximation

Now let us consider the mean-field approximation for the configuration integral12,26,27 . At
first, we have to equate to zero the functional derivative of the functional (15), i.e.

δS[ϕ]
= 0, (19)
δϕ(r)

that yields the following equation

e(ϕ(r) − ϕ(r′ ))
Z Z
′
dr G−1
0 (r
′ ′
− r )ϕ(r ) − iρext (r) + 2nd e dr′ g(r − r′ ) sin
kB T

− i (q+ n+ exp [iβq+ ϕ(r)] + q− n− exp [iβq− ϕ(r)]) = 0. (20)

9
Further, introducing the electrostatic potential ϕ(r) = iψ(r)12,26,27 , we arrive at the following
nonlinear integro-differential self-consistent field equation:
e(ψ(r′ ) − ψ(r))
Z
− εε0 ∆ψ(r) = 2nd e dr′ g(r − r′ ) sinh
kB T
+ q+ n+ exp [−βq+ ψ(r)] + q− n− exp [−βq− ψ(r)] + ρext (r). (21)

The electrostatic free energy in the mean-field approximation is

(M F )
Fel = kB T S[iψ]. (22)

C. Point-like charge in dipole environment

As an application of equation (21), let us calculate the electrostatic potential of the

electric field created by a point-like test ion with charge q surrounded by dipolar particles
in the absence of ions (i.e., n± = 0). Placing the point-like ion at the origin and taking into
account that ρext (r) = qδ(r), we write the self-consistent field equation in the form
e(ψ(r′ ) − ψ(r))
Z
′ ′
− εε0 ∆ψ(r) = 2nd e dr g(r − r ) sinh + qδ(r). (23)
kB T
In order to understand how the dipolar surrounding changes the Coulomb electrostatic po-
tential of the bare point-like charge, we consider the regime of weak electrostatic interaction,
i.e. assume that eψ(r)/kB T ≪ 1, which yields
2nd e2 2nd e2
Z
− εε0 ∆ψ(r) = − ψ(r) + dr′g(r − r′ )ψ(r′ ) + qδ(r). (24)
kB T kB T
Further, passing to the Fourier-image of the electrostatic potential
Z
ψ̃(k) = drψ(r)e−ikr ,

we can rewrite the equation (24) in the following algebraic form

2nd e2
εε0 k 2 ψ̃(k) = − (1 − g(k))ψ̃(k) + q, (25)
kB T
where k = |k|. Thus, after some algebraic transformations, we arrive at
q
ψ̃(k) = , (26)
εε0 (k 2 + κ 2 (k))
where
2nd e2
κ 2 (k) = (1 − g(k)) (27)
εε0 kB T

10
is the so-called screening function and
Z
g(k) = dre−ikr g(r) (28)

is the characteristic function of distribution30 . Therefore, the electrostatic potential of the

point charge takes the standard form of the linear response theory31–33 :
q dk eikr
Z
ψ(r) = . (29)
εε0 (2π)3 k 2 + κ 2 (k)
In order to evaluate the electrostatic potential, we should specify the probability distribution
function g(r). For the dipolar particles with permanent dipole moments with the PDF
determined by relation (16) the characteristic function takes the form
sin kl
g(k) = . (30)
kl
However, relation (30) does not allow us to obtain the analytical expression for electrostatic
potential. Nevertheless, we can use another model characteristic function
1
g(k) = 2 2 , (31)
1 + k 6l
which allows us to obtain the analytic results, determining the following ’Yukawa’-type PDF
√ !
3 6r
g(r) = 2
exp − , (32)
2πl r l

where r = |r|. We would like to stress that the distributions determined by characteristic
functions (30) and (31) give the same mean-square distances between the charge centers of
√
dipolar particles l/ 6.
Thus, substituting eq. (31) with eq. (27) and taking integral (29), we arrive at the
following result  
r
q 1 + yd exp − ls
ψ(r) = , (33)
4πεε0r 1 + yd
where yd = nd e2 l2 /(3εε0kB T ) = l2 /6rD
2
is the electrostatic ”strength” of dipole-dipole elec-
p
trostatic interaction and ls = l/ 6ε(1 + yd ) is the length, whose physical sense will be
−1/2
clarified below; rD = (2nd e2 /εε0kB T ) is the effective Debye screening length attributed
to the charge centers of dipolar particles. It is interesting to investigate the electrostatic
potential behavior for very long dipolar particles, i.e when yd ≫ 1 (or l ≫ rD ). In that
regime, relation (33) takes the form of the well known Debye-Hueckel potential
 
q r
ψ(r) ≃ exp − . (34)
4πεε0r rD

11
Thus, when the mean-square dipole length is much bigger than the Debye screening length,
the charge centers of dipolar particles can be considered as unbonded ions, screening the
charge of the test ion.
It is worth noting that relation (33) allows us to formally introduce the local dielectric
permittivity as
ε(1 + yd )
ε(r) =  . (35)
1 + yd exp − lrs
Therefore, length ls determines the radius of the sphere around the point charge in which
the dielectric constant is smaller than its bulk value10,12,34,35

nd e2 l2
εb = ε(1 + yd ) = ε + . (36)
3ε0kB T

In other words, length ls can be considered as the effective solvation radius of the point
charge in the dipole environment within the linear theory.
The mean-field electrostatic free energy in the linear approximation is:

1
Z Z
(M F )
Fel = kB T S[iψ] = − dr dr′ ψ(r)G−1 ′ ′
0 (r, r )ψ(r )
2

nd e2
Z Z Z
′ ′ ′ 2
− dr dr g(r − r )(ψ(r) − ψ(r )) + drρext (r)ψ(r) =
2kB T

1
Z Z Z
′
− dr dr ψ(r)G−1 ′ ′
d (r, r )ψ(r ) + drρext (r)ψ(r), (37)
2
where the reciprocal Green function renormalized by the presence of dipolar particles

2nd e2
G−1 ′ −1 ′
d (r, r ) = G0 (r, r ) + (δ(r − r′ ) − g(r − r′ )) (38)
kB T

is introduced. Further, taking into account that

Z
ρext (r) = dr′ G−1 ′ ′
d (r, r )ψ(r ),

we obtain the final relation:

1
Z
Fel = drρext (r)ψ(r). (39)
2
To calculate the solvation free energy of the point-like test ion, we should subtract the
electrostatic self-energy of point-like charge, i.e.

1
Z
∆Fsolv = drρext (r)(ψ(r) − ψext (r)) (40)
2

12
from (39). Taking into account that
q
ψext (r) = , (41)
4πεε0r
we arrive at √
q2 6 y
∆Fsolv =− √ d . (42)
8πεε0 l 1 + yd

D. Point-like charge in ion-dipole environment

Now let us calculate the electrostatic potential of the point-like charge at the non-zero
ionic concentrations in the bulk (n± 6= 0) in the linear approximation eψ/kB T ≪ 1. In that
case the self-consistent field equation takes the following linearized form
2Ie2 2nd e2 2nd e2
Z
− εε0 ∆ψ(r) = − ψ(r) − ψ(r) + dr′ g(r − r′ )ψ(r′ ) + qδ(r), (43)
kB T kB T kB T
2 2
n− /2e2 is the solution ionic strength.


where I = q+ n+ + q−
Moving on to the Fourier representation, after some algebra, we obtain the standard
relation for the Fourier-image of the potential
q
ψ̃(k) = (44)
εε0 (k 2 + κ 2 (k))
with the following screening function
2nd e2
κ 2 (k) = κ2s + (1 − g(k)) , (45)
εε0 kB T
p
where κs = 2e2 I/(εε0kB T ) is the inverse Debye screening length attributed to the ionic
species.
Using the same Yukawa-type PDF, i.e., using the same characteristic function g(k) =
1/ (1 + k 2 l2 /6) as in the previous subsection and performing the inverse Fourier transforma-
tion, we arrive at the following analytic relation for the electrostatic potential
q
u(yd, ys )e−κ1 (yd ,ys )r + (1 − u(yd, ys )) e−κ2 (yd ,ys )r ,


ψ(r) = (46)
4πεε0 r
where
√ 
3 p 1/2 κ2 l2
κ1,2 (yd , ys ) = 1 + ys + yd ± (1 + ys + yd )2 − 4ys , ys = s , (47)
l 6
p
ys + yd + (1 + ys + yd )2 − 4ys − 1
u(yd, ys ) = p . (48)
2 (1 + ys + yd )2 − 4ys

13
In the absence of ions (ys = 0), expression (46) will transform into the above expression
(33). In the absence of dipolar particles (yd = 0), we obtain the standard Debye-Hueckel
potential
q
ψ(r) = e−κs r . (49)
4πεε0r
In order to obtain the solvation free energy of the test charge in the ion-dipole envi-
ronment, we execute the similar algebraic transformations as in the previous subsection.
In that case, we have the same relation for the electrostatic free energy in the mean-field
approximation, i.e.
1 1
Z Z Z Z
(M F ) ′ −1 ′ ′
Fel =− dr dr ψ(r)G (r, r )ψ(r ) + drρext (r)ψ(r) = drρext (r)ψ(r), (50)
2 2
where
2Ie2 2nd e2
G−1 (r, r′) = G−1 ′
0 (r, r ) + δ(r − r′ ) + (δ(r − r′ ) − g(r − r′ )) (51)
kB T kB T
is the reciprocal Green function renormalized by the presence of ions and dipolar particles.
Subtracting the electrostatic self-energy from (50) (see the previous subsection), we obtain
the following analytic expression
q2
∆Fsolv = − (u(yd, ys )(κ1 (yd , ys ) − κ2 (yd , ys )) + κ2 (yd , ys )) , (52)
8πεε0
which will turn into the above relation (42) in the ion-free case (ys = 0).

E. Random phase approximation

Now we proceed to the derivation of the electrostatic free energy of the ion-dipole mixture
within the Random phase approximation (RPA). Expanding the functional S[ϕ] in (14) into
a power series near the mean-field ϕ(r) = iψ(r) and truncating it at the quadratic term, we
obtain
β
ϕG −1 ϕ ,


S[ϕ] ≈ S[iψ] + (53)
2
where
δ 2 S[iψ]
G −1 (r, r′ |ψ) = kB T
δϕ(r)δϕ(r′)
1 2 2
= G−1 ′
n− exp (−βq− ψ(r)) δ(r − r′ )


0 (r, r ) + q+ n+ exp (−βq+ ψ(r)) + q−
kB T
2 Z
e (ψ(r) − ψ(r′′ ))
 
2nd e ′′ ′′
+ dr g(r − r ) cosh (δ(r − r′ ) − δ(r′′ − r′ )) (54)
kB T kB T

14
is the renormalized reciprocal Green function with the mean-field electrostatic potential ψ(r)
satisfying equation (21). Therefore, taking the Gaussian functional integral26 , we obtain the
following general relation for the configuration integral in the RPA
 
1
Q ≈ exp −S[iψ] + tr (ln G − ln G0 ) , (55)
2
where the symbol tr(..) means the trace of the integral operator26,27 . In the absence of
external charges (ρext (r) = 0), the mean-field electrostatic potential ψ(r) = 0, so that the
(M F )
mean-field contribution to the electrostatic free energy Fel = kB T S[0] = 0. Therefore,
the electrostatic free energy is fully determined by the electrostatic potential fluctuations
near its zero value. In that case the renormalized reciprocal Green function takes the form
2Ie2 2nd e2
G −1 (r, r′|0) = G−1 (r, r′ ) = G−1 ′
0 (r, r ) + δ(r − r′ ) + (δ(r − r′ ) − g(r − r′ )), (56)
kB T kB T
whereas the electrostatic free energy is26
kB T V kB T dk G0 (k)
Z
Fel ≈ tr (ln G0 − ln G) = 3
ln , (57)
2 2 (2π) G(k)
where G0 (k) = 1/(εε0k 2 ) and G(k) = 1/(εε0(k 2 + κ 2 (k))) are the Fourier-images of the
Green functions. Hence, the electrostatic free energy of the ion-dipole mixture in the RPA
can be written in the standard form27,33
κ 2 (k) κ 2 (k)
   
V kB T dk
Z
Fel = ln 1 + − , (58)
2 (2π)3 k2 k2
where the screening function κ 2 (k) is determined by relation (45). Note that we have
subtracted the electrostatic self-energy of the system from the electrostatic free energy33 .
We would like to stress that integral (58) converges at the ultraviolet limit.
For the Yukawa-type PDF, determined by eq. (31), the integral in (58) can be taken
analytically only in the absence of ions in the system (n± = 0), yielding
V kB T
Fel = − σ(yd ), (59)
l3
where the auxiliary dimensionless function
√
6
σ(yd ) = (2(1 + yd )3/2 − 2 − 3yd ) (60)
4π
is introduced. It is instructive to analyze the limiting regimes following from relations (59-
60). Thus, we get
√

− 6e4 ln2
d
, yd ≪ 1

Fel 48π(εε0 kB T )2
= (61)
V kB T − 1 3 , yd ≫ 1.

12πrD

15
 √
The first regime determines the case when the mean-square distance l/ 6 is much less than
−1/2
the Debye screening length rD = (2nd e2 /εε0kB T ) introduced above and attributed to
the charged groups of dipolar particles. In that case, the electrostatic correlations consist of
the pairwise effective Keesom interaction of the dipolar particles36,37 . In the opposite case,
√
when the mean-square distance l/ 6 is much bigger than the Debye screening length rD ,
the electrostatic free energy can be described by the standard Debye-Hueckel (DH) limiting
law. The latter strictly justifies the idea proposed first in work6 that charged groups of very
long dipolar particles can be modelled as unbonded point-like ions.
Now let us consider the limiting behavior of the electrostatic free energy in the presence
of salt at yd ≪ 1 and yd ≫ 1. At yd ≪ 1 we have
√ √ √
Fel κ3s 3 6 ys 3 6 1 + 3 ys + ys 2
=− − √ yd − √ y + O(yd3), (62)
V kB T 12π 2πl3 1 + ys 16πl3 (1 + ys )3 d

where the first term in the right hand side describes the DH contribution to the electrostatic
free energy from the ionic component of the mixture. The second and third terms describe,
respectively, the contributions of ion-dipole and dipole-dipole pairwise correlations. In the
opposite regime when yd ≫ 1, we obtain the standard DH limiting law
3/2
2(nd + I)e2

Fel 1
≃− . (63)
V kB T 12π εε0 kB T

In the latter case, dipolar particles charged groups manifest themself as unbonded ions
participating in the charge screening together with salt ions.

III. CONCLUDING REMARKS AND PERSPECTIVES

In conclusion, we have formulated a nonlocal statistical field theory of the dilute elec-
trolyte solutions with admixture of a small amount of dipolar particles, consideringmodelling
the latter as dimers comprising two oppositely charged groups located at the fluctuating
distances relative from each other. Attributing to every dipolar particle an arbitrary proba-
bility distribution function of the distance between its charge centers and using the standard
Hubbard-Stratonovich transformation, we have represented the configuration integral of the
system in the functional integral form with the functional in integrand exponent, gener-
alizing the Poisson-Boltzmann-Langevin functional10–12 obtained earlier for the ions mixed
with point-like dipolar particles. Within the mean-field approximation we have derived a

16
nonlinear integro-differential equation for the mean-field potential. Linearizing the obtained
self-consistent field equation, we have derived the general expressions for the electrostatic
potential around the point-like test ion and its solvation free energy in the solution of salt-
free dipolar particles, as well as in the solution with the nonzero salt concentration. For
the ’Yukawa’-type probability distribution function we have obtained analytic expressions
for the electrostatic potential of the point-like test ion and its solvation free energy. We
have obtained a general relation for the bulk electrostatic free energy in the Random phase
approximation expressed through the integral over the wave vectors of the reciprocal space
converging at the ultraviolet limit. For Yukawa-type probability distribution function we
have obtained an analytic relation for the electrostatic free energy of the salt-free solution of
dipolar particles, analyzing all limiting regimes that follow from it. We have also analyzed
the limiting regimes, following from the general relation for the electrostatic free energy of
the dipolar particles solution in case of the Yukawa-type probability distribution function in
the presence of salt.

Finally, we would like to discuss further applications of the present theory. As is well
known, in the dipolar fluids at a rather low temperature chalin-like clusters are formed,
preventing the expected liquid-gas phase separation6,38,39 . In contrast to the first, quite
eclectic theory6,39 taking into account the chain-like cluster formation in the dipolar fluids,
the present theory allows us, in principle, to calculate the electrostatic free energy of the
dipolar chain clusters mixture within one formalism. Thus, it is interesting to apply the
present formalism to the description of dipolar fluids phase behavior taking into account for
the cluster formation of the dipolar particles and their dispersion and excluded volume in-
teractions. Especially, it is interesting to answer the following question: How will accounting
for the chain-like cluster formation change the self-consistent field equation (21)? Answer-
ing that question will give us an understanding of how the presence of chain-like dipolar
clusters in a polar fluid will renormalize the electrostatic potential of the test ion and its
solvation free energy. There are also two advantages of the present theory over the previ-
ously published statistical theory of polar fluids6,39 . Firstly, in contrast to that theoretical
model, the present theory can be easily generalized to the case of dipolar particles mixture
with an arbitrary number of components. Secondly, our theory can be generalized to the
particles with an arbitrary electric structure. For this purpose, it is necessary to introduce
a probability distribution function of distance for each pair of oppositely charged groups of

17
the fluid molecule. A systematic study of all these issues will be the subject of forthcoming
publications.

ACKNOWLEDGMENTS

The author has benefited from conversations with I.Y. Erukhimovich and A.A. Kornyshev.

18
REFERENCES

1
Onsager L 1936 J. Am. Chem. Soc. 58 (8) 1486
2
Kirkwood J 1939 J. Chem. Phys. 7 911
3
Hoye J S and Stell G 1976 J. Chem. Phys. 64 1952
4
Chandler D 1977 J. Chem. Phys. 67 1113
5
Iukhnovskii I R, Golovko M F 1980 Statistical theory of classical equilibrium systems (Kiev,
Izdatel’stvo Naukova Dumka, In Russian)
6
Levin Y 1999 PRL 83 (6) 1159
7
Fisher M, Levin Y 1993 PRL 71 3826
8
Nienhuis G and Deutch J M 1972 J. Chem. Phys. 56 235
9
Weiss V C and Schroer W 1998 J. Chem. Phys. 108 7747
10
Abrashkin A, Andelman D and Orland H 2007 PRL 99 077801
11
Coalson R D, Duncan A 1996 J. Phys. Chem. 1996 100 2612
12
Budkov Yu A, Kolesnikov A L and Kiselev M G 2015 EPL 111 28002
13
Kornyshev A A and Vorotyntsev M A 1980 Surface Science 101 23
14
Levy A, Andelman D and Orland H 2012 PRL 108 227801
15
Dean D S and Podgornik R 2012 J. Chem. Phys. 136 154905
16
Budkov Yu A and Kolesnikov A L 2016 Eur. Phys. J. E 39 110
17
Budkov Yu A, Kalikin N N and Kolesnikov A L 2017 Eur. Phys. J. E 40 47
18
Martin J M, Li W, Delaney K T, and Fredrickson G H 2016 J. Chem. Phys. 145 154104
19
Canchi D R and Garcia A E 2013 Annu. Rev. Phys. Chem. 64 273
20
Haran G 2012 Curr. Opin. Struct. Biol. 22 14
21
Lowe A B and McCormick C L 2002 Chem. Rev. 102 4177
22
Kudaibergenov S, et al 2006 Advances in Polymer Science 201 157
23
Heldebrant D J, et al 2010 Green Chemistry 12 713
24
Budkov Y A, Kolesnikov A L, Kiselev M G 2016 J. Chem. Phys. 144 184703
25
Gongadze E, Iglic A. 2015 Electrochim. Acta 178 541
26
Podgornik R 1989 J. Chem. Phys. 91 5840
27
Lue L 2006 Fluid Phase Equilibria 241 236
28
Netz R R 2001 Eur. Phys. J. E 5 557
29
Efimov G V, Nogovitsin E A 1996 Physica A 234 506

19
30
Gnedenko B V 2001 The theory of probability and the elements of statistics (Fifth edition,
AMS Chelsea Publishing)
31
Barrat J L and Hansen J P 2003 Basic Concepts for Simple and Complex Liquids (Cam-
bridge: University Press)
32
Khokhlov A R, Khachaturian K A 1982 Polymer 23 (12) 1742
33
Borue V Yu and Erukhimovich I Ya 1988 Macromolecules 21 3240
34
Ramshaw J D 1978 J. Chem. Phys. 68 5199
35
Ramshaw J D 1980 J. Chem. Phys. 73 3695
36
Reinganum M 1912 Ann. Phys. (Leipzig) 38 649
37
Keesom W H 1916 Proc. K. Ned. Akad. Wet. 18 636
38
Weis J J and Levesque D 1993 PRL 71 2729
39
Levin Y, Kuhna P S, Barbosa M C 2001 Physica A 292 129

20