1

1 Introduction
In this paper we will study the liquid vapor interface in microscopic level. Here we
will study the system where it has density variation only in one direction, i.e the
one dimensional case. We consider one component system at constant tempera-
ture and volume. As we all know from the thermodynamics that in equilibrium
the free energy that minimizes at constant temperature and volume is Helmholtz
free energy. So the density profile in one dimension of such a system will be ob-
tained by minimizing the free energy. We express the Helmholtz free energy as a
integral of local free energy density which is a functional of local molecular num-
ber density and its derivative. Here the determination of equilibrium molecular
number density profile becomes a problem of calculus of variation. Here the ex-
tremization of the integral is constrained by total number of molecule is constant,
i.e. vapour becoming liquid or liquid becoming vapour, no external ingrained is
being mixed to the system. From that variational method the Euler-Lagrange
equation can be obtained with a Lagrange undetermined multiplier. Solving the
Euler-Lagrange equation a density function is obtained in a given chemical po-
tential. From the analysis of the one dimension differential equation we obtain
description of the cases of a single phase and two coexisting phases.

2 Density variation in one dimension; The flat

interface
In one-component system, below its critical temperature, liquid and vapor phase
can coexist in equilibrium. In presence of gravity the interface separating the
liquid and vapour phase is perpendicular to the direction of gravity and the density
will only vary in one direction, i.e. x direction here. We know from our day by

Figure 1: Density profile for a pure system in the neighborhood of the gas-liquid interface

day experience that the density interior to the liquid is almost constant and as we
go towards the surface it start to decreasing and in the flat interface the vapour
density is high but as we go further the vapour density also start decreasing and
after some distance it also get constant(figure 1). But the density is no where
discontinuous. We are starting from the integral of Helmholtz free energy density.
We can see in appendix 1 that the Helmholtz free energy can be expressed as the

2
integral of local free energy density of the form.
Z
F = d3 r ψ(ρ(r), ∇ρ(r), r) (1)

When the molecular number density ρ(r) is sufficiently slowly varying in its spacial
dependence on a scale set by the interaction range, we find ψ(r) is given by
1
ψ(ρ(r), ∇ρ(r), r) = U (r)ρ(r) + ψ̃(ρ(r)) + A(ρ(r))(∇ρ(r))2 (2)
2
U(r) is the external potential energy function and ψ̃(ρ) is the free energy density
of an equilibrium uniform system having density ρ. A(ρ) is related to the direct
correlation function C(r, ρ) of a uniform system having density ρ by
Z
kT
A(ρ) = d3 r r2 C(r, ρ) (3)
6

2.1 Minimization of free energy in one dimensional case

In one dimensional case, where the density only varies in x direction equation 1
can be written as Z b
F = ã dx ψ(ρ(x), ρ0 (x), x) (4)
a
where ã is the cross sectional area of that flat surface.
Equation 2 reduces to
1
ψ(ρ(x), ρ0 (x), x) = U (x)ρ(x) + ψ̃(ρ(x)) + A(ρ(x))(ρ0 (x))2 (5)
2
Here we will study first where there is no external potential, i.e. U (x) = 0. We
wish to determine the density function ρ(x) which minimizes the definite integral
in the equation 4. The system is closed with total volume V with fixed boundaries
at a and b and total number of particle N. So here the constraint equation is
Z b
ã ρ(x) = N (6)
a

We are introducing a Lagrange undetermined multiplier µ, and for our purpose

we have to minimize the integral
Z b Z b
I= dx[ψ(x) − µρ(x)] ≡ P0 (x) dx (7)
a a

The minimization on leads to the Euler-Lagrange equation

∂P0 (x) d ∂P0 (x)
− =0 (8)
∂ρ(x) dx ∂ρ0(x)
3
From here we can get a relation like
1
A(ρ(x))ρ00 (x) + A0 (ρ(x))(ρ0 (x))2 − µ(ρ(x)) + µ = 0 (9)
2
Where,
∂A(ρ)
A0 (ρ) = (10)
∂ρ
∂ψ(ρ)
µ(ρ) = (11)
∂ρ
We know from thermodynamics that the derivative of Helmholtz free energy

Figure 2: A typical chemical potential isotherm for a one component system

with respect to number density gives the chemical potential as in equation 11.
Since the equation 9 has no explicit x dependence so we can argue that if ρ(x)
is a solution of the differential equation then ρ(x + c) is also a solution of that
differential equation. Where c is an arbitrary constant. Since the boundaries of
the system are fixed at a and b so replacing ρ(x) by ρ(x+c) is equivalent to moving
the phase boundary to −c, i.e. we are decreasing one component and increasing
other and interestingly this is not effecting the phase equilibrium. So we can
conclude that the phase equilibrium does not depend on amount of substance in
each phase. Now let us analyse the equation 9.

Figure 3: µ(ρ) -µ for three different values of µ

Multiplying equation 9 by ρ0 (x) we get

d( 21 A(ρ0 )2 )
0
µ(ρ(x)) − µρ (x) = (12)
dx
4
Figure 4: (a) g1 (ρ) for a chemical potential for which µ(ρ) − µ has only one zero at ρ0 . (b) Density profiles for
solutions which approaches ρ0 asymptotically.

as ρ0 (x)dx = dρ, so we can have

Z
1 02
Aρ = {µ(ρ) − µ}dρ + C1 ≡ g1 (ρ) (13)
2
In a single phase interior we must have ρ0 and ρ00 approaching zero. From the
equation 9 we can see that µ(ρ) − µ approaches zero at boundaries. Equivalently,
02)
both Aρ02 and d(Aρdρ . In view of equation 13 we have the liberty to choose the
integration constant C1 such that g1 (ρ) become zero. A typical chemical potential
isotherm is shown in figure 2. µ will be determined by fixing the average density
of the system N/V . Different values of µ will give us three zones a, b, c as in
figure 4. Different values of µ causes µ(ρ) − µ to have one, two, three zones in
figure 4a, 4b, 4c respectively.

2.2 Study the three different zones for chemical potential

Case A: µ(ρ) − µ has one zero at ρ0 , as in figure 3a.
The plot of g1 (ρ) has one extremum at ρ0 and by suitable choice of C1 we
make that extremum pint at zero as in the figure 4a. The physical solution must
satisfy the condition ρ0 approaches zero at both ends. The only physical solution
in this case is a constant solution, ρ = ρ0 , the one phase state.
Case B: µ(ρ) − µ has two zeros at ρ1 and ρ2 as in figure 3b.
The plot of g1 (ρ) vs ρ has one minima at ρ1 and a point of inflection at ρ2 . With
the help of the integration constant we can make zero of g1 (ρ) either at ρ = ρ1
or at ρ = ρ2 . However only the solution ρ = ρ1 is stable since compressibility is
infinite at ρ = ρ2 . Thus ρ = ρ1 is the only physical solution. See fig 5.
Case C: µ(ρ) − µ has threee zeros at ρG , ρI , ρL respectively as in figure 3c.
The plot of g1 vs ρ will have two minima and one maxima as in figure 6a.
Either ρG or ρL have the stable solution, the choice depends on the absolute
minimum of the integral. The other solution corresponds to the metastable state.
The solution corresponds to ρI is unstable, the compressibility being negative.
We can achieve zeros ρG0 and ρL0 simultaneously, as in figure 8. The condition

5
 Figure 5: (a) g1 (ρ) can be made to be zero at ρ1 or at ρ2 (b) Density profiles which approaches ρ1 or ρ2

for having two minima at the same height is

Z ρI Z ρL
0 0
{µ(ρ) − µ}dρ + {µ(ρ) − µ}dρ = 0 (14)
ρG 0 ρ I0

Equation 14 is known as the Maxwell’s

equal area construction. We can write equa-
tion 14 as
Z ρL
0 = dρ[µ(ρ) − µ] (15)
ρG
= [ψ(ρL0 ) − µ(ρL0 )ρL0 ]
−[ψ(ρG0 ) − µ(ρG0 )ρG0 ] (16)

We know the from Grand potential that

Φ(T, V, N ) = U − T S − µN . We can write
it in the fashion Φ(T, V, N ) = F − µN =
(ψ(ρ) − ρµ(ρ))V . As we know from thermo-
∂Φ
dynamics P = −( ∂V ), so P (ρ) = ρµ(ρ) −
ψ(ρ). Thus equation 16 says that the equi-
librium pressure of the coexisting phases are
same. Figure 6: (a) g1 (ρ) showing two minima at ρG and
ρL . For different choices of integration constant
Equation 12 can be written as. g1 (ρ) can be made to be zero at either at ρG or
ρL . (b) Density profile which approaches either ρL
d{µρ(x) − ψ(ρ(x)) + 12 A(ρ(x)(ρ0 (x))2 } or ρG
= 0 (17)
dx
1
i.e.P = µρ(x) − ψ(ρ(x)) + A(ρ(x)(ρ0 (x))2 = constant
(18)
2

6
Figure 7: (a) g1 (ρ) when µ is chosen so that the two minima are at the same height. (b) The density profile for
physical solution. (c)Maxwell equal area construction

In the single phase interior where ρ0 → 0 and

µ(ρ) → µ, it is
P = P (ρ) = µ(ρ)ρ − ψ(ρ) (19)
So it is clear that P in equation 19 is the pressure. Integrating over total
volume in equation 19 gives

P V = µN − P V + σ0 ã (20)

where the surface tension is σ0 is given by

Z b
σ0 = dx A(ρ(x))(ρ0 )2 (21)
a

Thus the interestingly we have not consider the surface as true discontinuity
but the surface contribution to the free energy has come automatically.

2.3 Conclusion
So we have derived the surface energy term in terms direct correlation function
and the derivative of density. So for any particular type of potential we can
find the direct correlation function then from the equation 3 we can derive a
functional form of A. And on other hand solving for a particular type of chemical
potential isotherm numerically equation 9 we can find the density profile.With
the functional form of ρ(x) and A(ρ) we can find the surface energy functional.

7
3 Appendix
3.1 Derivation of free energy density

we assume that the Hamiltonian of the system is described in the form...

N
X |pi |2
H= + Vint + Vext (22)
i=1
2m

where,
N
X
Vext = Φ(ri ) (23)
i=1
i.e. the external field is the sum of the single particle contributions. The micro-
scopic one particle density is
N
X
ρM = δ(r − ri ) (24)
i=1
so, Z
Vext = d3 ρM Φ(r) (25)

The grand partition function

Ξ = T r exp {−β(H − µN )} (26)
where,
N Z
X 1
Tr = d3 rd3 p (27)
i=1
N!
Now,
H − µN (28)
N N
|pi |2
X Z X
3
= + Vint + d r ρM Φ(r) − µ (29)
i=1
2m i=1
N
|pi |2
X Z Z
= + Vint + d3 r ρM Φ(r) − d3 r ρM µ (30)
i=1
2m
N
|pi |2
X Z
= + Vint + d3 r ρM (Φ(r) − µ) (31)
i=1
2m
so,
N
|pi |2
X Z
Ξ = T r exp {−β( + Vint ) + d3 r ρM β(µ − Φ(r))} (32)
i=1
2m

8
We define f (rN , pN ) as the probability density function,i.e having first particle at
position r1 and momentum p1 , second particle at position at r2 with momentum p2
etc. So, having the probability density for the positions are i.e having first particle
at position r1 and second particle at r2 and so on irrespective of all momentum
is given by Z
n(rN ) = d3N p f (rN , pN ) (33)

Now,
P |pi |2
T r ρM exp {−β( N
R 3
i=1 2m + V int ) + d r ρM u(r)}
< ρM > = (34)
Ξ
δ ln Ξ
= (35)
δu
Now, the probability that particle 1 will be in the vicinity 1 irrespective of the
other particle is
Z
1
n(1/n) (r1 ) = d3 r1 d3 r2 .... d3 rN n(rN ) = (36)
V
Now, the probability of any particle at r1 is
N
X Z
(1/N ) (1/N )
ρ = n (r1 ) = ρM n( 1/N ) d3 r (37)
i=1
N
= = ρM n(1/N ) (r1 ) =< ρM > (38)
V
So,
δlnΞ
< ρM >= (39)
δu
Now, The Helmholtz free energy is defined by
1
F = µN − P V = µN − lnΞ =⇒ βF = βµN − lnΞ (40)
β
In our case the Hamiltonian itself depends the one particle number like µN de-
pends in the case of Grand Canonical partition function. Here µ is replaced by
u(r) and defined. We define a dimension less free energy as
Z
F0 = d3 r u(r) ρ(r) − lnΞ (41)

9
from equation 40 and 41 we get
Z
βF = βµN − F 0 − d3 r u(r) ρ(r) (42)
Z Z
= β d3 r ρ(r) µ + F 0 − d3 r u(r) ρ(r) (43)
Z
= d3 r (βµ − u(r)) ρ(r) + F 0 (44)
Z
= F0 + d3 r β Φ(r) ρ(r) (45)

Ideal gas situation:

In an ideal gas there is no interaction term in Hamiltonian.
N
p2i
X Z
Ξid = T r exp {−β( )+ d3 r ρM u(r)} (46)
i=1
2m
N N
X p2i X
= T r exp {−β( )+ u(ri )} (47)
i=1
2m i=1
N N N N
p2i
Z
X 1 Y X X
= dpi dqi exp {−β( )+ u(ri )} (48)
i=1
N ! i=1 i=1
2m i=1
N Z Z
X 1
= [λ−3 d3 r exp {u(r)}]N (49)
i=1
N!
Z
−3
= exp {λ d3 r exp {u(r)} (50)

So,
Z
−3
lnΞ = λ d3 r exp {u(r)} (51)
δlnΞ
ρ(r) = (52)
δu(r)
exp {u(r)}
= (53)
λ3
=⇒ u(r) = ln{λ3 ρ(r)} (54)
so,
Z
lnΞ = ρ(r) d3 r (55)
Z Z
F00 = d3 r ln[λ3 ρ(r)]ρ(r) − d3 r ρ(r) (56)
Z
= d3 r ρ(r)[ln{λ3 ρ(r)} − 1] (57)

10
Now, let consider the interaction in the system and in that case for the density
ρ(r) let the free energy is

F 0 = F00 + Fext (58)

=⇒ F 0 = F00 − φ (59)
=⇒ φ = F00 − F 0 (60)

for our case,

δF 0
|ρ(r) = u(r) (61)
δρ(r)
Note, F00 (ρ(r)) is the ideal gas free energy at density ρ(r).
F 0 (ρ(r)) is the free energy at density ρ(r) but system have interaction.
If ρ(r) → 0, then the interaction become negliable and every system look like
ideal.
As we define
φ(ρ) = F00 (ρ) − F 0 (ρ) (62)
We define the correlation function
∂C(r1 )
C(r1 , r2 ) = = C(r1 , r2 ) (63)
∂ρ(r2 )
∂C(r1 )
C(r1 , r2 , r3 ) = = C(r1 , r2 ) = C(r2 , r1 , r3 ) (64)
∂ρ(r2 )∂ρ(r3 )
∂C(r1 )
C(r1 , r2 , r3 , r4 ) = = C(r1 , r2 , r3 , r4 ) (65)
∂ρ(r2 )
When,

ρ→0 (66)
F00 = F 0 (0) (67)
φ(0) = 0 (68)

and1

φ(ρ) = φ(ρ) − 0 (69)

= φ(ρ) − φ(0) (70)
Z 1
d
= ds φ(sρ) (71)
ds
Z0 1
∂φ(ρ̄) dρ̄
= ds (72)
∂ ρ̄ ds
Z0 Z 1
= ds C(r1 , sρ)sρ(r1 )d3 r1 (73)
0
1
where ρ̄ = sρ

11
 So,
Z Z 1
3
φ(ρ) = d r1 sdsC(r1 , sρ)ρ(r1 ) (74)
0
Z Z 0 Z Z 1
1
= ∗2 d3 r1 sds d3 r2 C(r1 , r2 , ss0 ρ) ρ(r1 )ρ(r2 )(75)
2
Z Z1 Z 1 Z 01
1
= ∗2 d3 r1 d3 r2 sds s0 ds0 C(r1 , r2 , ss0 ρ)ρ(r1 )ρ(r(76)
2)
2Z Z 0 0
1
= C̄(r1 , r2 )ρ(r1 )ρ(r2 )d3 r1 d3 r2 (77)
2Z Z
1
=⇒ F00 − F 0 = C̄(r1 , r2 )ρ(r1 )ρ(r2 )d3 r1 d3 r2 (78)
2 Z Z
1
=⇒ F 0 = F00 − C̄(r1 , r2 )ρ(r1 )ρ(r2 )d3 r1 d3 r2 (79)
2Z Z
1
=⇒ F 0 = F00 − C̄(r1 , r1 + r2 )ρ(r1 )ρ(r1 + r2 )d3 r1 d3 r2 (80)
Z 2
=⇒ F 0 = d3 r ρ(r)[ln{λ3 ρ(r)} − 1]
Z Z
1
− C̄(r, r + r2 )ρ(r)ρ(r + r2 )d3 rd3 r2 (81)
Z 2
=⇒ F 0 = d3 r ρ(r)[{ln{λ3 ρ(r)} − 1}
Z
1
− d3 r2 C̄(r, r + r2 )ρ(r + r2 )] (82)
Z 2
=⇒ F 0 = d3 r f¯(r) (83)

Where,
Z
1
f¯(r) = ρ(r)[ln{ρ(r)λ − 1} − 3
d3 r2 C̄(r, r + r2 )ρ(r + r2 )] (84)
2

Taylor expansion of C̄ about ρ(r) is given by

Z 
δ C̄(r, r2 ) 
C̄(r, r2 ) = C̄(r, r2 ; ρ(r)) + [ρ(r3 ) − ρ(r)]dr3
δρ(r3 ) ρ(r)
δ 2 C̄(r, r2 ) 
Z 
1
+ [ρ(r3 ) − ρ(r)][ρ(r4 ) − ρ(r)]dr3 dr4
2! δρ(r3 )δρ(r4 )  ρ(r)
+ ................ (85)

12
 Z
C̄(r, r2 ) = C̄(r, r2 ; ρ(r)) + C̄(r, r2 , r3 ; ρ(r))[ρ(r3 ) − ρ(r)]dr3
Z
1
+ C̄(r, r2 , r3 , r4 ; ρ(r))[ρ(r3 ) − ρ(r)][ρ(r4 ) − ρ(r)]dr3 dr4
2!
+ ................ (86)

where

Z 1 Z 1
C̄(r, r2 ) = 2 sds C(r, r2 ; ss0 )d0 (87)
Z0 1 Z0 1
C̄(r, r2 , r3 ) = 2 sds C(r, r2 , r3 ; ss0 )ds0 (88)
0 0
Z 1 Z 1
C̄(r, r2 , r3 , r4 ) = 2 sds C(r, r2 , r3 , r4 ; ss0 )ds0 (89)
0 0

C(r, r2 ), C(r, r2 , r3 ), C(r, r2 , r3 , r4 ) are the two-, three-, four-particle direct

correlation functions respectively.
In accordance with translational invariance, the correlation functions depend
only upon the inter-particle distance:

C(r, r2 ) = C 0 (r2 − r) (90)

C(r, r2 , r3 ) = C 0 (r2 − r, r3 − r) (91)
C(r, r2 , r3 , r4 ) = C 0 (r2 − r, r3 − r, r4 − r) (92)

In our case, in accordance with translational invariance, we have

C̄(r, r + r2 ; ρ(r)) = C̄(r2 ; ρ(r)) (93)

C̄(r, r + r2 , r + r3 ; ρ(r)) = C̄(r2 , r3 ; ρ(r)) (94)
C̄(r, r + r2 , r + r3 , r + r4 ; ρ(r)) = C̄(r2 , r3 , r4 ; ρ(r)) (95)

Taylor expansion of C̄(r, r + r2 ) about ρ(r), we have

Z
C̄(r, r + r2 ) = C̄(r, r + r2 ; ρ(r)) + C̄(r, r + r2 , r + r3 ; ρ(r))
[ρ(r + r3 ) − ρ(r)]dr3
Z
1
+ C̄(r, r + r2 , r + r3 , r + r4 ; ρ(r))[ρ(r + r3 ) − ρ(r)]
2!
[ρ(r + r4 ) − ρ(r)]dr3 dr4 + ................ (96)

In order to make the above equation valid under translational invariance, we

can write

13
 Z
C̄ 0 (r2 ) = C̄ 0 (r2 ; ρ(r)) + C̄ 0 (r2 , r3 ; ρ(r))[ρ(r + r3 ) − ρ(r)]dr3
Z
1
+ C̄ 0 (r2 , r3 , r4 ; ρ(r))[ρ(r + r3 ) − ρ(r)][ρ(r + r4 ) − ρ(r)]
2!
dr3 dr4 + ........................ (97)

Under translational invariance, Eq.(84) becomes

Z
1
f¯(r) = ρ(r)[ln{λ ρ(r)} − 1] − ρ(r) C̄ 0 (r2 )ρ(r + r2 )dr2
3
(98)
2
Substitute Eq.(97) into Eq.(98), we have

Z  Z
1
f¯(r) = ρ(r)[ln{λ ρ(r)} − 1] − ρ(r)
3
C̄ (r2 ; ρ(r)) + C̄ 0 (r2 , r3 ; ρ(r))
0
2
Z
1
[ρ(r + r3 ) − ρ(r)]dr3 + C̄ 0 (r2 , r3 , r4 ; ρ(r))[ρ(r + r3 ) − ρ(r)]
2!

[ρ(r + r4 ) − ρ(r)]dr3 dr4 ρ(r + r2 )dr2 (99)

Z
1
f¯(r) = ρ(r)[ln{λ ρ(r)} − 1] − ρ(r) C̄ 0 (r2 ; ρ(r))ρ(r + r2 )dr2
3

Z 2
1
− ρ(r) C̄ 0 (r2 , r3 ; ρ(r))ρ(r + r2 )[ρ(r + r3 ) − ρ(r)]dr2 dr3
2 Z
1
− ρ(r) C̄ 0 (r2 , r3 , r4 ; ρ(r))ρ(r + r2 )[ρ(r + r3 ) − ρ(r)]
4
[ρ(r + r4 ) − ρ(r)]dr2 dr3 dr4 (100)

Taylor Expansion
A Taylor series is a series expansion of a function about a point. A one
dimensional Taylor expansion of a single variable real function f (x) about a point
x = x0 is given by

00 (3)
0 2f (x0 ) 3 f (x0 )
f (x) = f (x0 ) + (x − x0 )f (x0 ) + (x − x0 ) + (x − x0 ) +
2! 3!
(n)
nf (x0 )
................ + (x − x0 ) + .................. (101)
n!
An alternate form of one dimensional Taylor series may be obtained by letting

14
 x − x0 = ∆x (102)
x = x0 + ∆x (103)

Substitute Eq.(102),(103) into Eq.(101), we have

00 (3)
0 2f (x0 ) 3 f (x0 )
f (x0 + ∆x) = f (x0 ) + ∆x f (x0 ) + (∆x) + (∆x) +
2! 3!
(n)
nf (x0 )
................ + (∆x) + .................. (104)
n!
In general
∞   n 
X 1 ∂
f (x0 + ∆x) = ∆x f (x) (105)
n=0
n! ∂x x=x0

A Taylor series expansion of a real function of two variables f (x, y) about

(x0 , y0 ) is given by

 
∂f (x, y) ∂f (x, y)
f (x0 + ∆x, y0 + ∆y) = f (x0 , y0 ) + ∆x + ∆y
∂x ∂y x=x0 ,y=y0
 2
1 ∂f (x, y) ∂f (x, y)
+ ∆x + ∆y
2! ∂x ∂y x=x0 ,y=y0
 3
1 ∂f (x, y) ∂f (x, y)
+ ∆x + ∆y
3! ∂x ∂y x=x0 ,y=y0
+ .................................. (106)

f (x0 + ∆x, y0 + ∆y) =

 
∂f (x, y) ∂f (x, y)
f (x0 , y0 ) + ∆x + ∆y
∂x ∂y x=x0 ,y=y0
2 2
∂ 2 f (x, y)
 
1 2 ∂ f (x, y) 2 ∂ f (x, y)
+ (∆x) + (∆y) + 2∆x∆y
2! ∂x2 ∂y 2 ∂x∂y x=x0 ,y=y0
3 3
∂ 3 f (x, y)

1 3 ∂ f (x, y) 3 ∂ f (x, y) 2
+ (∆x) + (∆y) + 3(∆x) ∆y
3! ∂x3 ∂y 3 ∂x2 ∂y
3

2 ∂ f (x, y)
+ 3∆x(∆y) + .................................. (107)
∂x∂ 2 y x=x0 ,y=y0

In general

15
 ∞   n 
X 1 ∂ ∂
f (x0 + ∆x, y0 + ∆y) = ∆x + ∆y f (x, y) (108)
n=0
n! ∂x ∂y x=x0 ,y=y0

A Taylor series expansion of a real function of three variables f (x, y, z) about

(x0 , y0 , z0 ) is given by

f (x0 + ∆x, y0 + ∆y, z0 + ∆z) =


∂f (x, y, z) ∂f (x, y, z)
f (x0 , y0 , z0 ) + ∆x + ∆y
∂x ∂y
 
∂f (x, y, z) ∂f (x, y, z) ∂f (x, y, z)
+ ∆z + ∆x + ∆y
∂z x=x0 ,y=y0 ,z=z0 ∂x ∂y
2 
∂f (x, y, z) ∂f (x, y, z) ∂f (x, y, z)
+ ∆z + ∆x + ∆y
∂z x=x0 ,y=y0 ,z=z0 ∂x ∂y
3
∂f (x, y, z)
+ ∆z + ........................................... (109)
∂z x=x0 ,y=y0 ,z=z0

In general

f (x0 + ∆x, y0 + ∆y, z0 + ∆z) =

∞   n
X 1 ∂ ∂ ∂
∆x + ∆y + ∆z
n=0
n! ∂x ∂y ∂z

f (x, y, z) (110)
x=x0 ,y=y0 ,z=z0

In vector form
∞   n 
X 1
f (r0 + ∆r) = ∆r· ∇ f (r)
n=0
n! r=r0

The only contribution to the integrals in Eq.(100) comes from the points near
to the field point r. Thus we can Taylor expand ρ(r + r2 ) about the point r:

1
ρ(r + r2 ) = ρ(r) + r2 · ∇ρ(r) + (r2 · ∇)2 ρ(r) + .................... (111)
2!

Keeping above equation upto second order in gradient and substitute it into
Eq.(100), we obtain

16
 Z 
 1
f¯(r) = 3 0

ρ(r) ln{λ ρ(r)} − 1 − ρ(r) C̄ (r2 ; ρ(r)) ρ(r) + r2 · ∇ρ(r)
2
 Z 
1 1
+ (r2 · ∇)2 ρ(r) dr2 − ρ(r) C̄ 0 (r2 , r3 ; ρ(r)) ρ(r) + r2 · ∇ρ(r)
2! 2

1
(r2 · ∇)2 ρ(r) ρ(r + r3 ) − ρ(r) dr2 dr3
 
+
2!
Z  
1 0 1 2
− ρ(r) C̄ (r2 , r3 , r4 ; ρ(r)) ρ(r) + r2 · ∇ρ(r) + (r2 · ∇) ρ(r)
4 2!
  
ρ(r + r3 ) − ρ(r) ρ(r + r4 ) − ρ(r) dr2 dr3 dr4 (112)
Taylor expand all the terms. The above equation becomes

Z
1
f¯(r) = ρ(r) ln{λ3 ρ(r)} − 1 − ρ(r) C̄ 0 (r2 ; ρ(r))ρ(r)dr2
 

Z  2 
1 0 1 2
− ρ(r) C̄ (r2 ; ρ(r)) r2 · ∇ρ(r) + (r2 · ∇) ρ(r) dr2
2 2!
Z  
1 1
− ρ(r) C̄ 0 (r2 , r3 ; ρ(r)) ρ(r) + r2 · ∇ρ(r) + (r2 · ∇)2 ρ(r)
2 2!
 
1
ρ(r) + r3 · ∇ρ(r) + (r3 · ∇)2 ρ(r) − ρ(r) dr2 dr3
2!
Z  
1 1
− ρ(r) C̄ 0 (r2 , r3 , r4 ; ρ(r)) ρ(r) + r2 · ∇ρ(r) + (r2 · ∇)2 ρ(r)
4 2!
 
1
ρ(r) + r3 · ∇ρ(r) + (r3 · ∇)2 ρ(r) − ρ(r)
2!
 
1 2
ρ(r) + r4 · ∇ρ(r) + (r4 · ∇) ρ(r) − ρ(r) dr2 dr3 dr4 (113)
2!

Z  
1 1
f¯(r) = f0 (ρ(r)) − ρ(r) C̄ (r2 ; ρ(r)) r2 · ∇ρ(r) + (r2 · ∇) ρ(r) dr2
0 2
2 2!
Z  
1 1
− ρ(r) C̄ 0 (r2 , r3 ; ρ(r)) ρ(r) + r2 · ∇ρ(r) + (r2 · ∇)2 ρ(r)
2 2!
 
1
r3 · ∇ρ(r) + (r3 · ∇)2 ρ(r) dr2 dr3
2!
Z  
1 0 1 2
− ρ(r) C̄ (r2 , r3 , r4 ; ρ(r)) ρ(r) + r2 · ∇ρ(r) + (r2 · ∇) ρ(r)
4 2!
  
1 1
r3 · ∇ρ(r) + (r3 · ∇)2 ρ(r) r4 · ∇ρ(r) + (r4 · ∇)2 ρ(r)
2! 2!
dr2 dr3 dr4 (114)

17
 where
Z
 1
f0 (ρ(r)) = ρ(r) ln{λ3 ρ(r)} − 1 − ρ(r) C̄ 0 (r2 ; ρ(r))ρ(r)dr2

(115)
2
Z  
1 1
f¯(r) = f0 (ρ(r)) − ρ(r) C̄ 0 (r2 ; ρ(r)) r2 · ∇ρ(r) + (r2 · ∇)2 ρ(r) dr2
2 2!
Z  
1 1
− ρ(r) C̄ 0 (r2 , r3 ; ρ(r))ρ(r) r3 · ∇ρ(r) + (r3 · ∇)2 ρ(r) dr2 dr3
2 2!
Z  
1 0 1 2
− ρ(r) C̄ (r2 , r3 ; ρ(r)) r2 · ∇ρ(r) + (r2 · ∇) ρ(r)
2 2!
  Z
1 1
r3 · ∇ρ(r) + (r3 · ∇)2 ρ(r) dr2 dr3 − ρ(r) C̄ 0 (r2 , r3 , r4 ; ρ(r))
2! 4
  
1 2 1 2
ρ(r) r3 · ∇ρ(r) + (r3 · ∇) ρ(r) r4 · ∇ρ(r) + (r4 · ∇) ρ(r)
2! 2!
Z 
1
dr2 dr3 dr4 − ρ(r) C̄ 0 (r2 , r3 , r4 ; ρ(r)) r2 · ∇ρ(r) +
4
 
1 1
(r2 · ∇)2 ρ(r) r3 · ∇ρ(r) + (r3 · ∇)2 ρ(r) r4 · ∇ρ(r) +
2! 2!

1
(r4 · ∇)2 ρ(r) dr2 dr3 dr4 (116)
2!
Dropping all terms of higher than second order in gradients from above equa-
tion, we obtain
Z  
1 1
f¯(r) = f0 (ρ(r)) − ρ(r) C̄ 0 (r2 ; ρ(r)) r2 · ∇ρ(r) + (r2 · ∇)2 ρ(r) dr2
2 2!
Z  
1 1
− ρ(r) C̄ 0 (r2 , r3 ; ρ(r))ρ(r) r3 · ∇ρ(r) + (r3 · ∇)2 ρ(r) dr2 dr3
2 2!
Z
1
ρ(r) C̄ 0 (r2 , r3 ; ρ(r)) r2 · ∇ρ(r) r3 · ∇ρ(r) dr2 dr3



−
2 Z
1
ρ(r) C̄ 0 (r2 , r3 , r4 ; ρ(r))ρ(r) r3 · ∇ρ(r) r4 · ∇ρ(r)



−
4
dr2 dr3 dr4 (117)
Using rotational invarience we can write equation 117 as
f¯(r) = f0 (ρ(r)) − f1 (ρ(r))∇2 ρ(r) − f2 (ρ(r))(∇ρ(r))2 (118)
Where,
Z Z
ρ(r)
f1 (ρ(r)) = d3 r2 r22 C¯0 (r2 , ρ) + ρ d3 r3 C 0 (r2¯, r3 , ρ) (119)
12
18
and
Z Z
ρ(r) ρ
f2 (ρ(r)) = d32 r d r3 r2 .r3 {C¯0 (r2 , r3 , ρ) +
3
d3 r4 C¯0 (r2 , r3 , r4 , ρ)}
6 2
(120)
With the help of the symmetry properties of correlation functions, f2 (ρ) can be
written as
Z
ρ
f2 (ρ) = d3 r1 d3 r2 d3 r3 (r2 − r1 ).(r3 − r1 )
6V Z
ρ
∗{C̄(r1 , r2 , r3 , ρ) + d3 r4 C̄(r1 , r2 , r3 , r4 , ρ)} (121)
Z 2
ρ
= d3 r1 d3 r2 d3 r3 (r12 − r1 r2 )
6V Z
ρ
∗{C̄(r1 , r2 , r3 , ρ) + d3 r4 C̄(r1 , r2 , r3 , r4 , ρ)} (122)
Z 2
ρ
= d3 r1 d3 r2 d3 r3 (r1 − r2 )2
12V Z
ρ
∗{C̄(r1 , r2 , r3 , ρ) + d3 r4 C̄(r1 , r2 , r3 , r4 , ρ)} (123)
Z 2
ρ
= d3 r2 d3 r3 r22
12V Z
ρ
∗{C¯0 (r2 , r3 , ρ) + d3 r4 C¯0 (r2 , r3 , r4 , ρ)} (124)
2
Calculation for symmetricity in equation 121 to 124
let,
Z
I = d3 r1 d3 r2 d3 r3 (r2 − r1 ).(r3 − r1 )
Z
ρ
∗{C̄(r1 , r2 , r3 , ρ) + d3 r4 C̄(r1 , r2 , r3 , r4 , ρ)} (125)
2
(126)

Now,

(r2 − r1 ).(r3 − r1 ) (127)

= r12 − r1 r2 − r1 r3 + r2 r3 (128)
(129)

interchange r1 → r2 and r2 → r1 equation 109 becomes

r22 − r2 r1 − r2 r3 + r2 r3 (130)

As the correlation functions are symmetric, so interchange r1 → r2 and r2 → r1

will not affect the integral I.

19
 So, from this argument we can say
Z
2I = d3 r1 d3 r2 d3 r3 (r12 + r22 − 2r1 r2 )
Z
ρ
∗{C̄(r1 , r2 , r3 , ρ) + d3 r4 C̄(r1 , r2 , r3 , r4 , ρ)} (131)
Z 2
= [ d3 r1 d3 r2 d3 r3 (r12 − 2r1 r2 )
Z
ρ
∗{C̄(r1 , r2 , r3 , ρ) + d3 r4 C̄(r1 , r2 , r3 , r4 , ρ)}]
Z 2
+[ d3 r1 d3 r2 d3 r3 (r22 )
Z
ρ
∗{C̄(r1 , r2 , r3 , ρ) + d3 r4 C̄(r1 , r2 , r3 , r4 , ρ)}] (132)
2
Again using the symmetric properties we can write
Z Z
ρ
I= d3 r1 d3 r2 d3 r3 (r12 − 2r1 r2 ) ∗ {C̄(r1 , r2 , r3 , ρ) + d3 r4 C̄(r1 , r2 , r3 , r4 , ρ)}
2
(133)
putting equation 133 in equation 121 we get equation 122.
Again
r12 − r1 r2 (134)
can be written as
r22 − r2 r1 (135)
Adding equations 134 and 135 we get

(r1 − r2 )2 (136)

and by this we get equation 123

Now,
We can write the equation 118 as
1
f¯(r) = f0 (ρ(r)) + A(ρ(r))(∇ρ(r))2 − ∇.{f1 (ρ(r))∇ρ(r)} (137)
2
Where,
1
A(ρ) = f10 (ρ) − f2 (ρ) (138)
2
From equation 83 we can see that it is a volume integral. So the divergence
term in equation 136 become the surface integral(using Gauss divergence theorem)
where ∇ρ = 0 for physical solution. Hence the divergence will not contribute.
Putting the value of f¯(r) from equation 136 in equation 83 we get
Z
1
F0 = d3 r[f0 (ρ(r)) + A(ρ(r))(∇ρ(r))2 ] (139)
2
20
Again putting the value of F 0 from equation 139 in equation 45 we get
Z
1
βF = d3 r[f0 (ρ(r)) + A(ρ(r))(∇ρ(r))2 + βΦ(r)ρ(r)] (140)
2
So, equation 140 can put in the form of equation 2 by replacing Φ(r) = U (r),
f0 (ρ(r)) = ψ̃(ρ(r)) and A(ρ) = βA(ρ)
So,

ψ(ρ(r), ∇ρ(r), r) = U (r)ρ(r) + ψ̃(ρ(r)) + 12A(ρ(r))(∇ρ(r))2 (141)

∂ C¯0 (r2 , ρ)
Z
= d3 r3 C¯0 (r2 , r3 , ρ) (142)
∂ρ
and
∂ C¯0 (r2 , r3 , ρ)
Z
= d3 r4 C¯0 (r2 , r3 , r4 , ρ) (143)
∂ρ
Substituting equation 142 and 143 for in equation 119, 120 and 138
ρ2 ∂ 2 ¯0
Z
1 ∂
A(ρ) = d3 r2 r22 (1 + 2ρ + )C (r2 , ρ)
6 ∂ρ 2 ∂ρ2
1 ∂2 2
Z
= 2
ρ d3 r2 r22 C¯0 (r2 , ρ) (144)
12 ∂ρ
Now,
1 1
1 ∂ 2 2 ¯0 ∂2 2 0
Z Z
{ρ C (r2 , ρ)} = sds ds 2 {ρ C (r2 , ss0 ρ)}
0
2 ∂ρ2 ∂ρ
Z0 1 0
Z 1
1 d 1 d
= ds ds0 0 {ss0 ρ2 C 0 (r2 , ss0 ρ)}
0 ρ ds ρ 0 ds
0
= C (r2 , ρ) (145)

Now we can write from equation 144 and 145

Z
1
A(ρ) = d3 r2 r22 C 0 (r2 , ρ) (146)
6

3.2 numerical analysis

One can study the nature of the surface tension for a particular potential. I have
consider the hard sphere type potential for which I have the numerical values as
below.
Programming
IMPLICIT NONE
Double Precesion PI, RHO,R,C,A,B,ETAA,CPY

21
INTEGER I
PI = 3.1415926
RHO = 1.1200
ETAA = RHO*PI/6D0
DO I = 0, 10000
R = 0.0002D0*FLOAT(I)
C = (1.0D0 - ETAA)*(1.0D0 - ETAA)*(1.0D0 - ETAA)*(1.0D0 - ETAA)
A = -(1.0D0 + 2*ETAA)*(1.0D0 + 2*)/C
B = 6*ETAA*(1.0D0 + 0.5D0*ETAA)*(1.0D0 + 0.5D0*ETAA)/C
IF(R.LT.1D0) THEN
CPY = A + B*R + 0.5D0*ETAA*A*R*R*R
ELSE
CPY = 0D0
END IF
PRINT*,R, CPY
WRITE(10,*)R, CPY
ENDDO
END

With the value of correlation function C one can calculate A(ρ(r)) and solving
the differential equaltion withe the boundary condition one can find the density
profile. And with that value of A(ρ(r)) and ρ(r) one can plot the surfacetension
w.r.t r.

22