The second virial coefficient and critical point behavior of the Mie Potential

D. M. Heyes, G. Rickayzen, S. Pieprzyk, and A. C. Brańka

Citation: The Journal of Chemical Physics 145, 084505 (2016); doi: 10.1063/1.4961653
View online: http://dx.doi.org/10.1063/1.4961653
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/145/8?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in

Second virial coefficient of a generalized Lennard-Jones potential
J. Chem. Phys. 142, 034305 (2015); 10.1063/1.4905663

Second virial coefficient for the dipolar hard sphere fluid

J. Chem. Phys. 135, 044514 (2011); 10.1063/1.3615723

Viscosity, Second p V T -Virial Coefficient, and Diffusion of Pure and Mixed Small Alkanes C H 4 , C 2 H 6 , C
3 H 8 , n - C 4 H 10 , i - C 4 H 10 , n - C 5 H 12 , i - C 5 H 12 , and C ( C H 3 ) 4 Calculated by Means of an
Isotropic Temperature-Dependent Potential. I. Pure Alkanes
J. Phys. Chem. Ref. Data 35, 1331 (2006); 10.1063/1.2201308

A Monte Carlo study of effects of chain stiffness and chain ends on dilute solution behavior of polymers. II.
Second virial coefficient
J. Chem. Phys. 119, 1257 (2003); 10.1063/1.1579682

Predicting the gas–liquid critical point from the second virial coefficient
J. Chem. Phys. 112, 5364 (2000); 10.1063/1.481106

Reuse of AIP Publishing content is subject to the terms: https://publishing.aip.org/authors/rights-and-permissions. Downloaded to IP: 200.62.146.210 On: Wed, 31
Aug 2016 20:59:32
 THE JOURNAL OF CHEMICAL PHYSICS 145, 084505 (2016)

The second virial coefficient and critical point behavior of the Mie Potential
D. M. Heyes,1,a) G. Rickayzen,2,b) S. Pieprzyk,3,c) and A. C. Brańka3,d)
1
The Physics Department, Royal Holloway, University of London, Egham, Surrey TW20 0EX, United Kingdom
2
School of Physical Sciences, University of Kent, Canterbury, Kent CT2 7NH, United Kingdom
3
Institute of Molecular Physics, Polish Academy of Sciences, M. Smoluchowskiego 17, 60-179 Poznań, Poland
(Received 4 July 2016; accepted 15 August 2016; published online 30 August 2016)

Aspects of the second virial coefficient, b2, of the Mie m : n potential are investigated. The Boyle
temperature, T0, is shown to decay monotonically with increasing m and n, while the maximum
temperature, Tma x , exhibits a minimum at a value of m which increases as n increases. For the 2n : n
special case T0 tends to zero and Tma x approaches the value of 7.81 in the n → ∞ limit which is
in quantitative agreement with the expressions derived in Rickayzen and Heyes [J. Chem. Phys. 126,
114504 (2007)] in which it was shown that the 2n : n potential in the n → ∞ limit approaches Baxter’s
sticky-sphere model. The same approach is used to estimate the n−dependent critical temperature of
the 2n : n potential in the large n limit. The ratio of T0 to the critical temperature tends to unity
in the infinite n limit for the 2n : n potential. The rate of convergence of expansions of b2 about
the high temperature limit is investigated, and they are shown to converge rapidly even at quite
low temperatures (e.g., 0.05). In contrast, a low temperature expansion of the Lennard-Jones 12 : 6
potential is shown to be an asymptotic series. Two formulas that resolve b2 into its repulsive and
attractive terms are derived. The convergence at high temperature of the Lennard-Jones b2 to the
m = 12 inverse power value is slow (e.g., requiring T ≃ 104 just to attain two significant figure
accuracy). The behavior of b2 of the ∞ : n and the Sutherland potential special case, n = 6, is
explored. By fitting to the exact b2 values, a semiempirical formula is derived for the temperature
dependence of b2 of the Lennard-Jones potential which has the correct high and low temperature
limits. Published by AIP Publishing. [http://dx.doi.org/10.1063/1.4961653]


I. INTRODUCTION π D/2 ∞
b2(D, β) = − (e−φ(r )β − 1)r D−1dr, (2)
Γ(D/2) 0
The second virial coefficient, b2, is the starting point
of many analytic expressions for the equation of state of which in 3D reduces to5–8
molecular systems.1 Many different types of small molecules  ∞
have had their b2 calculated over certain temperature ranges. b2 = −2π (e−φ(r )β − 1)r 2dr. (3)
It can be measured experimentally, and by inversion of 0
its temperature dependence aspects of the pair potential An alternative form based on the virial expression is6,9
between molecule pairs can be ascertained. More recently
it has been proposed as a useful indicator of the liquid-  ∞
2π
vapor critical point, especially in the context of colloidal b2 = − β φ ′(r)e−φ(r )βr 3dr. (4)
3 0
liquids.2–4
The second virial coefficient is defined through the density The advantage of Eq. (4) is that (unlike Eq. (3)) the repulsive
expansion of the compressibility factor, Z, and attractive parts of b2 can be computed directly from
it. Although the second virial coefficient formulas given in
P
Z≡ = 1 + b2 ρ + O(ρ2), (1) Eqs. (3) and (4) can be evaluated by numerical integration, it
ρk BT is useful to have an explicit analytic expression for b2 from
where k B is the Boltzmann’s constant and ρ is the molecule which generic trends can be more readily deduced.
number density. If ϵ is a characteristic energy of the pair The Mie m : n potential is of main interest here,
potential, the reduced temperature, T ∗ = k BT/ϵ ≡ β −1. As ( σ ) m ( σ ) n
model interaction potentials are considered here, k B can be φ(r) = λϵ − , m > n > 3, (5)
conveniently set to unity, and the “∗” superscript omitted. r r
In D dimensions the Mayer function definition of the where
second virial coefficient of a central potential is5
( m ) ( m ) n/(m−n)
λ= . (6)
a)Electronic address: david.heyes@rhul.ac.uk m−n n
b)Electronic address: gerald.rickayzen@physics.org
c)Electronic address: slawomir.pieprzyk@ifmpan.poznan.pl The widely used Lennard-Jones (LJ) potential is a special case
d)Electronic address: branka@ifmpan.poznan.pl of Eq. (5),

0021-9606/2016/145(8)/084505/11/$30.00 145, 084505-1 Published by AIP Publishing.

Reuse of AIP Publishing content is subject to the terms: https://publishing.aip.org/authors/rights-and-permissions. Downloaded to IP: 200.62.146.210 On: Wed, 31
Aug 2016 20:59:32
 084505-2 Heyes et al. J. Chem. Phys. 145, 084505 (2016)

σ ) 12 ( σ ) 6
( 
which reproduces the second virial coefficient of the Lennard-
φ(r) = 4ϵ − . (7)
r r Jones potential is given in Sec. II G. Conclusions are made in
Sec. III.
This particular potential is important because the second
term describes the van der Waals force of attraction between
molecules. The r −12 term in the potential represents the finite II. THEORY
repulsion at small finite distances. The LJ parameters ϵ and σ
have been determined for many real gases.10 The higher virial In this section, various expressions for the Mie potential
coefficients of the LJ potential have also been computed second virial coefficient are derived, and their characteristics
as a function of temperature by Mayer sampling Monte examined.
Carlo.11
In this work, an extensive study on the properties of the A. High temperature series expansion
second virial coefficient of the generic m : n Mie potential is
carried out, which includes as special cases the generalized First a series expansion expression for b2 is derived
Lennard-Jones potential (i.e., m = 2n) and the original LJ starting from the Mie potential definition in Eq. (5). The
potential (m = 2n = 12). In particular, the behavior of b2 in following reduced variables are employed:
the low-temperature and high-temperature limits is studied r k BT T∗
in great detail, as well as the Boyle temperature and the x= , T∗ = , α= . (8)
σ ϵ λ
temperature where b2 has a maximum.
Closed expression formulas for the LJ b2 in terms of Then from Eq. (3),
Bessel and parabolic cylinder functions have been derived  ∞   ( ) 
b2 1 1 1
(see, for example, Refs. 5, 6, 12, and 13) which are discussed =− 2
x dx exp − − − 1 . (9)
2πσ 3 α xm xn
in Sec. II B. Lennard-Jones,14 and later others,8,9,15 derived 0

an infinite power series expansion representation of the m : n Because it could lead to a divergence at the origin, one cannot
potential b2 in terms of the gamma function, which is derived simply expand the exponent in inverse powers of α which is
again here by a more compact route, and reveals that the first proportional to T ∗, which is replaced by “T” for the rest of
term in the series expansion (which is the high temperature this work. Instead the variable of integration is changed to y,
limit) is the inverse power (IP) potential ∼r −m second virial where
coefficient. Also expressions for the high and low temperature
1 dx 1 dy
limit of b2 are derived directly from Eqs. (3) and (4) (in y= , x = (α y)−1/m , =− . (10)
the latter case for a continuous potential of arbitrary analytic αx m x m y
form). Then,
The convergence characteristics of the series representa-
tions of b2 in the high (HL) and low temperature (LL) limits 1
= (α y)n/m (11)
are explored in Secs. II A and II C, respectively. xn
General trends of the Mie potential second virial and
coefficient characteristics are discussed, especially in relation 
b2 α −3/m ∞ −3/m d y
to the Boyle and maximum temperature in Sec. II D. The = − y
large n behavior of the second virial coefficient and associated 2πσ 3 m 0 y
   
characteristic temperatures are quantitatively reproduced × exp −( y − α y
(n−m)/m n/m
) −1 . (12)
using a treatment which for n → ∞ reduces to Baxter’s sticky
sphere model, as discussed in Sec. II E. Exact decompositions When the temperature is high or the coupling is weak, the
of b2 into the contributions from the repulsive and attractive integrand may be expanded in powers of T (n−m)/m which is
parts of the potential are derived in Sec. II F. A fit formula small. This expansion leads to


α −3/m ∞
  1  
b2

 
= y (−3−m)/m
dy  exp(− y) − 1 + exp(− y) y
(n−m)/m n/m j  

−   (α )  
2πσ 3 m 0  j=1
j! 
 
 
 ∞ 
α −3/m 1 (n−m) j/m ∞

  

=− y (−3−m)/m
y y) + α y (−3−m+n j)/m exp(− y) d y 
 
 d [exp(− − 1]
m 
 0 j=1
j! 0


 
 ∞
α −3/m 1 (n−m) j/m −3 + n j 

  ( )
=− y (−3−m)/m d y [exp(− y) − 1] + α .
 
 Γ (13)
m 
 0 j=1
j! m 

 

Reuse of AIP Publishing content is subject to the terms: https://publishing.aip.org/authors/rights-and-permissions. Downloaded to IP: 200.62.146.210 On: Wed, 31
Aug 2016 20:59:32
 084505-3 Heyes et al. J. Chem. Phys. 145, 084505 (2016)

( ) 3/m (
The first term in Eq. (13) can be evaluated using integration λ
)
b2, I P (T) 3
by parts to give = Γ 1− . (20)
b0 T m

α −3/m ∞ (−3−m)/m Setting m = 12, Eq. (20) gives
− y d y [exp(− y) − 1]
m ( ) 1/4 ( )
0
 √
( )
b2 1 4 1 3
α −3/m ∞ −3/m =− Γ − = 2T −1/4Γ , (21)
= y dy exp(− y) b0 4 T 4 4
3 0
which is the first term in Eq. (15) for m = 12. It is also the
α −3/m
( )
3
= Γ 1− (14) high temperature limit of the 12 : 6 potential second virial
3 m coefficient, as may be seen by inspection.
and from Eq. (13), The Sutherland potential (SP), φ(r) = +∞, r ≤ σ;
= −ϵ(σ/r)6, r > σ,16 is a limiting case of the Mie potential
α −3/m
( )
b2 3 which has attracted some interest as the hard core would lead
= Γ 1 −
2πσ 3 3 m one to think that it would form the basis of the van der Waals
α −3/m 
(
1 (n−m) j/m n j − 3
) equation of state. To obtain b2 of the SP, the limit of m → ∞ is
− α Γ . (15) taken in Eq. (15) for the m : n potential, keeping λ fixed. This
m j=1 j! m
reduces to taking the limit Γ[(n j − 3)/m]/m, for m → ∞. To
determine this note that
When m = 2n and λ = 4 then ( ) ( ) ( )
nj − 3 nj − 3 nj − 3
α −3/2n α −3/2n Γ 1+ = .
( )
b2 3 Γ (22)
3
= Γ 1 − − m m m
2πσ 3 2n 2n
 1 (
nj − 3
) Hence,
× α − j/2 Γ ( ) ( ) ( )
j! 2n nj − 3 nj − 3 nj − 3
j=1 Γ =Γ 1+ /
( ) m m m
1  1 −(n j+3)/2n nj − 3
=− α ,
( )
Γ (16) nj − 3
2n j=0 j! 2n = mΓ 1 + / (n j − 3) (23)
m
using Γ(1 − x) = −xΓ(−x). For the LJ potential, n = 6 and and
( ) ( )
1 nj − 3 nj − 3
= lim Γ 1 + / (n j − 3)
( ) (1+2 j)/4 (
lim Γ
)
b2 1  1 4 2j − 1
=− Γ , (17) m→ ∞ m m m→ ∞ m
2πσ 3 12 j=0 j! T 4
1
= . (24)
(n j − 3)
which if b0 = 2πσ 3/3 may be written as
Therefore the m → ∞ limit of Eq. (15) using Eq. (24) is for
( ) (2 j+1)/4 ( ) the ∞ : n potential,
b2 3b2 1 1 4 2j − 1
= = − Γ . (18)
b0 2πσ 3 4 j=0 j! T 4 b2  α− j
= − , (25)
2πσ 3 j=0
j!(n j − 3)
See the supplementary material for a list of the first 40 terms
of the expansion. which for the SP or n = 6 special case is
Normalization of b2 by b0 is informative as in certain
Mie exponent and temperature regions b2 approaches the b2  α− j
=− , (26)
second virial coefficient of the hard sphere potential which is b0 j=0
j!(2 j − 1)
b0 for a hard sphere diameter, σ. The original derivation of
these formulas by Lennard-Jones7,14 was in terms of inverse which is an expression obtained elsewhere.17
distance, x −1. The derivation of Eqs. (15) and (16) for the
general Mie potential, and Eq. (18) for the LJ special case, B. Closed expressions
is made much shorter by expressing the distance, x, in terms
of the composite parameter, y, which is used instead as the A closed exact analytical solution for the LJ second virial
expansion parameter. The derivation above also shows that the coefficient has been derived by Eu,18
first ( j = 0) term in the expansion which emerges separately √
π 2x 1/4σ 3
( )
3
is the contribution from the m or IP part of the potential. This b2 = − 4Γ
6 4
term dominates at high temperature.  ( ) ( )
For the IP potential, written in the form λϵ(σ/r)m in 7 3 3 1
× 6x M , , x − (1 + 4x) M , , x
dimension D (see the supplementary material), 4 2 4 2
√ 3/4 3 ( )
π 2x σ 1
( ) D/m D (
π D/2 λ σ D
) − 2Γ
b2, I P (D,T) = Γ 1− . (19) 6 4
Γ(D/2) T D m 
10
(
9 5
) (
5 3
)
× x M , , x + (1 − 4x) M , , x , (27)
In three dimensions and using Γ(3/2) = π 1/2/2, 3 4 2 4 2

Reuse of AIP Publishing content is subject to the terms: https://publishing.aip.org/authors/rights-and-permissions. Downloaded to IP: 200.62.146.210 On: Wed, 31
Aug 2016 20:59:32
 084505-4 Heyes et al. J. Chem. Phys. 145, 084505 (2016)

where x = βϵ and any dimension and any n) had been published previously as
∞
Eq. (7.87) in Ref. 19 and more recently in Ref. 20.
 (a)n x n It is straightforward to show from the recurrence relations
M(a, b, x) = (28)
n=0
(b)n n! of the Kummer functions given in Eqs. (13.4.1) and (13.4.4)
of Ref. 21 that the formulas in Eqs. (27) and (29) are identical.
is Kummer’s confluent hypergeometric function and (a)n
The asymptotic form of Kummer’s function allows us to obtain
= Γ(a + n)/Γ(a) are the Pochhammer symbols. Another exact
directly from Eq. (29) a high temperature limiting series for
closed-form formula for b2 of the LJ fluid has recently been
the second virial coefficient,13
derived by González-Calderón and Rocha-Ichante,13 √ 1
2 2πσ 3 1 −1/4  n 2n − 1
 
πx 1/4σ 3 b2(T → ∞) = −
( ) ( )
1 1 1 T 2 Γ
b2 = − √ Γ − M − , ,x 3 4 4
3 2 4 4 2 n=0
√ ∞
2πx σ3/4 3
( )
1
(
1 3
)  ((2n − 1)/4)i T −i−n/2
− Γ M , ,x . (29) × . (30)
3 4 4 2 i=0
((2n + 1)/2)i i!
Another closed expression for b2 in even more compact form It can be proved that the series for b2 in the high temperature
is presented as Eq. (14) in Ref. 13 in terms of the parabolic limit defined in Eqs. (18) and (30) are identical. The first five
cylinder function. Such a compact expression (generalized to terms of Eq. (30) are

√
Γ 41 Γ 14



2 2πσ 3 3 −1/4 *
( )
1 1
b2(T → ∞) = Γ T 1− 3

1/2 − − 3

3/2 − 2 · · · + . (31)
3 4 , 2Γ 4 T 2T 12Γ 4 T 8T -

Figure 1 shows the convergence of the b2 expansion of Eq. (18) virial coefficient of any of them can be found in the following
or Eq. (30) for several temperatures in the range 0.1–100. As way. Consider such a potential, φ(r), which is defined in terms
T decreases more terms are needed to produce the same of a suitable length, σ, and energy, ϵ. Let x = r/σ and k BT/ϵ
fractional deviation from the exact result. Four or five terms be replaced by “T” again. σ and ϵ are chosen so that the
are sufficient to obtain b2 with an accuracy better than 0.0001 minimum of φ(x) is −1. The potential has a single minimum.
for T > 10. The high temperature limit of b2 converges quite The definition of b2 in Eq. (4) can be rewritten in terms of
quickly and can be used as an almost exact representation of these dimensionless variables,
b2, even at very low T. For example, an accuracy of 10−6 is 
1 ∞ 3 ′ φ(x)
( )
b2
obtained with only 20 terms for T as low as 0.5. B2 = =− x φ (x) exp − dx. (32)
b0 T 0 T
The low temperature limit of b2 for an arbitrary continuous
potential with a minimum is considered below. Now, when T is small the integrand is dominated by the
exponential near its maximum at, say, x 0, where

C. Low temperature expansion φ ′(x 0) = 0, (33)

The Mie potential is repulsive when the molecules are and it falls away rapidly as x moves away from this maximum.
close, attractive when they are far apart, and has a minimum Hence, φ(x) − φ(x 0) is positive for all x and can be written
in between. There are many potentials with this general with s a real number,
behavior, and the low temperature dependence of the second s2 = φ(x) − φ(x 0),
s = ± φ(x) − φ(x 0).

(34)
Now change the integration variable to s. Equation (34)
implicitly defines x as a function of s. The ambiguity in the
sign in Eq. (34) is removed by choosing
s = sign(x − x 0) φ(x) − φ(x 0).

(35)
Then s increases as x increases near x 0 and
)  s∞
φ(x 0)
( ( 2)
1 s
B2 = − exp − χ(s) exp − ds, (36)
T T s0 T
where
dx dφ(x(s))
χ(s) = x 3φ ′(x(s)) = x3 . (37)
ds ds
The limits of the integration are
FIG. 1. The convergence of b 2 expansions from Eq. (18) or Eq. (30). Several
s0 = − φ(0) − φ(x 0), s∞ = −φ(x 0).
 
temperatures in the range 0.1–100 indicated on the figure are considered. (38)

Reuse of AIP Publishing content is subject to the terms: https://publishing.aip.org/authors/rights-and-permissions. Downloaded to IP: 200.62.146.210 On: Wed, 31
Aug 2016 20:59:32
 084505-5 Heyes et al. J. Chem. Phys. 145, 084505 (2016)

Because the exponential falls away rapidly as s increases from Hence, since the potential has been normalized so that its
zero, the limits of the integral can be extended to ±∞ and the minimum value is −1,
dominant behavior of B2 as T → 0 is given by 1
) ∞ m ( m ) m−n (m − n)
1
(
φ(x 0)
( 2)
s x 0m−n = , x 0 = , φ(x 0) = −α = −1 (45)
B2 = − exp − χ(s) exp − ds. (39) n n mx 0n
T T −∞ T
and
(For potentials which are strongly repulsive as r → 0 such
n(m − n) mn
as the LJ one, the lower limit is −∞ anyway.) For the same φ ′′(x 0) = α = 2. (46)
reason we can expand χ(s) in powers of s and keep only the x 0n+2 x0
leading non-zero term. Differentiating Eq. (34) with respect Thus
to s, ( m ) 3/(m−n) 2T π ( )
b2 1
dφ(x) B2 = = −3 exp . (47)
= 2s (40) b0 n mn T
ds
If we specialize further by choosing m = 2n,
is obtained, so this term is already first order in s. Because
√
( )
the exponential is an even function of s, a constant term in b2 3 1
= − 23/n πT exp (48)
the other factors in the integral yields zero for the integral. b0 n T
To obtain the leading term in B2, we need to expand the in the low temperature limit. If the potential contains several
remaining factor to first order in s. Let ∆x = x − x 0, so that minima, all will contribute to the virial but the dominant
near the minimum behavior at low T is given by the global minimum. As the
 12 
reduced temperature is inversely proportional to the coupling
φ ′′(x 0)

1 2 ′′
s ≈ sign(∆x) ∆x φ (x 0) = ∆x . (41) constant, the same formula applies when the interaction
2 2
strength (i.e., ϵ) is extremely strong.
Then, Including the next term in the series expansion for the
case of m = 2n,

2
x 3 = (x 0 + ∆x)3 ≈ x 30 + 3x 20∆x = x 30 + 3x 20 s . (42) b2 3 1 √
( ) 
(3 + n)(3 + 2n)

φ ′′(x 0) = − 23/n exp πT 1 + T + O(T 2
) ,
b0 n T 4n2
Combining Eqs. (39)-(42), the following result is obtained at
(49)
low temperatures:
 ) ∞ so for the LJ potential (n = 6),
φ(x 0)
( ( 2)
6 2 2 s
B2 = − x 0 exp − 2
s exp − ds ( ) 12
φ ′′(x 0) πT
( ) 
T T −∞ T b2 1 15
( ) ∞ =− exp 1+ T ··· , (50)
b0 2 T 16

2T 1
= −6x 02
t 2 exp −t 2 dt


exp as the limiting form of b2 as T → 0, which is the same
φ (x 0)
′′ T −∞
 formula obtained by Cachadiña,22 and in Ref. 13 by expansion
2T π of Eq. (29). See the supplementary material for a list of
( )
1
= −3x 20 exp . (43) the first 16 terms of the expansion. The advantage of the
φ (x 0)
′′ T
above derivation leading to Eq. (47) is that it is more
Equation (43) has been previously derived for arbitrary general in applying to any reasonable continuous potential,
dimension by another route, given in Eq. (3.62) in Ref. 20. φ, rather than just the Mie potential or its LJ special
In the specific case of the m : n potentials, case.
( )
1 1 From Eq. (29) the low temperature limit is13
φ(x) = α m − n , m > n > 3,
x x √ ∞ ( ) ( )
2ππσ 3 1 √  5 3 Ti
( )
b2(T → 0) = − . (51)
( m n ) exp T
φ ′(x) = α − m+1 + n+1 , (44) 3 T 4 i 4 i i!
( x x i=0
m(m + 1) n(n + 1)
)
φ ′′(x) = α − . The first few terms in the low temperature limit expansion
x m+2 x n+2 are13

√
2ππσ 3 1 √
( ) ( )
15 945 2 45 045 3 11 486 475 4 916 620 705 5
b2(T → 0) = − exp T 1+ T + T + T + T + T . (52)
3 T 16 512 8192 524 288 8 388 608

Figure 2 shows the low T convergence of the second virial are shown. Up to 14 terms the series expansion approaches
coefficient from Eq. (51) denoted by “LL”, measured by the closer to the exact value (shown by the long arrow pointing
relative deviation from the exact result, as a function of T −1 to the left) but adding more terms (16–20) the same accuracy
for low T. Data for i terms in the series ranging from 0 to 20 is achieved only for smaller T (the temperature range shrinks

Reuse of AIP Publishing content is subject to the terms: https://publishing.aip.org/authors/rights-and-permissions. Downloaded to IP: 200.62.146.210 On: Wed, 31
Aug 2016 20:59:32
 084505-6 Heyes et al.
FIG. 2. The convergence of b 2 in the low temperature regime using Eq. (51).
The index, i, means the number of terms used in the low temperature
expansion of Eq. (51). The arrows indicate the first normal or expected
converging trend (arrow pointing to the left) and the shorter right pointing
arrow highlights the diverging behavior with increasing index.
which is indicated by the short arrow pointing to the right). The
figure illustrates that Eq. (51) is an asymptotic series. The LL
expansion does not converge for any T. By adding successive
terms, i = 0, 1, 2, 3 to i = k, a more accurate representation
of the function is seen but then the addition of further terms
beyond a particular value of k leads to a decrease in the
accuracy of the series representation. Figure 2 shows that
k = 14 is required for an accuracy of 10−6 for T < 0.05. In
general, the value of k depends on T and the accuracy required,
but we could not obtain an accuracy of ≤10−6 for T > 0.05
from the LL series with any k, and for k = 0 (the leading term)
an accuracy of 10−5 was only achieved for T < 0.000 01.
D. Second virial coefficient limits, Boyle
and maximum temperatures
The temperature dependence of the second virial
coefficient using several values of n for the 2n : n special
case of the Mie potential is given in Fig. 3. This figure
illustrates the general shape of b2(T; m : n) in that it has a
maximum at T = Tmax, goes through zero once, and has two
limits, one which is positive for T → ∞ (from the dominance
of the repulsive part of the potential) and −∞ for T → 0
(representing the relative importance of the attractive part of
FIG. 3. The second virial coefficient, b 2, calculated using Eq. (3) for the Mie
2n : n potential. Representative n values are given on the figure. The orange
dotted line is the locus of the maximum value in b 2 (i.e., Tmax values). The
open circle represents the limiting value Tmax(n → ∞) = 7.8098.
Reuse of AIP Publishing content is subject to the terms: https://publishing.aip.org/authors/rights-and-permissions. Downloaded to IP: 200.62.146.210 On: Wed, 31
Aug 2016 20:59:32
J. Chem. Phys. 145, 084505 (2016)
FIG. 4. The m and n dependence of the Boyle temperature, T0, for the
Mie m : n potential defined in Eq. (5). The lines from top to bottom are
n = 4, 5, 6, 7, 8, 9, 11, 15, 25, and 35. The inset shows a finer resolution of the
data in the large m region in the bottom right hand corner of the main figure.
the potential). The Boyle temperature, T0 (i.e., b2(T0) = 0),
decreases on increasing n. The orange dashed line and points
on the figure trace out the Tmax line. The figure shows that as
n increases the hard-sphere limit of b2 → b0 is approached for
an increasingly wide temperature range, for not too small T.
Figure 4 shows the dependence of the Boyle temperature,
T0, as a function of the Mie potential m and n exponents
calculated using Eq. (3). The figure reveals that T0 decreases
monotonically with increasing m and n. For the LJ potential,
T0 = 3.418. In contrast, the maximum temperature (i.e.,
db2/dT = 0) exhibits a minimum at a particular value of
m for each n, as shown in Fig. 5. The figure shows that the
value of m at the minimum increases as n increases. For the
LJ potential Tmax = 25.153.
Some of the behavior of the second virial coefficient of
the 2n : n potential case in the large n limit is considered in
Sec. II E.
E. Hard sphere and sticky hard sphere models
Equation (4) can be written in terms of the reduced
distance, x, as

b2 1 ∞ 3 dφ −φ(x)/T
=− x e dx. (53)
b0 T 0 dx
FIG. 5. As for Fig. 4 except the maximum temperature, Tmax, is shown. The
orange squares and dashed line trace out the m value where the minimum
value of Tmax occurs, for a given value of n.
 084505-7 Heyes et al. J. Chem. Phys. 145, 084505 (2016)

For the 2n : n potential, φ(x) = 4(z 2 − z), where x = z −1/n and

hence,

b2 1 ∞ −3/n dφ −φ(z)/T
= z dz e
b0 T 0 dz
 ∞
4
= dz exp(−[3/n] ln z)(2z − 1)e−φ(z)/T
T 0

4 ∞
( )
3 9
= dz 1 − ln z + 2 (ln z) − · · ·
2
T 0 n 2n
× (2z − 1)e−φ(z)/T . (54)
Therefore,
b2 3 9
= 1 − A + 2 B, (55)
b0 n 2n
on taking the n → ∞ limit for the first term in the last line of
Eq. (54). The constants are

4 ∞
A= dz ln(z)(2z − 1)e−φ(z)/T (56)
T 0
and

4 ∞
B= dz (ln(z))2(2z − 1)e−φ(z)/T . (57)
T 0
The temperature dependent quantity, A, does not depend on n,
has one minimum at T = 7.8098, and diverges to +∞ as T → 0.
It was shown in Ref. 23 that the 2n : n potential goes over to
the Baxter sticky sphere model,13,16,24 when n becomes very
FIG. 6. The top frame (a) shows the Boyle temperature as a function of n
large; the relationship between Baxter’s adhesive parameter, obtained by numerical integration from Eq. (3) (open circles or “D1”) and
τ, and the parameters of the 2n : n potential was shown in using Eq. (16) (crosses or “D2”) for the Mie 2n : n potential. The orange
Ref. 23 to be 1/12τ = A/n. The second virial coefficient of asterisks are for the ∞ : n or “IN” potential obtained by the summation in
Eq. (25). The diamond in the top left of the main figure is T0 of the ∞ : 6 or
Baxter’s model is b2/b0 = 1 − 1/4τ and using Eq. (2.10) in Sutherland (“SP”) potential special case obtained from an alternative closed
Ref. 23, form expression formula derived by Levi and Llano.38 The blue line in
b2 3 the main figure is a fit to guide the eye. The inset focuses on the large n
= 1 − A, (58) behavior, with the symbols referring to the same quantities as in the main
b0 n figure. The red points are the Baxter adhesive sphere model (“BM”) for T0
as in the relevant ρ → 0 limit the cavity function has specific from A(T0) = n/3, where A is defined in Eq. (56) and obtained by numerical
integration. The red line is a fit to guide the eye. The bottom frame (b) gives
values, y(σ) = 1 and y ′(σ) = 0 for large n. Equation (58) is the corresponding Tmax values. In the inset the magenta line is a fit (“fit”) to
just the first two terms on the right hand side of Eq. (55). the numerical data with the form c 0 + c 1 n −1 + 163.2667n −2 for n > 40, where
The high and low temperature limits of b2 for the 2n : n c 0 = 7.8098 = Tmax,0 and c 1 = 37.1798 (see the main text).
potential for large n are therefore determined by high and low
temperature limits of the variables, A and B. We use the term
7.8098 (=Tmax,0). The behavior of b2 in this large n region is
“Baxter model” (BM) to refer to Eqs. (55) and (58).
well captured by equating the temperature derivative of the
We know by letting n → ∞ in Eq. (48) and compar-
right hand side of Eq. (55) to 0, which gives
ing with Eq. (55) √ that in the low temperature limit,
A(T) = n/12τ = πT e1/T , a relation which gives a specific dA 3 dB
= , (59)
temperature dependence to Baxter’s parameter τ in the low dT 2n dT
temperature limit. which can be solved numerically for Tmax(n). When n is large,
From the basic definition, b2(T0) = 0, the relation, we can assume that its effect on Tmax is to produce a small
Eq. (58), gives A(T0) = n/3, so T0 → 0 as n → ∞. The change, Tmax = Tmax,0 + δ. Then Eq. (59) becomes
critical temperature, Tc , of Baxter’s model is τ = 0.1133
± 0.0005,25–27 and therefore, A(Tc ) = n/12τ = n/1.3596, and dA(Tmax,0 + δ) dA(Tmax,0) d 2 A(Tmax,0)
= +δ
hence A(T0)/A(Tc ) = 0.4532. dT dT dT 2
Figure 6 shows the n–dependence of T0 (top frame, a) 3 dB(Tmax,0 + δ)
= , (60)
and Tmax (lower frame, b) of the Mie 2n : n potential, which 2n dT
is seen to decay monotonically with increasing n in the main terminating the expansion at the O(δ) order. As the parameter,
figures. The insets focus on the large n limit behavior. For T0 δ, is of order 1/n its effect on the right hand side of Eq. (60)
the formula A(T0) = n/3 represents the directly evaluated data can be neglected in this limit. Therefore
very well (the range n > 100 is shown).
The maximum temperature from Eq. (58) in the limit of 3 dB(Tmax,0) d 2 A(Tmax,0)
δ= / , (61)
n → ∞ is the solution to dA/dT = 0, which has the value of 2n dT dT 2
Reuse of AIP Publishing content is subject to the terms: https://publishing.aip.org/authors/rights-and-permissions. Downloaded to IP: 200.62.146.210 On: Wed, 31
Aug 2016 20:59:32
 084505-8 Heyes et al.
which gives Tmax = c0 + c1/n, where c0 = Tmax,0 and c1
= 3(dB/dT)/2(d 2 A/dT 2) = 37.18, with both derivatives eval-
uated at T = Tmax,0. The bottom frame of Fig. 6 gives the Tmax
data. In the inset, the red dots are the numerical solution of
Eq. (59) for Tmax(n). The magenta line is a fit to the numerical
data using c0 + c1n−1 + 163.3n−2 for n > 40. We found that
treating c1 as a free parameter, the value of 37.0 for the
coefficient of the second term was obtained from the fitting
procedure, which is statistically indistinguishable from the
value given by the above formula. The numerical analysis
therefore shows that Tmax is nearly linear in n−1 in the n → ∞
limit, in quantitative agreement with the above theory.
Figure 6 also shows T0 of the ∞ : n potential, which
indicates that this potential forms a lower bound of the Boyle
temperature of the m : n potential. Note that in taking the
m → ∞ limit of the Mie potential the strength parameter, λ,
is held constant.
Figure 7 shows the m−1 dependence of T0 and Tmax for the
m : 6 potential, concentrating on the large m limit. Note that
Tmax goes to ∞ as m → ∞, which is a qualitatively different
behavior from the large n limit of the 2n : n potential. In the
m → ∞ limit, the repulsive core tends to a hard sphere. The
second virial coefficient of the m : 6 potential for arbitrarily
large m tends to 0 as T → ∞ for finite m, but as m increases
b2 tends to b0 over a wider and higher temperature range.
The second virial coefficient of the Sutherland potential is a
monotonically increasing function of T, and as for all ∞ : n
potentials more generally, has no Tmax. This is why there are
no ∞ : n potential Tmax points on frame (b) of Fig. 6. Some
related discussion on the various high temperature trends for
different potentials is given in Chap. 3 of Ref. 20.
Figure 8(a) presents the critical temperature of the
2n : n potential as a function of n in the large n limit.
The red dots are calculated from the BM derived formula,
A(Tc ) = n/1.3596. The figure shows that Tc tends to zero
in the n → ∞ limit. The open square in the top right of
the figure is the value of Tc = 0.425 for n = 18 obtained by
FIG. 7. The top frame (a) shows the Boyle temperature for the m : 6 potential
as a function of 1/m obtained by numerical integration from Eq. (3) (open
circles or “D1”) and using Eq. (16) (crosses or “D2”). The orange asterisk
is for the ∞ : n or “IN” potential obtained by the summation in Eq. (25).
The diamond is T0 of the ∞ : 6 or Sutherland (“SP”) potential special case,
obtained from an alternative closed form expression formula for the SP b 2
derived by Levi and Llano.38 The dashed line is to guide the eye. The bottom
frame (b) shows Tmax as a function of 1/m obtained numerically by the same
b 2 formulas. The dashed line is to guide the eye.
Reuse of AIP Publishing content is subject to the terms: https://publishing.aip.org/authors/rights-and-permissions. Downloaded to IP: 200.62.146.210 On: Wed, 31
Aug 2016 20:59:32
J. Chem. Phys. 145, 084505 (2016)
FIG. 8. Frame (a) shows the critical temperature of the 2n : n potential as a
function of 1/n in the large n limit. The red dots are determined numerically
from the Baxter Model (“BM”) formula, A(Tc ) = n/1.3596. The open square
is for n = 18 obtained from simulation,2,28 using the Gibbs ensemble method
invented by Panagiotopoulos.29 The dashed line is to guide the eye. Frame
(b) shows the ratio of the Boyle to the critical temperature. The red dots are
calculated from the Baxter Model (“BM”) formula, A(T0)/ A(Tc ) = 0.4532.
The dashed line is to guide the eye.
Monte Carlo simulation,2,28 using the Gibbs ensemble method
invented by Panagiotopoulos.29 The BM formula prediction
of Tc is very close to this number, and therefore the BM
formula is quantitatively accurate even for relatively small n
values, which is better than might be expected. Substituting
A(Tc ) = n/1.3596 in Eq. (58) gives b2/b0 = −1.2075 or
b2/v0 = −4.83, where v0 = πσ 3/6 (the volume of the sphere)
at the critical temperature, which is independent of n for not
too small values. Therefore when b2 = −4.83v0, the system
is at its critical point temperature according to the theory. It
was concluded in Ref. 2 based on the simulation liquid-vapor
coexistence data that b2 = −6.3v0 at Tc , which is not too
far from our result. This figure is important for determining
the optimum conditions for protein crystallization (b2 can
be measured experimentally) which are thought to be in the
metastable fluid-fluid region of the phase diagram, just below
the critical temperature.2 Experimental optimum conditions
for protein nucleation have been found to be where b2 is more
negative than −4v0 (see Table V in Ref. 2), which is very close
to the value of b2(Tc ) = −4.83v0 obtained from the present
simple treatment.
Figure 8(b) shows the ratio of the Boyle to the
critical temperature, T0/Tc , derived from the BM formula,
A(T0)/A(Tc ) = 0.4532. As this ratio is less than unity, T0 > Tc ,
which is the same trend as observed in experiments on small
molecules. The ratio decreases to unity as n → ∞. Therefore
from Fig. 8, it can be concluded that both T0 and Tc tend to
zero in the n → ∞ limit. Also from Fig. 7 it is the case that
Tmax/T0 and Tmax/Tc → ∞ in the same limit.
The BM second virial coefficient has been considered in
Ref. 13 and has the low temperature limit, b2 = −12∆e1/T ,30
where ∆ is the width of the attractive region in units of the
core diameter. A comparison with Eq. (50) indicates that the
prefactor in front of the exponential could be interpreted as
representing a temperature dependent effective square well-
width, on assuming harmonic motion about the minimum in
the continuous potential case.
 084505-9 Heyes et al. J. Chem. Phys. 145, 084505 (2016)

F. Repulsive and attractive components of b2

Apart from the virial-based route of Eq. (4) another path
to the repulsive and attractive parts of b2 is via the formula
for the excess internal energy per molecule, u β, which can be
obtained from the thermodynamic relationship,31
 ρ
∂P(x)
( ) 
dx
u(ρ) β = P(x) − T . (62)
0 ∂T x x
2

At the two body or b2 level,11

)(
db2
u β = −T ρ, (63)
dT
using Eqs. (1) and (62). If the average potential energy per LJ FIG. 9. The second virial coefficient of the Lennard-Jones 12 : 6 potential as
a function of temperature. The repulsive and attractive components of b 2 are
particle is u, the repulsive (n = 12) part is ur and the attractive shown, as is the IP component from Eq. (21). Note the log-lin scale.
(n = 6) part is ua , then Pc = ρ(4ur + 2ua ),32,33 where Pc is
the interaction part of the pressure. Therefore, ur = Pc /2ρ − u
and ua = 2u − Pc /2ρ, respectively. So up to the second virial with temperature is extremely slow. For example, at T = 1000
component level and using these definitions of ur and ua , and the LJ b2 is 0.614 92 while the IP second virial coefficient
u from Eq. (63), is 0.645 44. Convergence to look indistinguishable on the
( ) scale of the figure is only achieved for ca. T > 104, which
b2 db2
ur β = +T ρ (64) would be physically unattainable for any noble gas atom in an
2 dT equilibrium state without it ionizing.
and
( )
b2 db2 G. Parametrized fits
ua β = − + 2T ρ. (65)
2 dT
Semi-empirical equations of state34–37 typically employ
Also as the repulsive and attractive components of the pressure an empirical polynomial approximation for b2(T) which is
are 4ur ρ and 2ua ρ, respectively, for the LJ potential, the accurate over a limited temperature range, from about the
repulsive and attractive parts of the second virial coefficient triple point temperature to T ≃ 10. One example of this form
are is34,35,37
x2 x3 x4 x5
( )
b2 db2
b2,r = 4 +T (66) b2(T) = x 1 + √ + + 2 + 3, (69)
2 dT T T T T
and where x i are constants. This function applies in the
( ) temperature range of ca. 0.6 < T < 10 relevant for most
b2 db2
b2,a = −2 + 2T , (67) equation of state studies but is incorrect in the low and
2 dT high temperature limits. It is more demanding to increase the
respectively, which can be computed from Eq. (29). temperature range of such a representation. The approach here
Alternatively, the first temperature derivative of b2(T) is is to exploit the known low and high temperature limits, which
( ) 5/4 ( ) are given in Eqs. (51) and (30), respectively. Then b2(T) can
πσ 3 1
( )
db2 1 1 1 1
= √ Γ − M − , , be expressed in terms of a “bridging” function, f (T),
dT 12 2 T 4 4 2 T
√
2πσ 1 3
( ) 7/4 ( )
1
(
1 3 1
) b2f it (T) = f (T)b2L L (T) + (1 − f (T))b2H L (T), (70)
+ Γ M , ,
4 T 4 4 2 T where LL and H L refer to the low and high temperature limits,
( ) 9/4 ( )
1 2 3 1 1 5 1 respectively, which note are both functions of temperature.
+ πσ e 0F1 ,
2T
The function f (T) is defined as
6 T 4 16T 2
( ) 11/4
b2(T) − b2H L (T)
( )
1 2 3 1 1 7 1
+ πσ e 0F1 ,
2T , (68) f (T) = , (71)
12 T 4 16T 2 b2L L (T) − b2H L (T)
where 0F1 is the hypergeometric series defined as which must go from 0 as T → ∞ to 1 as T → 0, with a
(1/Γ[a]) ∞ k=0 1/(a)k z /k!
 k
transitional range between the extremes mainly occupying
Figure 9 shows the temperature dependence of b2 the interval ca. 0.04 < T < 0.5. This constraint ensures
decomposed into the repulsive and attractive components of the correct limiting behavior of b2(T → 0) → b2L L (T) and
the LJ potential. They increase in magnitude with decreasing b2(T → ∞) → b2H L (T). An empirical function, f fit(T), which
temperature, and the total b2 value is the result of part satisfies these requirements is
cancellation of these two terms. The figure also shows the
convergence of b2 to the IP value (given in Eq. (21)) at high exp(ab) + 1
f fit(T) = . (72)
temperature. Convergence of the LJ b2 to the IP behavior exp(a(b − T c )) + 1

Reuse of AIP Publishing content is subject to the terms: https://publishing.aip.org/authors/rights-and-permissions. Downloaded to IP: 200.62.146.210 On: Wed, 31
Aug 2016 20:59:32
 084505-10 Heyes et al.
FIG. 10. The figure shows the function f (T ) defined in Eq. (71). It is
represented by f fit(T ) defined in Eq. (72) with a = −2754, b = 0.9784, and
f it
c = 0.0094. In the inset, the relative deviation, ∆ = (b 2(T ) − b 2 (T ))/b 2(T )
f it
is shown, and b 2 (T ) is calculated from Eq. (70) in which f fit for f (T ) is
used.
Figure 10 compares the exact function f (T) determined from
Eq. (71) with the least squares determined functional form
given in Eq. (72). The relative deviation given in the inset of
the figure is shown to be less than 10−5 throughout the whole
temperature range.
III. CONCLUSIONS
Various aspects of the second virial coefficient, b2, of
the Mie m : n potential have been examined, in particular the
2n : n special case, and notably for n = 6, the Lennard-Jones
potential. Closed form expressions are employed, and high
and low temperature series expansions are derived from first
principles or taken from the literature. For the low temperature
expansion, a general expression for a continuous potential of
arbitrary analytic form with a minimum is derived, which
is then applied to the Mie potential. The high temperature
expansion converges quite rapidly even at relatively low
temperatures. For example at T = 0.05, just 100 terms in
the expansion of the Lennard-Jones b2/b0 series of Eq. (18)
gives the value −1.430 87 × 108 which is exact to at least 5
significant figures. The leading term in the low temperature
expansion of Eq. (50) is −1.359 67 × 108. In fact the low
temperature series is an asymptotic expansion which therefore
does not converge at any temperature.
The Boyle and maximum temperatures are plotted as
a function of m and n, which shows that the former is
monotonically decaying and the latter exhibits a minimum
as these exponents increase. The large n limit of the second
virial coefficient of the 2n : n potential is calculated and a
simple accurate representation of this region is established
drawing a parallel with Baxter’s sticky sphere model. This
treatment gives formulas for the n–dependence of the Boyle,
maximum and critical temperature for large n, which even
represents closely a simulation value for n = 18 in the critical
temperature case. The m → ∞ (hard-sphere core) limit of
the maximum temperature of the m : n potential is shown to
diverge.
Two formulas which provide a resolution of b2 into
its repulsive and attractive terms are derived. The slow
Reuse of AIP Publishing content is subject to the terms: https://publishing.aip.org/authors/rights-and-permissions. Downloaded to IP: 200.62.146.210 On: Wed, 31
Aug 2016 20:59:32
J. Chem. Phys. 145, 084505 (2016)
convergence with increasing temperature of the LJ b2 towards
the repulsive inverse power limit is demonstrated. A semiem-
pirical formula for b2 which has the correct limits in the two
temperature extremes is constructed for the LJ special case.
To conclude, a number of new features of the second
virial coefficient of the Mie potential have been brought
to light. This includes methods for calculating it, and also
trends and new representations in certain parameter extremes,
notably for large values of the exponents. Of particular
note, we have derived accurate expressions for the Boyle,
maximum temperature of the second virial coefficient, and
the critical temperature for the 2n : n potential in the region
of large n. Such information has heretofore been lacking,
and currently inaccessible from simulation. Knowledge of
this short range attractive region is relevant to the behavior
of proteins in solution (e.g., for optimizing nucleation and
growth of crystals), and for colloid aggregation and gelation,
where the colloidal particle interactions can be represented
approximately by a 2n : n potential, with large n typically.
SUPPLEMENTARY MATERIAL
See the supplementary material for the Lennard-Jones
second virial coefficient temperature expansion coefficients,
and the derivation of the inverse power second virial coefficient
in arbitrary dimension.
ACKNOWLEDGMENTS
D.M.H. would like to thank Dr. T. Crane (Department
of Physics, Royal Holloway, University of London, UK)
for helpful software support. We thank the reviewers for
suggesting improvements to the work.
1Y. Song and E. A. Mason, J. Chem. Phys. 91, 7840 (1989).
2G. A. Vliegenthart and H. N. W. Lekkerkerker, J. Chem. Phys. 112, 5364
(2000).
3S. Zhou, Mol. Simul. 33, 1187 (2007).
4V. L. Kulinskii, J. Chem. Phys. 134, 144111 (2011).
5M. L. Glasser, Phys. Lett. A 300, 381 (2002).
6P. Vargas, E. Muñoz, and L. Rodriguez, Phys. A 290, 92 (2001).
7T. Kihara, Rev. Mod. Phys. 25, 831 (1953).
8B. A. Mamedov and E. Somuncu, J. Mol. Struct. 1068, 164 (2014).
9M. Edalat, S. S. Lan, F. Pang, and G. A. Mansoori, Int. J. Thermophys. 1,
177 (1980).
10J. D. Dymond, K. N. Marsh, and R. C. Wilhoit, in Virial Coefficients of
Pure Gases and Mixtures, edited by M. Fenkel and K. N. Marsh (Springer,
Landord-Bornstein, 2003), ISBN: 978-3-540-44340-7.
11A. J. Schultz and D. A. Kofke, Mol. Phys. 107, 2309 (2009).
12W. Witschel, Int. J. Thermophys. 11, 1075 (1990).
13A. González-Calderón and A. Rocha-Ichante, J. Chem. Phys. 142, 034305
(2015).
14J. E. Jones, Proc. R. Soc. A 106, 463 (1924).
15I. Nezbeda and W. R. Smith, Fluid Phase Equilib. 216, 183 (2004).
16J. Largo, M. A. Miller, and F. Sciortino, J. Chem. Phys. 128, 134513 (2008).
17N. M. Laurendeau, Statistical Thermodynamics: Fundamentals and Appli-
cations (Cambridge University Press, Cambridge, 2005), p. 378.
18B. C. Eu, “Exact analytic second virial coefficient for the Lennard-Jones,”
e-print arXiv:physics/0909.3326v1 (2009).
19A. Santos, in Proceedings of the 5th Warsaw School of Statistical Physics
(Warsaw University Press, Warsaw, 2014), p. 254, Eq. (7.87), and
employed in interactive software (2012), http://demonstrations.wolfram.
com/SecondVirialCoefficientsForTheLennardJones2nNPotential/; e-print
arXiv:1310.5578 (2013).
20A. Santos, A Concise Course on the Theory of Classical Liquids (Springer,
Berlin, 2016).
 084505-11 Heyes et al. J. Chem. Phys. 145, 084505 (2016)

21M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions 29A. Z. Panagiotopoulos, Mol. Phys. 61, 813 (1987).
(Dover Publications, New York, 1964), p. 504, Eq. (13.1.2). 30R. D. Rohrmann and A. Santos, Phys. Rev. E 89, 042121 (2014).
22I. Cachadiña, “Comment on ‘Exact analytic second virial coefficient for the 31D. Boda, T. Lukács, J Liszi, and I. Szalai, Fluid Phase Equilib. 119, 1

Lennard-Jones fluid”’, e-print arXiv:physics/1310.3586v1 (2013). (1996).

23G. Rickayzen and D. M. Heyes, J. Chem. Phys. 126, 114504 (2007), There 32R. Zwanzig and R. D. Mountain, J. Chem. Phys. 43, 4464 (1965).

is a misprint in that the parameter A should be divided by n each time it 33D. M. Heyes and A. C. Brańka, J. Chem. Phys. 143, 234504 (2015).

appears in Eq. (2.10).
24R. J. Baxter, J. Chem. Phys. 49, 2770 (1968).
25M. G. Noro and D. Frenkel, J. Chem. Phys. 113, 2941 (2000).
26M. A. Miller and D. Frenkel, Phys. Rev. Lett. 90, 135702 (2003).
27M. A. Miller and D. Frenkel, J. Chem. Phys. 121, 535 (2004).
28G. A. Vliegenthart, J. F. M. Lodge, and H. N. W. Lekkerkerker, Phys. A 263,
378 (1999).
34J. J. Nicolas, K. E. Gubbins, W. B. Streett, and D. J. Tildesley, Mol. Phys.
37, 1429 (1979).
35J. K. Johnson, J. A. Zollweg, and K. E. Gubbins, Mol. Phys. 78, 591 (1993).
36T. Sun and A. S. Teja, J. Phys. Chem. 100, 17365 (1996).
37H.-O. May and P. Mausbach, Phys. Rev. E 85, 031201 (2012); 86, 059905
(2012).
38D. Levi and M. de Llano, J. Chem. Phys. 63, 4561 (1975).

Reuse of AIP Publishing content is subject to the terms: https://publishing.aip.org/authors/rights-and-permissions. Downloaded to IP: 200.62.146.210 On: Wed, 31
Aug 2016 20:59:32