Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Molecular Physics

ISSN: 0026-8976 (Print) 1362-3028 (Online) Journal homepage: http://www.tandfonline.com/loi/tmph20

Thermodynamic properties for the triangular-well


fluid

F. F. Betancourt-Cárdenas , L. A. Galicia-Luna & S. I. Sandler

To cite this article: F. F. Betancourt-Cárdenas , L. A. Galicia-Luna & S. I. Sandler (2007)


Thermodynamic properties for the triangular-well fluid, Molecular Physics, 105:23-24,
2987-2998, DOI: 10.1080/00268970701725013

To link to this article: http://dx.doi.org/10.1080/00268970701725013

Published online: 01 Dec 2010.

Submit your article to this journal

Article views: 59

View related articles

Citing articles: 16 View citing articles

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=tmph20

Download by: [University of California, San Diego] Date: 15 March 2016, At: 07:43
Molecular Physics, Vol. 105, Nos. 23–24, 10 December–20 December 2007, 2987–2998

Thermodynamic properties for the triangular-well fluid


F. F. BETANCOURT-CÁRDENASy, L. A. GALICIA-LUNA*y and S. I. SANDLERz
yInstituto Politécnico Nacional, ESIQIE, Laboratorio de Termodinámica,
Edif. Z, Secc. 6, 1er piso, UPALM Zacatenco, 07738, Lindavista, México, D.F., México
zDepartment of Chemical Engineering, University of Delaware, 19716-3119, Newark, DE, USA

(Received 24 August 2007; in final form 25 September 2007)

The thermodynamic properties of the triangular-well fluid with a well range of up to twice the
hard sphere diameter were studied by means of a new developed equation of state and
molecular simulation. This EoS is based on the perturbation theory of Barker and Henderson
Downloaded by [University of California, San Diego] at 07:43 15 March 2016

with the first and second-order perturbation terms evaluated by molecular simulation and then
a fit with a simple function based on the radial distribution function of the reference fluid. The
thermodynamic properties for the triangular-well fluid were also obtained directly by Gibbs
ensemble and NPT Monte Carlo simulations. Good agreement is observed between the
proposed EoS and the molecular simulation results. A model for the triangular-well solid is
also presented; this has been used to calculate the solid–liquid transition line. Very good
agreement is obtained with previously report values for this line and for the triple point
temperature and pressure.

Keywords: Triangular-well potential; Monte Carlo simulation; Perturbation theory; Vapour–


liquid; Solid–fluid

1. Introduction where x ¼ r/ is the reduced distance, " is the well depth,
 is the diameter of hard core and  is the range of the
The square-well (SW) potential is one of the simplest potential, is another spherical potential. With this
spherically shaped molecules to exhibit a liquid–gas potential the interaction energy decreases as the inter-
transition and a critical point. Also the simple algebraic molecular distance between particles increases,
form of this potential allows analytical determination of which mimics the situation observed in real molecules.
the properties of a fluid interacting with this potential. Also its form is algebraically simple so it is possible to
As a result this potential is one of the most widely obtain analytical expressions for its thermodynamic
studied using a variety of techniques, including integral properties. This potential is graphically represented on
equations, molecular simulation and perturbation the- figure 1.
ories [1–15]. While the SW potential is a crude This potential has not been widely studied in the
approximation for the interactions between real mole- literature. The first results for this potential were the
cules, most other potentials lead to non-analytical evaluation of the second-, third-, and four-virial
expressions for the theoretically obtained thermody- coefficients [16–18]. Later, Card and Walkey employed
namic properties. the second-order perturbation theory of Barker and
The triangular-well (TW) potential Henderson (BH) to calculate up to the fourth virial
coefficient [19], and obtained good agreement with
8 previous results. In 1974, Card and Walkey reported
< 1 x1
uðxÞ ¼ "ðx  Þ=ð  1Þ 1 < x   ð1Þ the first Monte Carlo NVT (MCNVT) results for the
: TW fluid of  ¼ 2 and used the BH theory to obtain an
0 <x
equation of state (EoS) for this fluid [20]. Shortly
thereafter Smith et al. [21] evaluated the first- and
second-order terms of the BH perturbation theory for
this potential with  ¼ 2 and confirmed the good
agreement between the theory and the MCNVT results
*Corresponding author. Email: lgalicial@ipn.mx of Card and Walkey [20].

Molecular Physics
ISSN 0026–8976 print/ISSN 1362–3028 online  2007 Taylor & Francis
http://www.tandf.co.uk/journals
DOI: 10.1080/00268970701725013
2988 F. F. Betancourt-Cárdenas et al.

2. Theory

The BH perturbation theory is base on a expansion in


the inverse of reduced temperature, T ¼ kT/", from a
given reference state, this can be written as
Ares A0 1 A1 1 A2
¼ þ þ ð2Þ
NkT NkT T NkT T 2 NkT
where the subscript 0 refers to the free energy of a refe-
rence system, and Ai is the contribution of ith order to
the perturbation series. The first-order term is given by
Z1
A1 
¼ 12 g0 ðxÞup ðxÞx2 dx: ð3Þ
NkT 0
For the second-order term, A2 , will be taken as the so
Figure 1. Triangular-well potential. called macroscopic compressibility approximation [1]
Downloaded by [University of California, San Diego] at 07:43 15 March 2016

Z 1 h  i2
A2
¼ 6KHS g0 ðxÞ up ðxÞ x2 dx: ð4Þ
NkT 0
Clippe [22] and Rosenfeld and Thieberger [23] studied
convergence of the Zwanzig perturbation theory con-
In equations (3) and (4),  is the packing
cluding that the second-order term becomes less
fraction,  ¼  /6,  is the reduced density, g0(x) is
significant than the first as the density increases.
the radial distribution function (RDF) of the reference
Ausloos et al. [24] used the method presented by Clippe
hard-sphere fluid, u (x) is the reduced perturbation
and Evrard [25] to express the perturbation terms of the 
potential, up ðxÞ ¼ up ðxÞ=", and KHS is the hard-sphere
Zwanzig perturbation theory as polynomials in density;
isothermal compressibility.
they provided a numerical comparison with the fitted
results of Card and Walkey [20] and found some
discrepancies for the small second-order term. ð1  Þ4
KHS ¼ : ð5Þ
Recently, Largo and Solana [26] employed a 128-term 1 þ 4 þ 42  43 þ 4
expansion for the radial distribution function of the
hard sphere fluid [27] and coupled with the BH theory As can be observed from equations (3) and (4), it is
obtained an analytical EoS for the TW fluid. They necessary to know the RDF of the reference fluid to
compared this EoS with the second virial results and evaluate the contributions of the perturbation term to the
the MCNVT data of Card and Walkey [20] and found series. Based on the Percus–Yevick theory, analytical
good agreement. In 2002, Adhikari and Kofke [28] expressions for the RDF [29–32] of the hard sphere fluid
presented the solid–liquid and vapour–liquid phase are available; however, these expressions lead to com-
transition lines of the triangular well fluid for  up to plicated equations when used on perturbation theories.
2.5. They compared the result with a first-order Largo and Solana [27] proposed a simple expression
BH perturbation theory. Up to now the only avail- to expand the RDF fluid in the form of a double sum in
able molecular simulation data in literature for the terms of the packing fraction and reduced distance, but
triangular well fluid is the one presented by Card and their expression used 128 terms to fit the RDF values
Walkey [20] for  ¼ 2, and Adhikari and Kofke [28] from molecular simulation results. Following the work
for  up to 2.5. of Jiuxun et al. [33], the RDF can be expanded in non-
In this paper, an equation of state for the trian- linear functions of the distance. The form selected here is
gular well fluid is presented; this EoS is based on  i
1X 3

the second-order perturbation theory of Barker and gðxÞ ¼ Ci ð6Þ
 i¼1 1
Henderson. The contributions to the residual
X  
Helmholtz free energy are calculated by molecular 3
1 j
simulation and compared with the EoS developed Ci ¼ Dij x  4 ð7Þ
j¼0
x
here. Monte Carlo NPT (MCNPT) and Gibbs ensemble
(MCGE) simulations are also used to obtain the where the constants Dij were obtained by fitting
thermodynamic properties for the triangular well molecular simulation results for the RDF of the hard-
fluid and the results compared with those from the sphere fluid [34–35] for distances up to two times the
proposed EoS. hard-sphere diameter. The Dij values are presented on
Thermodynamic properties for the triangular-well fluid 2989

table 1. The absolute average deviation obtained for this The final form of the proposed EoS is given by
fit was 1.18%. By substitution of equations (6) and (7)
into equations (3) and (4) the analytic results for the X 3  i
Ares A0 12 
integrals are ¼ þ  i ðÞ
NkT NkT T ð  1Þ i¼1 1
 i
A1 12 X 3
 X 3  i
¼ i ðÞ ð8Þ 6 HS 
NkT   1 i¼1 1   K i ðÞ
T 2ð  1Þ2 i¼1
1
X3  i
A2 6 HS  ð12Þ
¼ K i ðÞ : ð9Þ
NkT ð  1Þ2 i¼1
1
Here i() and  i() are functions of the potential with A0 given by the Carnahan–Starling EoS for
range obtained from the integration of equation (7), and the hard-sphere fluid. The results for A1 and A2, the
can be written as last two terms in equation (12) are plotted on figures 2
and 3, respectively, together with the results from
1 molecular simulation (other thermodynamic properties
i ðÞ ¼ 1 þ 2 4 þ 3 5 þ 4 8 þ 5 9 þ 6 10 þ 7 12
Downloaded by [University of California, San Diego] at 07:43 15 March 2016

8  can be easy derived by classical relationships). As can be


þ8 13 þ 9 14 þ 10 15 þ ð11 þ 12 Þ8 lnðÞ seen, the EoS reproduces the A1 term quite well.
ð10Þ However, the A2 term, which is an order of magnitude
1  smaller, is not accurately reproduced.
i ðÞ ¼ 7 1 þ 2 4 þ 3 5 þ 4 7 þ 5 8 þ 6 9 þ 7 10 Adhikari and Kofke [28] reported that the triangular-

well fluid has a solid–liquid transition with fcc and hcp
þ 8 12 þ 9 13 þ 10 14 þ 11 15
 crystal structures in the solid phase. To compare with
þ ð 12 þ 13 Þ8 lnðÞ : ð11Þ their results, an equation of state for the fcc solid has
The values of constants are given in table 2. been developed as described next. The properties of the
solid phase are also obtained based on the perturbation
theory, but now using a solid composed by hard-spheres
Table 1. Constants for the RDF of the hard-sphere fluid
(equation (6)).
as the reference. The residual Helmholtz free energy can
be calculated from
j
Z
i 0 1 2 3 Ares Ares
sol,hs 2 2
sol
¼ þ gsol
0 ðrÞuðrÞr dr ð13Þ
NkT NkT kT
1 1 0.03977 0.17642 0.08513
2 1.5267 0.2008 3.6097 1.4838
3 0.4672 5.4681 6.5609 1.9056
where Ares
sol =NkT is the excess Helmholtz free energy
of the solid phase and Ares sol,hs =NkT is the excess

Table 2. Constants for the i() and  i() functions.

i()  i()

i/ i 1 2 3 1 2 3

1 0.001182 0.020608 0.026468 0.000338 0.005888 0.007562


2 0.008821 0.180484 0.328045 0.005881 0.120323 0.218696
3 0.021281 0.37095 0.47642 0.021281 0.37095 0.47642
4 0.024736 4.8308 15.8597 0.213128 1.12319 1.09331
5 0.226346 2.82724 13.1842 0.129016 9.25996 20.7832
6 0.127689 2.2257 2.85852 0.049922 6.43692 19.7451
7 0.083333 0.127229 0.03893 0.085126 1.4838 1.90568
8 0.001988 0.010041 0.273403 0.033333 0.050892 0.015572
9 0.005881 0.120323 0.218696 0.000663 0.003347 0.091134
10 0.002027 0.035327 0.045373 0.00168 0.03478 0.062485
11 0.039772 0.20082 5.46806 0.000501 0.008832 0.011343
12 0.352849 7.21938 13.1218 0.079543 0.40164 10.9361
13 0.352849 7.21938 13.1218
2990 F. F. Betancourt-Cárdenas et al.

0
pffiffiffi
and ¼ 2=6, c0 ¼ 2.557696, c1 ¼ 0.1253077,
c2 ¼ 0.1762393, c3 ¼ 1.053308, c4 ¼ 2.818621,
-1 c5 ¼ 2.818621, and c6 ¼ 1.118413.
The integral in equation (13) can be evaluated using
-2 the result given by Choi et al. [39] for the RDF of the
hard-sphere fcc solid
A1/NkT

-3

X
1

-4 g0 ðxÞ ¼ gðiÞ ð18Þ


i¼1
A
-5 gð1Þ ¼ exp ½W1 ðr  r1 Þ=d2 ½W2 ðr  r1 Þ=d4 ð19Þ
r
W ni  2
-6 gðiÞ ¼ pffiffiffi exp ½Wðr  ri Þ=d2 , i  2 ð20Þ
0.0 0.2 0.4 0.6 0.8 24  rri
ρ*
Downloaded by [University of California, San Diego] at 07:43 15 March 2016

Figure 2. A1/Nkt for the triangular-well potential. Dots are where g(1) and g(i) represent the contributions from the
molecular simulation results; (),  ¼ 2; (5),  ¼ 1.8; (4), first and higher order neighbor shells, and the quantities
 ¼ 1.6; (s),  ¼ 1.4; (œ),  ¼ 1.2. Continuous line, this work; ri and ni (i  2) are the distance and number of particles
dotted line, result from Smith et al. [21]; and dot-dashed line, in the ith nearest-neighbor lattice sites [40], respectively.
results from Ausloos et al. [24] both for  ¼ 2.
The resulting functions are given by


h  2
i
Helmholtz free energy of the reference solid phase W1 ðÞ ¼ 1:5338=  0:37687 exp 989:6   0:52
calculated from  
2
 2:5146  1:3574  8:5038 ð21Þ
Z
Ares
sol,hs Ares
sol,hs,0 Zhs ð0 Þ  1 0   
2
¼ þ d : ð14Þ W2 ðÞ ¼ 0:80313=  1:208 exp 5:6128 þ 67:808
NkT NkT 0 0 
3
 67:918 ð22Þ
Here 0 is a reference packing fraction and   
2
WðÞ ¼ 1:9881=  3:5276 þ 6:9762  26:205 ð23Þ
the Ares
sol,hs,0 =NkT is already known hard-sphere solid

 
1  8:0521 þ 18:003 2
r1 ðÞ ¼ ð24Þ
1  8:2973 þ 20:546  13:828 3 þ 103:95 4  582:74 5 þ 1245:7 6
 2

 pffiffiffi
reference state. This reference state free energy is  ¼ 2=6  : ð25Þ
given by Polson et al. [36] for a defect-free fcc crystal
 Only five shell contributions for the RDF of the solid
Ares
sol,hs 
¼ 5:91889: ð15Þ has been used, as this was found to give an accurate
NkT ¼0:5450
description of the RDF for distances up to twice the
Then using the equation of state of Hall [37] for hard sphere diameter. Then by substitution of equation
the compressibility factor of the hard-sphere solid, (18) into equation (13), numerical integration was used
equation (14) can be analytically integrated [38] to to evaluate the integrals for several values of  the
obtain the expression for the excess free energy of the results were fitted to a fourth- order polynomial in
solid phase as [38] density in order to obtain the derivatives with respect to
Ares
sol,hs
density. This model was then used to calculate the solid–
¼ 5:91889 þ fðÞ  fð0:5450Þ: ð16Þ fluid transition.
NkT
Here f() given by
" ! # 3. Molecular simulation
X6
i
fðÞ ¼ 4 ci  1 lnðÞ  3 lnð  Þ
i¼0 Thermodynamics properties for the TW fluid were
6 X
X 6
j! ð1Þ i
i evaluated by MCNPT and MCGE. The individual
þ 4j c j ð17Þ contributions of the terms in the perturbation series
i¼1 j¼i
ðj  iÞ! i  i!
were also calculated.
Thermodynamic properties for the triangular-well fluid 2991

(a) 0.00 (b) 0.00

-0.05 -0.05
A2/NkT

A2/NkT
-0.10 -0.10

-0.15 -0.15

-0.20 -0.20
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
ρ* ρ*

(c) 0.00 (d) 0.00


Downloaded by [University of California, San Diego] at 07:43 15 March 2016

-0.05 -0.05
A2/NkT

A2/NkT
-0.10 -0.10

-0.15 -0.15

-0.20 -0.20
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
ρ* ρ*

(e)

Figure 3. A2/Nkt for the triangular-well potential. Dots are molecular simulation results; a) (œ),  ¼ 1.2. b) (s),  ¼ 1.4. c) (4),
 ¼ 1.6. d) (5),  ¼ 1.8. e) (),  ¼ 2; continuous line, this work; dotted line, result from Smith et al. [21]; and dot-dashed line, results
from Ausloos et al. [24]. Note the change in scale.

As pointed out by Barker and Henderson the terms in The first- and second-order perturbation terms can be
the perturbation series can be obtained by Monte Carlo expressed in terms of fluctuations in the energy of the
or molecular dynamics simulations on the reference system as [3,32]
system [3], based on the interaction energy resulting
from a ‘virtual’ well. In this case, the terms of the A1 hU i0
¼ ð26Þ
perturbation series depend on density and well range, . NkT N
2992 F. F. Betancourt-Cárdenas et al.

A2 hU  hUii20 shown in the figures. Excellent agreement between the


¼ ð27Þ
NkT 2N simulation and analytic results is obtained for the first-
order term, while for the much smaller second-order
here U is the interaction energy above that of the term there are obvious discrepancies that can be found
reference as result of the perturbation interaction among the different data. Smith et al. [21] reported in
potential. However as this perturbation interaction the text of their paper that the second-order term was
must not influence the simulation it is a ‘virtual’ well, very small at high densities, though the value obtained
the symbol h. . .i indicates averages over all sampled from their fitted function increases with increasing
configurations and the subscript zero indicates that the density.
averages are performed in the hard-sphere reference The residual internal energies and densities for the
system. The averages in equations (26) and (27) can be TW fluid were determined by MCNPT simulations. For
evaluated using Monte Carlo [3,32] or molecular these simulations the system was composed of N ¼ 512
dynamics [15] simulation. particles originally placed on a simple cubic lattice. The
The first- and second-order perturbation terms were simulation was carried out in cycles, each one consisting
obtained by molecular dynamics of a hard-sphere on average in a moving attempt for every particle of the
Downloaded by [University of California, San Diego] at 07:43 15 March 2016

system based on the algorithm given in [41]. The system and five volume change attempts. To avoid
simulations were carried out for 512 particles, initially initial configuration effects, the system was equilibrated
located on a simple cubic lattice. The system was then for 1  106 cycles, and then the same number of cycles
allowed to equilibrate for 2  107 collisions to avoid any was used in the production phase sampling properties
influence of the original configuration in the sampled every 5  104 cycles. The size of the maximum particle
properties. The energies generated by the ‘virtual’ well displacement and maximum volume change were
were sampled and recorded every 2000 collisions. The adjusted during simulation to obtain an acceptance
properties of the system were then calculated from the ratio of 50%. The error bars for all properties were
same number of collisions in the production phase. In all calculated by block averages [42]. The results for  ¼ 1.2,
cases the value of , the range of the potential, was less 1.5, 1.75, and 2 are presented in tables 4 to 7,
than the half of the simulation box length, so no long- respectively. The RDF were also recorded during
range corrections were needed. simulations. As an example, figure 4 shows the results
Different  values and densities were considered; obtained for  ¼ 2 and T ¼ 2 at three different densities.
the results obtained for the perturbation terms together Notice that, unlike for the square-well fluid, the RDF
with error bars (which are smaller than the symbols for the triangular-well fluid has only a single disconti-
used in the figures) are given in table 3 and plotted in nuity at the hard core diameter.
figures 2 and 3. Smith et al. [21] previously reported The determination of the vapour–liquid coexistence
values and a fitted function for  ¼ 2, these are also curve was done by MCGE simulation [43] for a total of

Table 3. First- and second-order term for the triangular-well fluid. Numbers in parenthesis indicate the uncertainty on the last
digit, i.e. 0.0826(5) is –0.0826 0.0005.

  ¼ 1.2  ¼ 1.3  ¼ 1.4  ¼ 1.5  ¼ 1.6  ¼ 1.7  ¼ 1.8  ¼ 1.9 ¼2

A1/NkT
0.1 0.0826(1) 0.1281(1) 0.1799(1) 0.2364(1) 0.3660(2) 0.2987(1) 0.4399(1) 0.5193(1) 0.6063(1)
0.2 0.1833(1) 0.2853(1) 0.3963(1) 0.5167(1) 0.7856(1) 0.6458(1) 0.9348(1) 1.0966(1) 1.2699(2)
0.3 0.3105(1) 0.4777(1) 0.6561(2) 0.8445(1) 1.2562(1) 1.0445(2) 1.4822(1) 1.7227(1) 1.9812(1)
0.4 0.4700(1) 0.7137(1) 0.9647(1) 1.2252(1) 1.7776(1) 1.4945(1) 2.0741(1) 2.3905(2) 2.7287(1)
0.5 0.6696(1) 0.9991(2) 1.3271(1) 1.6579(1) 2.3402(1) 1.9939(1) 2.7016(1) 3.0849(1) 3.4972(2)
0.6 0.9211(1) 1.3427(1) 1.7470(1) 2.1423(2) 2.9349(1) 2.5356(1) 3.3502(1) 3.7920(1) 4.2736(1)
0.7 1.2340(1) 1.7510(1) 2.2244(1) 2.6729(1) 3.5470(1) 3.1085(1) 4.0032(1) 4.4962(1) 5.0448(1)
A2/NkT
0.1 0.0238(1) 0.0353(1) 0.0458(1) 0.0565(1) 0.0788(1) 0.0660(3) 0.0871(2) 0.1013(3) 0.1088(4)
0.2 0.0481(3) 0.0631(1) 0.0777(2) 0.0888(1) 0.1090(3) 0.0981(2) 0.1111(3) 0.1241(2) 0.1331(2)
0.3 0.0680(3) 0.0845(3) 0.0949(2) 0.1008(3) 0.1042(1) 0.1039(2) 0.1047(1) 0.1049(2) 0.1094(1)
0.4 0.0861(3) 0.0992(2) 0.0997(4) 0.1013(4) 0.0870(2) 0.0886(4) 0.0776(3) 0.0763(2) 0.0751(4)
0.5 0.1023(3) 0.0998(1) 0.0929(2) 0.0764(2) 0.0557(2) 0.0671(2) 0.0506(2) 0.0456(1) 0.0435(3)
0.6 0.1113(1) 0.0985(2) 0.0741(2) 0.0594(1) 0.0370(2) 0.0443(2) 0.0307(1) 0.0259(1) 0.0249(2)
0.7 0.1072(1) 0.0776(2) 0.0534(1) 0.0380(1) 0.0223(2) 0.0273(1) 0.0183(1) 0.0166(1) 0.0161(1)
Thermodynamic properties for the triangular-well fluid 2993

Table 4. MCNPT results for the triangular-well fluid  ¼ 1.2. Table 6. MCNPT results for the triangular-well fluid
Numbers in parenthesis indicate the uncertainty on the last  ¼ 1.75. Numbers in parenthesis indicate the uncertainty on
digit, i.e. 0.2737(6) is 0.2737 0.0006. the last digit, i.e. 0.2737(6) is 0.2737 0.0006.

T P  Ures/NkT T P  Ures/NkT

0.5 0.5 0.765(3) 4.03(1) 1 0.5 0.621(1) 3.364(7)


0.5 1.5 0.963(3) 5.54(2) 1 1.5 0.7486(9) 4.113(5)
0.5 2.5 1.030(1) 6.22(1) 1 2.5 0.8095(8) 4.481(4)
0.5 3.5 1.089(2) 7.07(2) 1 3.5 0.8527(5) 4.739(3)
0.5 4.5 1.115(1) 7.30(1) 1 4.5 0.8860(5) 4.935(3)
1 0.5 0.326(2) 0.535(3) 1.5 0.5 0.345(1) 1.199(3)
1 1.5 0.587(1) 1.135(4) 1.5 1.5 0.5820(8) 2.060(3)
1 2.5 0.715(2) 1.538(5) 1.5 2.5 0.6703(8) 2.413(3)
1 3.5 0.796(1) 1.82(1) 1.5 3.5 0.7276(6) 2.647(2)
1 4.5 0.852(1) 2.06(1) 1.5 4.5 0.7690(5) 2.817(2)
Downloaded by [University of California, San Diego] at 07:43 15 March 2016

1.5 0.5 0.229(1) 0.200(1) 2 0.5 0.2289(4) 0.572(1)


1.5 1.5 0.447(1) 0.468(1) 2 1.5 0.4627(6) 1.188(2)
1.5 2.5 0.567(1) 0.661(2) 2 2.5 0.5681(6) 1.494(2)
1.5 3.5 0.648(1) 0.814(2) 2 3.5 0.634(1) 1.694(2)
2 4.5 0.6830(5) 1.844(2)
2 0.5 0.1808(3) 0.1046(3)
2 1.5 0.372(1) 0.256(1)
2 2.5 0.484(1) 0.372(1) Table 7. MCNPT results for the triangular-well fluid  ¼ 2.
2 3.5 0.558(1) 0.462(1) Numbers in parenthesis indicate the uncertainty on the last
2 4.5 0.619(1) 0.546(1) digit, i.e. 0.2737(6) is 0.2737 0.0006.

T P  Ures/NkT

5 2 0.2737(6) 0.368(1)
Table 5. MCNPT results for the triangular-well fluid  ¼ 1.5. 2.74 0.5 0.1769(3) 0.4495(7)
Numbers in parenthesis indicate the uncertainty on the last 2.74 1.5 0.3901(9) 0.995(2)
digit, i.e. 0.472(2) is 0.472 0.002. 2.74 2.5 0.4932(9) 1.272(2)
2.74 3.5 0.5601(8) 1.456(3)
T P  Ures/NkT 2.74 4.5 0.607(1) 1.588(3)
2.74 5.5 0.6477(9) 1.700(2)
1 0.5 0.472(2) 1.783(7)
2.74 6.5 0.678(1) 1.785(3)
1 1.5 0.710(2) 2.822(7)
2.74 7.5 0.705(1) 1.861(3)
1 2.5 0.794(1) 3.263(6)
1 3.5 0.847(1) 3.558(5) 2 0.5 0.281(1) 1.007(4)
1 4.5 0.885(1) 3.776(5) 2 1.5 0.514(1) 1.834(4)
2 2.5 0.606(1) 2.176(4)
1.5 0.5 0.273(1) 0.613(1)
2 3.5 0.6665(9) 2.404(3)
1.5 1.5 0.523(1) 1.260(4)
2 4.5 0.7094(9) 2.567(3)
1.5 2.5 0.640(1) 1.619(4)
2 5.5 0.7422(8) 2.692(3)
1.5 3.5 0.706(1) 1.843(4)
2 6.5 0.772(1) 2.807(4)
1.5 4.5 0.757(1) 2.023(2)
1.2 0.05 0.0520(1) 0.413(1)
2 0.5 0.2021(3) 0.314(1)
1.2 0.5 0.621(2) 3.746(8)
2 1.5 0.419(1) 0.707(2)
1.2 1 0.689(1) 4.16(1)
2 2.5 0.534(1) 0.950(3)
1.2 1.5 0.735(1) 4.455(7)
2 3.5 0.608(1) 1.124(2)
1.2 2.5 0.796(1) 4.837(7)
2 4.5 0.663(1) 1.262(2)
1.2 3.5 0.835(1) 5.082(7)
2 5.5 0.706(1) 1.374(3)
1.2 4.5 0.870(1) 5.300(6)
1 0.5 0.724(1) 5.27(1)
1 1.5 0.807(1) 5.886(8)
1 2.5 0.859(1) 6.279(5)
1 3.5 0.899(1) 6.585(6)
1 4.5 0.932(1) 6.835(5)
0.8 0.5 0.832(1) 7.603(8)
0.8 1.5 0.896(1) 8.206(9)
2994 F. F. Betancourt-Cárdenas et al.

4 Table 8. MCGE results for the triangular-well fluid.


Numbers in parenthesis indicate the uncertainty on the last
digit, i.e. 0.376(7) is 0.376 0.007.
3  
 T 1 2 Ures
1 =NkT Ures
2 =NkT

1.5 0.670(2) 0.376(7) 0.376(7)


0.65 0.174(3) 0.575(3) 1.57(5) 3.77(4)
g(x)

2
0.6 0.073(2) 0.741(2) 0.82(3) 5.15(2)
0.55 0.032(2) 0.819(2) 0.43(3) 6.31(2)
0.5 0.014(2) 0.8783(3) 0.25(3) 7.58(2)
1 0.45 0.005(1) 0.9282(4) 0.10(3) 9.04(1)
1.75 0.922(5) 0.322(7) 0.322(7)
0.9 0.190(4) 0.466(7) 1.71(4) 2.95(5)
0 0.85 0.098(3) 0.579(4) 0.92(3) 3.75(3)
0 1 2 3 4
0.8 0.060(1) 0.652(3) 0.65(3) 4.47(2)
x=r/σ 0.75 0.039(1) 0.703(3) 0.48(3) 5.15(2)
Downloaded by [University of California, San Diego] at 07:43 15 March 2016

Figure 4. RDF for the triangular-well fluid  ¼ 2 from 0.7 0.022(1) 0.741(2) 0.30(2) 5.84(2)
molecular simulations at T ¼ 2. Dot-dashed line,  ¼ 0.28; 0.65 0.012(1) 0.777(2) 0.18(2) 6.62(1)
dashed line,  ¼ 0.67; continuous line,  ¼ 0.77. 0.6 0.006(1) 0.816(2) 0.10(2) 7.56(2)
1.95 1.147(4) 0.295(8) 0.295(8)
N ¼ 1024 particles. Once again the simulation was 1.1 0.113(2) 0.495(2) 0.95(3) 3.14(2)
done in cycles, each cycle consisting, on average, of a 1.05 0.078(2) 0.564(3) 0.72(2) 3.69(2)
displacement attempt for each particle in the vapor and 1 0.0545(4) 0.6080(7) 0.54(2) 4.17(2)
liquid boxes, six volume change attempts and 100 parti- 0.95 0.0397(4) 0.6564(7) 0.43(2) 4.74(1)
cle interchange attempts. The selection of the particle 0.9 0.0262(4) 0.6883(7) 0.31(2) 5.26(1)
movements was made at random. The two simulation 0.85 0.0171(3) 0.7202(6) 0.23(2) 5.83(1)
0.8 0.0118(3) 0.7568(6) 0.17(2) 6.52(1)
boxes, of equal volumes, were started with the particles
on a simple cubic lattice. The system was then
equilibrated for 1  105 cycles, to avoid any influence
molecular simulation, as can be seen in figure 2.
of the initial configurations, and then the properties
The results Smith et al. [21] are also presented and
were sampled every 100 cycles for the next 2  105 cycles.
seen also to be in good agreement, while the results of
The changes in volume and the size of the particle
Ausloos et al. [24] deviates from the others at densities
displacements were adjusted to give an acceptance ratio
greater than 0.5. In figure 3 a similar comparison is
of approximately 50%.
made for the A2 term, and we see that the approach here
The critical points were estimated from the MCGE
of using the ‘macroscopic compressibility’ expression
data with the scaling law and the law of rectilinear
give values that are about an order of magnitude larger
diameters [42]
than those obtained from simulation (and goes out of
l  g ¼CðT  Tc Þ c ð28Þ scale for the figures). Figure 3(e) also includes the results
l þ g from Smith et al. [21] and Ausloos et al. [24], and we see
¼c þ DðT  Tc Þ ð29Þ that none of them can reproduce the values obtained
2
from simulation.
where l(g) is the density of the liquid (gas) phase, c and The second virial coefficient, pressure, residual inter-
Tc are the critical density and temperature respectively, nal energy and vapour–liquid equilibrium have been
C and D are constants that depend on the system and calculated with the proposed EoS. The predicted second
were obtained by a least squares procedure, and c is the virial coefficient is plotted in figure 5 together with the
critical exponent (considered to be c 1=3 for three- analytical solution of Feinberg and Rocco [16]. Excellent
dimensional systems). The critical points obtained (first agreement is seen for reduced temperatures greater than
entry at each  value) together with the MCGE data are 1; for lower temperatures large deviations are present as
reported on table 8. a consequence of the slow convergence of the perturba-
tion series at low temperatures.
4. Results The predicted pressure–density diagrams are presented
in figures 6 to 9, and we see that the prediction from the
The results for the first perturbation term from equation proposed EoS improves as the range of the triangular
(8) are in excellent agreement with those obtained from well increases. The value of  ¼ 1.2 was the smallest
Thermodynamic properties for the triangular-well fluid 2995

5 14

0 12

-5
10

-10
8
B2*

*
-15

P
6
-20

4
-25

-30 2

-35 0
0 1 2 3 4 5 6 0.0 0.2 0.4 0.6 0.8

T* ρ*
Downloaded by [University of California, San Diego] at 07:43 15 March 2016

Figure 5. Second virial coefficient for the triangular-well Figure 7. Pressure–density diagram for the triangular-well
fluid. Dots, analytic solution; (),  ¼ 2; (4),  ¼ 1.5; (œ), fluid  ¼ 1.5. (), T ¼ 2; (5), T ¼ 1.5; (4), T ¼ 1.
 ¼ 1.2. Continuous line, this work EoS. Continuous line, this work EoS.

14 14

12 12

10 10

8 8
*

*
P

6 6

4 4

2 2

0 0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
ρ* ρ*

Figure 6. Pressure–density diagram for the triangular-well Figure 8. Pressure–density diagram for the triangular-well
fluid  ¼ 1.2. (), T ¼ 2; (5), T ¼ 1.5; (4), T ¼ 1. fluid  ¼ 1.75. (),T ¼ 2; (5), T ¼ 1.5; (4), T ¼ 1.
Continuous line, this work EoS. Continuous line, this work EoS.

considered because, based on the data of Adhikari and pressures are larger than those obtained from the
Kofke [28], smaller values of the well do not have a molecular simulation, especially at densities greater
vapour–liquid coexistence region. Very good agreement than 0.8. Adhikari and Kofke do not report the phase
between the results from the EoS and simulation is seen in diagram for  ¼ 2, but from the interpretation of their
these diagrams. In the case of  ¼ 1.2, good agreement is data it appears that the lowest temperature plotted
obtained, though at T ¼ 1 which is close to the solid– (T ¼ 0.8) is close to the triple point of this fluid.
liquid transition according to the data of Adhikari and The residual internal energy was also predicted with
Kofke [28], there are deviations. the proposed EoS, and this is plotted in figure 10
For the case  ¼ 2 there is excellent agreement together with the results from molecular simulation. By
between our results with the reported data of Card using the second-order theory as presented here good
and Walkey [20]. The proposed EoS provides an agreement is obtained for the predicted residual internal
excellent representation for reduced temperatures energy except for the case  ¼ 1.2. As seen in figure 3, the
greater than 1.3. At lower temperatures the predicted values predicted from the macroscopic compressibility
2996 F. F. Betancourt-Cárdenas et al.

20

15

*
10
P

5
Downloaded by [University of California, San Diego] at 07:43 15 March 2016

0
0.0 0.2 0.4 0.6 0.8 1.0
ρ*
Figure 9. Pressure–density diagram for the triangular-well fluid  ¼ 2. Open symbols, this work. (), T ¼ 5; (5), T ¼ 2.74; (4),
T ¼ 2; ( þ ), T ¼ 1.5; (œ), T ¼ 1.2; (s), T ¼ 1; (§), T ¼ 0.8; closed symbols, Card and Walkey [20] at the same symbol
temperature.

0.0
The vapour–liquid equilibrium curves for the trian-
-0.5
gular well potential are presented in figure 11. The
results shown from the MCGE simulations excellent
-1.0 agreement with the results of Adhikari and Kofke [28].
The predicted coexistence curves from the EoS are also
presented there and also are in good agreement. The
U /NkT

-1.5

vapour–liquid critical point predicted from the EoS is at


res

-2.0
a higher temperature than that obtained from MCGE
-2.5
simulation, which is expected for an EoS based on the
BH perturbation theory. Here also the predictions from
-3.0 the EoS improve as the value of  increases.
The solid–liquid transition was calculated by simula-
-3.5 tion, and compared with the previous results of
0.0 0.2 0.4 0.6 0.8 1.0
ρ*
Adhikari and Kofke [28]. As this was not the focus of
this work, the solid–liquid transition was computed only
Figure 10. Residual internal energy for the triangular-well using simulation for  ¼ 1.5. However, by using the EoS
fluid at T ¼ 2 at different  values. (),  ¼ 2; (5),  ¼ 1.75;
for the triangular well fluid proposed in this work and
(4),  ¼ 1.5; (œ),  ¼ 1.2. Lines are the result of the EoS
proposed here, with the dashed line as a result of only the first- the procedure presented in the previous section to
order term in the expansion and continuous line resulting from calculate the properties of the solid phase, the solid–
including the second-order term. liquid transition has been calculated by solving the
equality of fugacity and pressure for both phases.
The computed pressure–temperature diagram is com-
approximation gives a A2/NkT values that are larger pared with the results of Adhikari and Kofke [28] in
than those obtained from molecular simulation. This is figure 12, where we observe excellent agreement with
because for small values of  the magnitude of the first- simulation results in that reference. The triple point
order term is small so that the second-order term is temperature and pressure has been estimated by having
significant. Consequently, we have also plotted the equal pressure and fugacity in each of the three phases.
residual internal energy obtained using only the first- The results, presented on table 9, are in very good
order perturbation term, and in this case we find agreement with the results of Adhikari and Kofke [28]
excellent agreement with simulation. There is a similar, obtained from MCGE simulation and Gibbs–Duhem
though smaller effect on the predicted pressure. integration.
Thermodynamic properties for the triangular-well fluid 2997

1.4
5. Conclusions

1.2 An EoS for the triangular-well fluid based on the BH


perturbation theory has been developed. This EoS is
analytic, and is a function of well width, packing
1.0
fraction (density), and temperature, and is applicable
*

for well widths  from 1.1 to 2.


T

0.8 The contributions of the first- and second-order terms


in the BH perturbation series were evaluated from
molecular simulation. Excellent agreement between the
0.6
EoS and simulation results was observed for the first-
order term. The contribution of the second-order term
0.4 determined from simulation is very small (|A2/
0.0 0.2 0.4 0.6 0.8 1.0
ρ*
NkT| 5 0.2) and is not accurately predicted from the
EoS. This inaccuracy mainly affects the residual internal
Downloaded by [University of California, San Diego] at 07:43 15 March 2016

Figure 11. Vapour–liquid equilibrium for the triangular-


energy for the case of  ¼ 1.2 where the EoS predicts
well fluid. Open Symbols, This work; Closed symbols,
Adhikari and Kofke [28]. (),  ¼ 2; (4),  ¼ 1.75; (5), values systematically greater than the results from
 ¼ 1.5. (s), Critical point from EoS; (§) Critical point molecular simulation. For larger values of  the
from simulation. magnitude of the second-order term is smaller than the
first-order term and this effect is less important.
The effect of the second-order term in the EoS on the
predicted pressure is smaller, and very good agreement is
obtained with the results of molecular simulation. Also
102
values of the second virial coefficient from the EoS are
101
in excellent agreement with the exact results for the
triangular-well potential.
100 The vapour–liquid equilibrium was calculated for
three values of  from MCGE simulations. The curves
10-1 predicted with the EOS are in generally good agreement
βPσ3

with those obtained from simulation and others


10-2
previously reported by others. The EoS predicts higher
10-3
critical temperatures, as is expected as a result of the
slow convergence of the perturbation series expansion
10-4 close to the critical point.
Finally, the solid–liquid phase transition was also
10-5 calculated for the triangular-well potential with the
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
β
proposed EoS and compared with previous results. The
results are in very good agreement for the estimated
Figure 12. Pressure-inverse temperature diagram ¼ "/kT triple point temperature and pressure with the results
for the triangular well fluid for  ¼ 1.5. The continuous lines
are vapour–liquid equilibria and solid–liquid equilibria from obtained by others. However, the thermodynamic
the proposed model, circles are simulation data from Adhikari properties for the fcc solid phase were estimated from
and Kofke [28], and the dashed line is the model presented by perturbation theory instead of the previous cell model
Adhikari and Kofke [28]. calculations [28].

Table 9. Triple point temperature and pressure for the triangular-well fluid  ¼ 1.5.

This work (EoS) Adhikari and Kofke (Theory) Adhikari and Kofke (Molecular Simulation)
3 3
Inverse Pressure ( P ) Inverse Pressure ( P ) Inverse Pressure ( P 3)
temperature ( ) temperature ( ) temperature ( )
2.035 0.0094 1.9 0.0180 2.04 0.0078
2998 F. F. Betancourt-Cárdenas et al.

Acknowledegements [18] W.C. Farrar and A.G. De Rocco, J. Chem. Phys. 54, 2024
(1971). Erratum: J. Chem. Phys. 55, 4683 (1971).
[19] D. A. Card and J. Walkley, Can. J. Phys. 50, 1419 (1972).
The authors thank CONACYT and IPN for their [20] D. A. Card and J. Walkley, Can. J. Phys. 52, 80 (1974).
financial support. Thanks are given to Professor D.A. [21] W. R. Smith, D. Henderson, and J. A. Barker, Can. J.
Kofke and Professor J. Adhikari for sending us their Phys. 53, 5 (1975).
simulation data for triangular-well fluid. Thanks are [22] P. Clippe, J. Chem. Phys. 65, 2982 (1976).
given to Intel Co. for grant us a non-commercial/ [23] Y. Rosenfeld and R. Thieberger, J. Chem. Phys. 63, 1875
(1975).
academic license for their Fortran compiler for Linux [24] M. Ausloos, P. Clippe, R. Evrard, and R. Verhaeghe,
Mol. Phys. 37, 643 (1979).
[25] P. Clippe and R. Evrard, J. Chem. Phys. 64, 3217 (1976).
[26] J. Largo and J. R. Solana, Physica A 284, 68 (2000).
References [27] J. Largo and J. R. Solana, Fluid Phse Equilbr. 167, 21
(2000).
[1] J. A. Barker and D. Henderson, J. Chem. Phys. 47, 2856 [28] J. Adhikari and D. A. Kofke, Mol. Sim. 100, 1543 (2002).
(1967). [29] M.S. Wertheim, Phys. Rev. Lett., 10, 321 (1963). E.
[2] W. R. Smith, D. Henderson, and J. A. Barker, J. Chem. Thiele, J. Chem. Phys., 39, 474 (1963).
Downloaded by [University of California, San Diego] at 07:43 15 March 2016

Phys. 55, 4027 (1971). [30] W. R. Smith and D. Henderson, Mol. Phys. 19, 411
[3] J. A. Barker and D. Henderson, Rev. Mod. Phys. 48, 587 (1970).
(1976). [31] J. Chang and S. I. Sandler, Mol. Phys. 81, 735 (1994).
[4] D. Levesque, Physica 32, 1985 (1966). [32] J. Largo and J. R. Solana, Mol. Sim. 29, 363 (2003).
[5] Y. Tago, J. Chem. Phys. 58, 2096 (1973). [33] J.-X. Sun, L.-G. Cai, Q. Wu, and F.-Q. Jing, Commun.
[6] W. R. Smith, D. Henderson, and R. D. Murphy, J. Chem. Theor. Phys. 41, 400 (2004).
Phys. 61, 2911 (1974). [34] J. Largo and J. R. Solana, Fluid Phase Equilibr. 212, 11
[7] W. R. Smith and D. Henderson, J. Chem. Phys. 69, 319 (2003).
(1978). [35] J. A. Barker and D. Henderson, Mol. Phys. 21, 187
[8] D. Henderson, W. G. Madden, and D. D. Fitts, J. Chem. (1971).
Phys. 64, 5026 (1976). [36] J. M. Polson, E. Trizac, S. Pronk, and D. Frenkel, J.
[9] W. R. Smith, D. Henderson, and Y. Tago, J. Chem. Phys. Chem. Phys. 112, 5339 (2000).
67, 5308 (1977). [37] K. R. Hall, J. Chem. Phys. 57, 2252 (1972).
[10] F. del Rı́o and L. Lira, Mol. Phys. 61, 275 (1987). [38] H. Adiharma, S. P. Tan, and M. Radosz, Mol. Phys. 100,
[11] F. del Rı́o and L. Lira, J. Chem. Phys. 87, 7179 (1987). 2559 (2002).
[12] A. L. Benavides, J. Alejandre, and F. del Rio, Mol. Phys. [39] Y. Choi, T. Ree, and F. H. Ree, J. Chem. Phys. 95, 7548
74, 321 (1991). (1991).
[13] E. Schöll-Paschinger, A. L. Benavides, and R. Castañeda- [40] J. D. Hirschfelder, C. F. Curtiss, and R. B. Bird,
Priego, J. Chem. Phys. 123, 234513 (2005). Molecular Theory of Gases and Liquids (Wiley,
[14] A. Rotenberg, J. Chem. Phys. 43, 1198 (1965). USA, 1964).
[15] B. J. Alder, D. A. Young, and M. A. Mark, J. Chem. [41] J. M. Haile, Molecular Dynamics Simulation: Elementary
Phys. 56, 3013 (1972). Methods (John Wiley and Sons Inc., USA, 1992).
[16] M. J. Feinberg and A. G. De Rocco, J. Chem. Phys. 41, [42] D. Frenkel and B. Smit, Understanding Molecular
3439 (1964). Simulations: From Algorithms to Applications, 2 edn.
[17] R. H. Fowler, H. W. Graben, A. G. De Rocco, and M. (Academic Press, Inc., USA, 2001).
J. Feinberg, J. Chem. Phys. 43, 1083 (1965). [43] A. Z. Panagiotopoulos, Mol. Phys. 61, 813 (1987).

You might also like