Ind. En$. Chem. Fundam.
1982, 21, 191-197 191
Solubilities of Hydrocarbon Solids in Supercritical Fluids. The
Augmented van der Waals Treatment
Keith P. Johnston, Davld H. Zlger, and Charles A. Eckert'
Department of Chemical En$lneering, Universlty of Illlnols, Urbana, Illinois 6 780 1
New data are presented for the solubility of several nonpolar hydrocarbon solids in supercritical ethylene, ethane,
and CO,. These data, plus others reported previously, are well correlated by perturbed hard-sphere equations
of state that do not require the w e of critical properties. The addition of a second-order term based on square-well
molecular dynamics results increases the range of applicability of the model significantly.
Introduction
Heavy nonvolatile organic solids dissolve in supercritical
fluids up to ten orders of magnitude in excess of the ideal
amount (Hannay and Hogarth, 1879). This enhanced
solubility is related primarily to the "liquid-like" density
of the fluid which promotes strong attractive forces. This
volatilization phenomenon is a basis for supercritical fluid
extraction, a new alternative to distillation and liquid
extraction. A unique property of this process is that the
dissolved extract and supercritical fluid may be separated,
thoroughly and reversibly, by decreasing the fluid density,
thus its dissolving power. This may be accomplished by
expansion to subcritical pressure or by a relatively small
increase in temperature.
In comparison to the conventional processes, supercri-
tical fluid extraction offers considerable flexibility for an
extractive separation through controlling pressure, tem-
perature, and choice of solvent (Todd and Elgin, 1955).
The energy requirements may be reduced significantly for
volatilization and separation of the heavy components
(Kohn and Savage, 1979).
Although numerous applications of supercritical fluid
extraction have appeared in the patent literature, with
several already at the production stage, there is little
quantitative fundamental understanding of the phase
behavior. The present goal is to develop an analytical
semitheoretical model to predict the thermodynamic be-
havior of supercritical solutions. This involves two major
difficulties: (1)these mixtures are often highly asymmetric,
in terms of the size and energy differences between the
Components, reflected for instance in the critical properties,
and (2) the isothermal compressibility of the fluid is order
of magnitude 103times that of a triple point liquid. These
complexities offer sufficient challenge to justify choosing
the simplest possible experimental system to characterize
the nature of the supercritical fluid phase. Consequently,
we have studied the solubility of nonpolar solids in su-
percritical nonpolar fluids, near infinite dilution (to avoid
the effects of solute-solute interactions). In this manner
we seek to characterize the fundamental properties of
supercritical fluid phases. Data were taken using a newly
developed flow technique, which greatly reduced the ex-
perimental run time, while simplifying sampling and
composition analysis. Experimental studies of solid-fluid
equilibria (SFE)have increased dramatically, perhaps due
to applications in food and pharmaceutical industries,
application in environmental control (De Filippi et al.,
1980), coal liquefaction, tar sands processing, and COz
enhanced oil recovery. Table I shows solubility data for
pure organic solids in supercritical ethylene, ethane, and
0198-43131a211021-01gi~0i2510
carbon dioxide within 20% of the solvent critical tem-
perature.
One of the requirements for the proposed model was to
avoid conventional corresponding states treatments based
on the critical properties, which are difficult to apply to
highly asymmetric systems. Instead, an alternate choice
of solid properties, namely the enthalpy of vaporization,
along with the vapor pressure, led to a reduction in the
number of adjustable mixture parameters. This provided
for the possibility of thermodynamic prediction of un-
measured systems. Johnston and Eckert (1981) investi-
gated this approach in reference to a first-order Carna-
han-Starling-van der Waals (CS-VDW) model (see eq 15)
for the relatively incompressible dense supercritical region.
In order to model the highly compressible or near su-
percritical region, the second-order term in perturbation
theory is needed to include the enhanced attractive forces
in the mixture, which is beyond the domain of cubic
equations of state. An augmented van der Waals (AVDW)
model was developed based on the molecular dynamics for
the square well fluid to include this effect which becomes
apparent experimentally at the lower densities, and at
near-critical temperatures. The AVDW model advanta-
geously complemented the CS-VDW model, as it was
successful down to the critical density, while converging
with the latter at higher reduced densities, in the range
of 1.3 to 1.5.
Experimental Section
The apparatus and procedure were described by John-
ston and Eckert (19811, Johnston (1979), and Ziger (1980).
In comparison to conventional static designs, this flow
apparatus offered greatly reduced run times due to in-
creased surface area of contact between the packed solid
and flowing gas. The solution was separated completely
by depressurization to one atmosphere, thus simplifying
sampling and composition analysis.
Solubility isotherms for fourteen hydrocarbon-super-
critical fluid systems are shown in Table I1 and Figure 1.
For comparison, several slightly subcritical solubility iso-
therms are shown for both ethane (T, = 32.3 OC) and
carbon dioxide (T, = 31.1 "C). These results complement
the similar data reported earlier (Johnston and Eckert,
1981).
Almost half the data listed in Table I1 are averaged from
duplicate or triplicate runs. The average deviation for
repeated runs was 3.5%, which compared favorably with
an assumed experimental uncertainty of 3-5%. Below lo4
mole fraction, the estimated uncertainty increased to 10
to 20%, and below 10" to 40%. This increase was in-
versely related to the weight of solid collected.
0 1982 American Chemical Society
192 Ind. Eng. Chem. Fundam., Vol. 21, No. 3, 1982

Table I. Available Solubility Data

system TIT, PIP, reference
Ethylene
anthracene 1.14-1.3 2-9.7 Johnston and Eclcert (1981)
benzoic acid 1.13-1.20 2.41-5.63 Kurnik and Reid (1981)
p-chloroiodobenzene 1.01-1.08 0.5-3 Ewald (1952)
hexachloroethane 1.0 2-1.0 5 0.5-3.4 Ewald (1952)
naphthalene 1.01-1.09 1-2 Diepen and Scheffer (1948)
Tsekhanskaya (1964)
2,3-dimethylnaphthalene 1.08-1.16 1.61-5.63 Kurnik and Reid (1981)
2,6-dimethylnaphthalene 1.08-1.1 6 1.61-5.63 Kurnik and Reid (1981)
phenanthrene 1.11 2.68-1 0.95 Eisenbeiss (1964)
1.05-1.21 1.1-5.5 Johnston and Eckert (1981)
Ethane
phenanthrene 1.03 2.9-11.3 Eisenbeiss (1964)
Carbon Dioxide
benzoic acid 1.05-1.1 1.65-3.85 Kurnik and Reid (1981)
biphenyl 1.02-1.08 1.4-4.5 McHugh and Paulaitis (1980)
diphenylamine 1.0-1.02 0.7-3.10 Tsekhanskaya (1962)
hexachloroethane 1.01-1.08 1.33-3.85 Kurnik and Reid (1981)
naphthalene 1-1.08 0.82-4.4 Tsekhanskaya (1964)
2,3-dimethylnaphthalene 1.01-1.08 1.33-3.85 Kurnik and Reid (1981)
2,6-dimethylnaphthalene 1.01-1.08 1.33-3.85 Kurnik and Reid (1981)
phenanthrene 1.05-1.1 1.65-3.85 Kurnik and Reid (1981)
phenol 1.02-1.09 1.07-3.43 van Iper and Paulaitis (1980)
p-chlorophenol 1.02 1.lo-3.24 van Leer and Paulaitis (1980)
2,4-dichlorophenol 1.02 1.09-2.80 van Leer and Paulaitis (1980)
I'
0
A T=318 K

6 8 IO 12 14 16
DENSITY (gmol/l)
PRESSURE lo5 PO
Figure 1. Solubility vs. pressure isotherms for phenanthrene in Figure 2. Enhancement factor vs. density isotherms for phenan-
ethylene. threne in ethylene.
Io6
Frequently it is useful to describe the solubility in terms
0 T = 308 K
of an enhancement factor E , which is the ratio of the actual A T.323 K
gas solubility, over the ideal gas solubility LL
0
0 T: 343 K
I- Augmented VOW
E = yzP/PzSat (1)
Such results are shown in Figures 2 and 3. In the range +
of infinite dilution, the solution density approaches the z
W
pure solvent density, which was obtained from Din (1961),
1962), Douslin and Harrison (1976), and Michels and
Q
Michels (1935,1937). These densities were used instead I
z
of one calculated from a mixture equation of state both W

in these figures, and in the model described below. il

lo3 1 I
Thermodynamic Modeling for Solid Fluid IO 12 14 16 18 20 22
Equilibria (SFE) DENSITY ( g m o l / l )
The fugacity for the fluid phase may be described using Figure 3. Enhancement factor vs. density isotherms for fluorene in
an "expanded liquid" or gas approach coz.
In each case, the uz is the crucial and most fundamental
macroscopic property for describing the high pressure
mixture behavior; however, such data as a function of T
and P are virtually nonexistent. Paulaitis, Johnston, and
The fugacity for the solid phase is comparatively simple Eckert (1981) demonstrated that o2 cannot be obtained
simply from the y2 vs. P data. In the gas case c $ ~ depends
u,(P - P p ) upon u2, which is related to the isothermal compressibility.
fzs = fios exp (4)
RT The dissolution of the heavy component in this highly

c I
101 I I I
8 1
Pi238K
p,
Figure 4. Isothermal compressibility of ethylene vs. reduced den-
sity.
compressible fluid leads to strong attractive forces pro-
moting a large contraction, with O2 approaching negative
infiiity at the pure solvent critical point. Therefore, SFE
is complicated severely due to the large compressibility of
the fluid. As shown in Figure 4, the compressibility for
pure ethylene at 15O above the T,reaches ten times that
of an ideal gas and lo00 times that of a fluid at the triple
point.
The original theoretical attempts to model SFE by
Ewald (1953), and Robin and Vodar (1953) utilized the
virial equation and were thus limited to approximately 1 / 4
the critical density. Mackay and Paulaitis (1979) used eq
2 to correlate successfully several SFE systems, using the
Redlich-Kwong equation with a quadratic energy mixing
rule to obtain D2. Kurnik et. al (1981) used eq 3, while
obtaining & from the Peng-Robinson equation. In these
methods, the attractive and repulsive parameters were
predicted from the critical properties, which are often
unavailable, and if available, hard to combine. The vastly
different critical properties of the asymmetric components
exacerbates the problem of choosing a corresponding state
that can describe the unlike or cross interactions. The
solvent interacts simultaneously with only part of the large
solute, making predictions of a kij difficult. Further dif-
ficulties encountered in highly asymmetric systems were
discussed by Kaul and Prausnitz (1977), Kreglewski and
Chen (1979), and Henderson (1974).
In the CS-VDW model, the crucial unlike pair energy
parameter was fit from the binary data. Instead of using
the solute critical properties with kij,this parameter was
related successfully to another property, the heat of va-
porization of the solute at the normal melting point. It
was this methodology, and not the choice of a particular
cubic equation of state, that led to the improvement. The
variations in the numerous first-order cubic equations are
trivial compared to the application of any of the better
ones to a complicated asymmetric or highly compressible
mixture.
Models for Highly Compressible Fluid Mixtures
That Go Beyond First Order
As the solvent density approaches the critical density,
the CS-VDW model underpredicts the attractive forces,
and thus the enhancement (see Figure 2). This discrep-
ancy becomes significant as the temperature decreases
toward the critical, where the compressibility is largest.
The description of this near supercritical region is phys-
ically interesting, yet also necessary, for it is the region for
the precipitation stage in an extraction process.
The physical nature of this region may be revealed by
an examination of high-temperature expansion perturba-
Ind. Eng. Chem. Fundam., Vol. 21, No. 3, 1982 193
tion theory (Hansen and McDonald, 1976). The pair po-
tential consists of a repulsive part uo(r),perturbed by an
attractive potential w(r)
u(r) = uo(r) + Xw(r)
This leads to the free energy expression
(5)
PA = PA, + P( wN)O - ‘/zP~((
wN2)o - (wN)oz) + 0 ~ 3 3 )
(6)
with the total perturbation energy
N
WN = EEw(ij)
i<j
(7)
The Carnahan-Starling hard-sphere equation of state
describes analytically the zeroth-order repulsive term. The
first-order or mean field term
AI = iJg~(1,2)w(1,2)
2 drlz (8)
involves an average value of the perturbing attractive en-
ergy integrated over the unchanged reference system. If
the fluid is sufficiently dense, the range of the attractive
forces exceeds the intermolecular separation resulting in
a uniform field with g(r) approaching 1. The integral in
eq 8 becomes van der Waals a constant. The second-order
term represents the fluctuations in the fluid structure,
caused by compression of the molecules into energetically
favorable locations by the attractive forces. This non-
randomness or clustering may affect & severely, with the
strongest impact at high compressibility. This was dem-
onstrated quantitatively with square-well clusters in a
mixture with hard spheres of identical size (Henderson,
1974). For highly compressible supercritical mixtures, it
would be desirable to shift the focus from cubic equations
of state, with oversimplified mixing rules, toward rigorous
statistical mechanical equations of state that describe this
nonrandomness (Chang et al., 1979).
The proposed AVDW model, which includes these sec-
ond-order effects, requires application of square-well fluid
results to complex molecules. This “scale-up” involved a
judicious balance between empirical and theoretical con-
siderations. An overly empirical model is not usually ex-
tendible to unmeasured systems, whereas a semitheoretical
one may allow for prediction. The empirical aspects of the
model are an attempt to compensate for the following
simplifying assumptions in the perturbation theory: (1)
the theory assumes spherical symmetry: (2) the internal
modes, along with intermolecular vibration and rotation
were neglected; and (3) a potential such as the Kihara or
6-12 does not represent the true potential.
Rogers and Prausnitz (1971) applied Barker-Henderson
perturbation theory to systems as complicated as meth-
ane-argon and methane-neopentane using a Kihara po-
tential and a kip In view of the sacrificesmade in the above
three-assumptions, especially for a system as complicated
as phenanthrene-ethylene, it is useful to shift from a
Kihara to a square-well potential. The square-well po-
tential, with an infinitely steep repulsive contribution,
allows for the advantage of an analytical AVDW expression
for the Helmholtz free energy. This expansion of the free
energy shown in eq 6 becomes analytical since the
square-well potential uncouples the temperature and
density effects on g
4 An
A/N=Ao/N+ E- (9)
n=l T**
T* = k T / e (10)
As explained by Alder (1972), A3 is significantly less than
194 Ind. Eng. Chem. Fundam., Vol. 21, No. 3, 1982

Table 11. Solid-Fluid Solubility Data

naphthalene- anthracene- anthracene- phenanthrene- phenanthrene-
ethane ethane carbon dioxide ethane carbon dioxide
P P P P P
( i o 5 Pa) 104y ( i o 5 Pa) 104y (io5 Pa) 104y ( i o s Pa) 104y (lo5 Pa) 104y
T = 20.0" T = 30.0" T = 30.0" T = 30.0" T = 30.0"
42.3 58.5 138.8 0.858 104.3 0.292 69.9 5.50 80.9 4.00
49.2 108 207.7 1.21 138.8 0.515 104.3 8.09 86.4 4.91
56.1 103 276.7 1.41 276.7 0.694 138.8 11.3 104.3 6.47
69.9 115 414.5 1.53 414.5 0.795 207.7 17.3 138.8 9.22
104.3 133 449.0 1.39 T = 50.0 242.2 17.9 242.2 12.2
138.8 135 483.5 1.81 276.7 18.3 276.7 12.3
173.3 140 90.6 0.035 380.1 14.2
T = 35.0 104.4 0.040 T = 40.0
T = 25.0" 414.5 14.6
104.4 0.768 118.1 0.161 138.8 24.0
56.1 99.6 221.5 1.21 138.8 0.496 207.8 37.4 T = 50.0
69.9 116 345.6 1.64 173.3 0.893 276.7 41.7 104.3 1.41
104.3 142 483.5 1.96 207.7 1.09 345.6 49.4 118.1 5.56
138.8 160 T = 50.0 276.7 1.40 T = 60.0 138.8 8.29
173.3 189 345.6 1.54 276.7 23.3
221.5 21 1 104.3 1.01 380.1 1.61 69.9 1.98 345.6 29.2
276.7 21 0 138.8 1.60 414.5 1.72 104.4 15.3 414.5 28.4
173.3 1.97 138.8 27.5
T = 35.0 207.7 2.03 T = 70.0 173.3 44.3 T = 70.0
35.4 1.8 242.2 2.47 118.1 0.142 207.8 46.2 104.3 0.228
63.0 151 276.7 2.90 129.2 0.222 242.2 50.5 118.1 1.68
83.7 189 311.1 3.05 138.8 0.358 414.5 75.3 138.8 5.09
104.3 225 345.6 3.33 152.6 0.630 207.7 16.3
138.8 263 414.5 3.84 173.3 0.706 phenanthrene- 276.7 35.0
207.7 319 449.0 3.72 207.7 1.68 ethylene 380.1 39.5
T = 45.0 T = 70.0 276.7 2.48 T = 25.0 414.5 41.3
345.6 3.09 276.8 0.249
49.2 5.6 138.8 2.38 414.5 3.49
56.1 14.3 173.3 3.58 290.6 0.269
63.0 103 207.7 3.43 311.3 0.277
69.9 118 242.2 5.41 T = 45.0
104.3 306 276.7 5.63 290.6 0.506
138.8 397 345.6 6.85 311.3 0.520
173.3 439 414.5 7.68
249.1 467 T = 70.0
311.3 1.12
triphenylmethane- triphenylmethane- hexamethylbenzene- hexamethylbenzene-
ethane carbon dioxide ethylene carbon dioxide
P P P P
( l o 5 Pa) 104y (10' Pa) 104y ( l o 5 Pa) 109 ( l o 5 Pa) 104y
T = 30.0" T = 30.0" T = 25.0 T = 30.0"
69.9 8.61 69.9 0.094 63.0 0.580 69.9 0.551
104.3 15.4 104.4 6.99 69.9 2.84 83.7 13.6
138.8 20.2 118.1 8.31 83.6 9.01 104.3 10.7
207.7 30.5 138.8 6.97 104.3 16.1 138.8 13.7
24 2.2 31.7 207.7 12.7 138.8 24.8 207.7 15.4
276.7 37.2 276.7 14.0 207.7 35.6 276.7 15.8
34 5.6 44.7 34 5.6 15.7 276.7 46.7 345.6 15.5
380.1 49.9 380.1 16.9 345.6 52.4 T = 50.0
T = 40.0 414.5 15.0 483.4 62.9 69.9 0.346
69.9 7.46 T = 40.0 T = 45.0 76.8 0.402
104.4 13.9 83.7 0.444 63.0 0.57 7 90.5 2.58
138.8 19.2 104.4 5.51 69.9 1.29 104.3 3.90
173.3 31.0 138.8 10.30 83.6 3.37 125.0 14.7
207.7 37.9 207.8 16.20 104.3 13.4 138.8 17.7
276.7 50.9 276.7 19.50 138.8 31.6 173.3 23.5
345.6 22.10 173.3 47.7 207.7 27.4
T = 50.0 32.0
T = 50.0 207.7 63.6 276.7
69.9 2.36 345.6 96.9 345.6 34.9
83.7 5.82 104.3 1.11
104.3 17.3 138.8 6.10 483.4 119 T = 70.0
138.8 29.8 173.3 11.70 T = 70.0 83.7 0.870
173.3 47.6 207.8 21.90 76.8 2.42 104.3 2.85
207.7 55.9 345.6 26.30 104.3 10.1 125.0 6.85
380.1 26.80 125.0 22.9 138.8 12.4
414.5 24.70 138.8 34.3 173.3 28.4
207.7 106 207.7 41.7
345.6 209 276.7 55.8
345.6 67.6
414.5 68.2
483.5 67.9
 Ind. Eng. Chem. Fundam., Vol. 21, No. 3, 1982 195

Table I1 (Continued)
fluorene-ethylene fluorene-carbon dioxide pyrene-eth ylene pyrene-carbon dioxide
P P P P
( l o s Pa) 109 ( l o 5 Pa) 104y (lo5 Pa) 104y ( l o 5 Pa) 109
T = 25.0 T = 30.0a T = 45.0 T = 35.0
76.8 3.80 83.6 9.11 104.3 0.604 83.6 0.303
104.3 12.4 104.3 11.7 138.8 2.37 104.3 0.796
138.8 18.0 125.0 13.9 173.3 3.36 125.0 1.19
207.7 28.8 138.8 14.8 207.7 5.77 138.8 1.34
345.6 44.0 207.7 18.6 276.7 8.01 207.7 2.10
483.4 54.5 345.6 21.6 345.6 11.3 34 5.6 2.56
T = 45.0 483.5 22.6 414.5 13.5 483.4 3.47
T = 35.0 483.6 15.9 T = 50.0
69.9 0.631
83.7 2.15 83.7 4.15 T = 75.0 104.3 0.153
104.3 8.85 104.3 11.8 83.7 0.076 125.0 0.666
138.8 20.7 138.8 15.3 138.8 2.11 138.8 1.23
207.7 44.4 207.7 22.6 207.7 7.78 207.7 2.79
345.6 80.2 345.6 24.3 276.7 14.1 276.7 3.75
483.4 112.0 414.5 27.2 345.6 20.8 345.6 4.48
T = 70.0 T = 50.0 414.5 28.2 483.4 5.40
483.5 33.6 T = 70.0
69.9 0.768 69.9 0.109
83.7 1.78 83.7 0.359 104.3 0.0270
104.3 5.10 104.3 4.28 125.0 0.320
125.0 14.8 138.8 15.4 138.8 0.562
138.8 23.1 173.3 23.9 173.3 1.89
187.0 61.3 207.7 30.8 207.7 3.15
207.7 76.6 276.7 38.7 276.7 5.54
27 6.7 140 345.6 40.8 414.5 8.62
345.6 203 414.5 43.6 483.4 9.38
414.5 257 T = 70.0
483.5 296
83.7 0.688
104.3 1.92
138.8 9.85
173.3 25.4
207.7 40.5
345.6 76.7
414.5 87.4
483.4 91.8
Liquid-solid isotherms.
in treating the asymmetricity as described in the CS-VDW
approach (Johnston and Eckert, 1981).
Application of t h e AVDW Model to Include the
Near-Supercritical Region
Differentiation of the free energy yields the chemical
potential

I I 1 I 1 where
50 55 60 65 70 75
AHyP ( kJ/gmol)
Figure 5. Cross-energy parameter q2w.solute heat of vaporization. The Carnahan-Starling repulsive contribution is
A2 except at conditiom extremely close to the critical point.
At these critical conditions, even A4 is large and the lim-
itation of 508 particles in the simulation cannot account
for the long-range fluctuations. In the present study, the The 42may be obtained for use in eq 3
free energy was truncated at second order (A2). In d2 = In 42r+ In - In z (14)
This Alder expression for the attractive forces has been
utilized in perturbed hard chain theory (PHCT) by Do- For example, in the CS-VDW model
nahue and Prausnitz (1978) and in the BACK equation
(Kreglewski and Chen, 1979). The limitations of the
BACK equation for large molecules in terms of the fits for
two binary parameters was explained, which is similar to A single effective b was used to avoid a solute size param-
the problem discussed above for kik The PHCT demon- eter that would be difficult to quantify. This was rea-
strates the necessity of going beyond first order for non- sonable since the repulsive term is significantly smaller
random mixtures. The intention in the proposed model than the attractive one. Furthermore, the use of two
is to incorporate the advantages of the Alder expression separate b parameters with the mixed hard-sphere model
for compressible fluids, while retaining the improvement of Mansoori et al. (1971) yielded similar results for the
196 Ind. Eng. Chem. Fundam., Vol. 21, No. 3, 1982

Table 111. Vapor Pressure, Heats of Vaporization, Cross-Energy Parameters, and % Uncertainties for Twelve Binary Systems
ethylene carbon dioxide
AH,'"~
vapor pressure constants a % %
melting point uncer- E,*/ uncer-
A B C ref kJ/mol ref T,"C k, K taintyC T,"C k, K taintyC
35 711
naphthalene 13.575 3729.3 0.0 h 50.8 h 25 717 b 45 709 b
25 813
hexamethyl- 13.134 3855.0 21.0 d 55.5 i 45 801 12 750o 765
742 5.6
benzene 70 784
25 839 35 826
fluorene 14.205 4561.8 0 I 62.2 1 45 825 15 50 807 12
70 802 70 783
50 816
anthracene 12.147 4397.6 0.0 f 62.2 g 70 50 812 12
85 802
800 9.0 7o 794
25 861
phenanthrene 14.631 4873.4 0.0 70.7 45 844 11 50 823 10
70 825 70 786
35 896
pyrene 77.1 45 905 14 50 87 5 12
75 884 70 853
a Vapor pressure equation: log Pst(Pa) = A - B / ( T - C), T (K). Most of the naphthalene data were above 1.5 mol %
and could not be modeled without eZ2. % uncertainity = ( l / N ) lyzicalc - y 2 i e x P / y Z i e x P Iwhere
, N = number of experi-
mental points. Ambrose et al. (1976). e Osborn and Douslin (1975);Bradley and Cleasby (1953). Wiedemann and
Vaughan 1970 . g Kudchadker et al. (1979). h Ambrose et al. (1975). ' MacDougall and Smith (1930). I Finke et al.
(1977). Smit)h et al. (1980).

solubility. A one-fluid theory was used for the attractive
parameter
= Y12€ll + 2YlY2€12 (16)
where the E~ term was eliminated for infinite dilution. The
results of the correlation suggested this ta was unnecessary
for solubilities below 1-2 mol %. Since only one size
parameter was used, this mixing rule would be essentially
the same for t or the van der Waals a.
The density fluctuations are described by the structure
factor or Fourier transform of the pair distribution function
s ( k )= 1 + p s exp(-ik.r) g(r) dr
At long wavelengths, or k = 0, S(k)is related to the
compressibility
Ro)= KT/K" GAS (18)
Ponce and Renon (1976) derived an analytical approxi-
mation for the second-order term that is directly propor-
tional to the hard-sphere compressibility. Notice that this
hard-sphere compressibility is independent of temperature,
thus limited for the temperature-dependent deviations
shown in Figure 2. Based on the implications of the above,
the actual experimental compressibility of the gas was
introduced. This modification for the second-order term
was expressed as KTIKT-, where this normalization factor
was the maximum KT at the given T. This term appears
to compensate for neglected nonrandomness in eq 16, to
include additional attraction at the low densities, and to
force the second order term to vanish at the high densities.
This approach appears reasonable based on fits of z vs. p
using eq 9 for pure ethylene and C 0 2 in a reduced density
range of 1-2. The resulta were comparable with the same
method without this factor; however, 2 is a great deal less
sensitive to p than &.
Using this factor, substitution of eq 9 and 16 into eq 11
yields
81.~2' = PCA1,(bpP(2t1z - + me) +
Table IV. Critical Properties, Energy and Size
Parameters, and Temperature Effects for the Solvent
be*,
pC, ell/ cm3/ d(elz/k)l
T,, K g-mol/L h , K g-mol dT
ethylene 282.4 7.75 310 28.0 0.75
carbon 304.2 10.6 333 21.0 1.3
dioxide
The universal constants A1,, A2, obtained from the sim-
ulation are given by Alder (1972). Further tests using
(17) another equivalent set of universal constants (Donohue et
al., 1978) gave similar results.
The unavailability of reliable vapor pressure data pre-
vented testing many of the systems in Tables I and I1
although considerable data acquisition is in progress. The
solids for which these data are reliable are shown in Table
I11 along with the AH" at the normal melting point. The
pure isothermal compressibility was obtained by numerical
differentiation of the density data using constant density
intervals. Again, for low solubility t / K was approximated
using e l l / K , which was obtained using eq 9, for z vs. p
isotherms (see Table IV). The t l l / K obtained was not
significantly affected by the inclusion of the KT/KT- factor.
The crucial unlike-pair energy t12was optimized using
linear least squares for each choice of b, until b was op-
timized (see Tables I11 and IV). The b was invariant for
a series of solids in a given gas, so the use of a b2 was
unnecessary. The model was used only for supercritical
fluid conditions above the critical density and temperature
of the solvent. The results for ethylene and carbon dioxide
shown in Figures 1-3, with the uncertainties shown in
Table I11 suggest that this method would be predictive if
t12 can be predicted. The cross-correlation with the solid
AH" at the normal melting point appears to fulfill this
requirement as shown in Figure 5. Notice that the en-
ergies me stronger in ethylene than in the nonhydrocarbon
COP The temperature dependence was fairly predictable
from T > TI, to about 350 K where
de12
€12 = t12(T = 323 K) - - ( T - 323 K) (20)
dT
 Ind. Eng. Chem. Fundam., Vol. 21, No. 3, 1982 197

Table V. Comparison of the CS-VDW Model, Augmented 0 = reference

VDW Model, and the Redlich-Kwong (RK) Equation r = repulsive
Using Critical Properties, Reported as Percent Uncertainty S = solid
cs- cs- sat = saturation
T.”C RK VDW VDW AVDW V = vapor
minimumplpc - 1.3 1.3 1.0 1.0 Literature Cited
Alder, 8. J.; Young, D. A.; Mark, M. A. J . Chem. Phys. 1972. 56, 6, 3013.
phenanthrene- 25 7.6 8.1 21 12 Ambrose, D.; Lawrenson, I. J.; Sprake, C. H. S. J. Chem. Thenno@.
ethylene 45 13 11 25 6.7 1975, 7, 1173.
70 30 11 17 14 Ambrose, D.; Lawrenson, I. J.; Sprake, C. H. S. J. Chem. Thermodyn.
1978, 8, 503.
anthracene- 50 38 9.1 9.1 7.5 Bradley, R. S.; Cleasby. T. G. J. Chem. Soc. 1953, 1690.
ethylene 70 24 14 17 7.3 Chang, J. I. C.; Hwu, F. S. S.; Leland, T. W. A&. Chem. Ser. 1979, No.
85 17 12 22 12 182, 71.
DeFlllppi, R. P.; Kyukonis, V. J.; Robey, R. J.; Modell. M., EPA Report-60012-
80-054, March 1980.
The slope shown in Table V was primarily related to the Dlepen, 0. A.; Scheffer, F. E. C. J . Am. Chem. SOC. 1948, 70. 4065.
solvent. Din, F. “ T v n a m i c s Functbns of Gases”;Butterworths: London, Vol. 1.
1962; Voi. 2, 1962; Vol. 3, 1961.
In Table V, the first two columns compare the corre- Donohue, M. D.; Prausnltz, J. M. A I M J. 1978, 24, 849.
sponding states treatment using critical properties Douslin, D. R.; Harrison, R. H. J. Chem. Thermodyn. 1978. 8 , 301.
(Mackay et al., 1979) to the CS-VDW model, in which the Elsenbeiss, J.; Contract No. OAl&lO&AMG244(A) SWRI Project No. 01-
1392, southwest Research Instme, San Antonio, TX, 1964.
latter has the advantage of being predictive. A 10% un- EwaM, A. H. Thesis, Manchester, U.K., 1952.
certainty is not far from the experimental uncertainty and Ewald, A. J.; Jepson, W. 6.; Rowilnson, J. S. Discuss. Faraday SOC.,1953,
75. 238.
is desirable considering the range of over two orders of Finke, H. L.; Messerly, J. F.; Lee, S. H.;Osborn, A. 0.; Douslln. D. R. J.
magnitude. However, the CS-VDW model is leas s u ~Chem. Thermodyn.
~ 1977. 9 , 937.
when the reduced density range is lowered from about 1.5 Hannay, J.; Hogarth. J. Roc. R . Soc.London Ser. A 187% 29#324.
Hansen, J. P.; McDonald, I. R. “Theory of Simple Liqulds”; Academic Press:
to 1 as shown in the third column. The AVDW model New York, 1976.
accomplished this needed extension to a reduced density Henderson, D. Ann. Rev. phvs. Chem. 1974, 461.
Henderson, D. J. Chem. phvs. 1974. 6 1 , 3, 926.
of 1.0, without any new adjustable parameters. Johnston, K. P. Master’s Thesis, University of Illlnois, Urbana, 1979.
Nomenclature Johnston, K. P.; Eckert, C. A. A I C M J. 1981, 27, 773.
Kaul, 8. K.; Prausnitz, J. M. Ind. Eng. Chem. Fundam. 1977, 76, 335.
AI,, Aam= universal constants from Alder’s results Kohn, P. M.; Savage, P. R. Chem. Eng., 1979, 86(6), 41.
E = enhancement factor Kreglewskl, A.; Chen, S. S. A&. Chem. Ser. 1979. No. 182, 197.
Kudchadker, S. A.; Kudchacker, A. P.; Zwolinski, B. J. J. Chem. Thennodyn.
g ( r ) = radial distribution function 1979, 1 1 , 1051.
k = wave vector Kurnlk. R. T.; Holla, S. J.; Reid, R. C. J. Chem. Eng. Data 1981, 26. 47.
kij = deviation from geometric mean temperature mixing rule Mackay, M. E.; Paulaitls, M. E. Ind. Eng. Chem. Fundam. 1979, 78, 149.
MacDougall, F. H.; Smith, L. I . J. Am. Chem. SOC. 1930, 52, 1998.
m = index for constant in polynomial fit of AI,, A z , Mansoori, G. A.; Carnahan, N. F.; Starling, K. E.; Leland, T. W. J. Chem.
S(k)= structure factor Phys. 1971, 54, 1523.
u = molar volume McHugh. M.; Paulaitls, M. E. J. Chem. Eng. Data 1980, 25, 326.
0 = partial molar volume Michels, A.; Mlchels, C. Roc. R. Soc. London. Ser. A 1935, 753, 201.
Mlchels, A.; Michels, C.; Wouters, H. Roc. R. Soc. London, Ser. A 1935,
w = perturbation potential 153, 214.
WN = total perturbation energy Mlchels, A.; Blaisse, B.; Michels, C. froc. R . Soc. London, Ser. A 1937,
x = mole fraction liquid 760, 358.
y = mole fraction vapor Osborn, A. G.;Douslln, D. R. J . Chem. Eng. Data 1975, 20, 229.
Pauialtis, M. E.; Johnston, K. P.; Eckert, C. A. J. Phys. Chem. 1981, 8 5 ,
z = compressibility factor 1770.
Ponce, L.; Renon, H. J. Chem. H y s . 1978, 6 4 , 2, 638.
Greek Letters Robin, S.; Vodar, B. DISCUSS. Farahy Soc. 1953, 15, 233.
@ = l / k T , where k is Boltzmann’s constant Rogers, B.; Prausnitz, J. Trans. Faraday SOC. 1971, 6 7 , 3474.
Smlth, N. K.; Stewart, R. C.; Osborn, A. G.; Scoit, D. W. J. Chem. Thermo-
y = activity coefficient e n . 1980, 12, 919.
e = energy parameter in square-well potential Todd, D.; Elgln. J. AIChE J. 1955, 1 , 20.
K = isothermal compressibility,-l/ V ( d V / a ~ ) ~ Tsekhanskaya, Y. V.; Iomtev, M. 6.; Mushkina, E. V. Russ. J. Phys. Chem.
A = coupling parameter (goes from 0 to 1) 1982, 36. 1177.
Tsekhanskaya, Y. V.; Iomtev, M. B.; Mushkina, E. V. Russ. J. Phys. Chem.
{ = reduced density 1984, 38, 1173.
4 = fugacity coefficient Van Leer, R. A.; Paulaitis, M. E. J . Chem. Eng. Data 1980. 25. 257.
Wiedemann, H. 0.;Vaughan. H. P. R o c . Toronto Symp. . . Therm. Anal. 3rd
Subscripts 1970, 233.
1 = solvent or first order Zlger, D. H. Master’s Thesis, University of Illinois, Urbana, 1980.
2 = solute or second order
ij = components Received for review June 3, 1981
n = order, i.e., first to fourth order Accepted November 30, 1981
0 = reference
This work was supported by the Environmental Protection Agency
Superscripts under U.S.EPA Cooperative Agreement CR806819 and by grants
a = attractive from Phillips Petroleum Company and Arc0 Oil and Gas Com-
L = liquid pany.