R&D NOTE ~ c @

Solubilities of Anthracene, Fluoranthene and

Pyrene in Organic Solvents: Comparison of
Calculated Values using UNIFAC and Modified
UNIFAC (Dortmund) Models with
Experimental Data and Values using
the Mobile Order Theory
Henrik K. Hansenl*,Carmen Rivero12 and William E. Acree Jr.3
‘Department of Geology and Geotechnical Engineering, Technical University of Denmark, 2800 Lyngby, Denmark
2Department of Chemical Engineering, University of Vigo, 36200 Vigo, Spain
3Department of Chemistry, University of North Texas, Denton, TX 76203-5068, USA

I
n the design of separation processes, it is of vital importance to know
the solubility of reagents and products. If the solubility is wrongly
estimated it could cause undesirable fouling and crystallization in
process equipment. Especially in the petroleum industry and similar
industry with heavy feedstock it is necesary to avoid that the flow lines
will be plugged.
Polycyclic aromatic compounds (PAH) (such as anthracene, fluoran-
thene, and pyrene) are of great importance to the petroleum industry,
and at the same time known to be carcinogenous. Therefore, crystalliza-
tion of these compounds must be avoided in chemical processes. Data
for the solubility of such compounds given in the literature are scarce,
and normally only available a t 25°C. In order to estimate the solubility
for these polycyclic aromatic compounds one can use predicting proce-
dures. Due to the fact that ideal solubility only is reasonable to accept in
systems with solutes of the same chemistry as the solvents, it is in most
cases necessary to find methods to calculate the activity coefficient of the
solute. To address this problem, researchers have turned to group contri-
butions methods and semiempirical expressions to predict the activity
coefficient. Even though the group contribution method of UNIFAC
(Fredenslund et al., 1975, 1977) earlier has shown its potential for calcu-
lating solid-liquid equilibria (Cmehling et al., 1978), no thorough work
using UNIFAC to estimate solubilities for PAHs has been published yet.
This is interesting especially after that both the UNIFAC (Hansen et al.,
1991) and the Modified UNIFAC (Dortmund) (Weidlich and Cmehling,
1987, Cmehling et al., 1998) have been revised recently, and are now
covering a wide range of functional groups.
The objective of this work is to analyze the predictability of UNIFAC
and modified UNIFAC (Dortmund) models to calculate solubilities of
anthracene, fluoranthene and pyrene in a series of organic solvents. A
comparison between the computed solubilities obtained here and values
calculated by the Mobile Order Theory (Acree, 1999; Roy et al., 1999;
Hernandez and Acree, 1998; Powell et al., 1994) will be given.
The solubility of anthracene in 43 organic solvents,
fluoranthene (45 solvents) and pyrene (30 solvents)
has been calculated using UNIFAC and Modified
UNIFAC (Dortmund) models to estimate the activity
coefficient of the solute. It was found that both
UNIFAC and Modified UNIFAC described better the
solubilities in polar solvents like alcohols, ketones,
esters and ethers than in nonpolar solvents like alkanes
and aromatic hydrocarbons. UNIFAC and the Mobile
Order Theory supplement each other well in calculat-
ing the solubilities, which means that one can choose
the right model depending on the solvent one is
using.
La solubilite de I’anthracene dans 43 solvants
organiques, du fluoranthhe (45 solvanh) et du
pyrhe (30 solvants) a 6te calculee B I’aide des
modeles UNIFAC et UNIFAC modifit5 (Dortmund) afin
d’estimer le coefficient d’activite du solut6. On a
trouve que ces deux derniers modeles decrivaient
mieux les solubilites dans les solvants polaires tels les
alcools, les chtones, les esters et les ethers, que dans
les solvants non polaires tels les alcanes et les hydro-
carbures aromatiques. L’UNIFAC et la theorie de
I’ordre mobile se completent bien pour calculer les
solubilites, ce qui signifie que I‘on peut choisir le
modele adequat selon le solvant utilise.
Keywords: solubility, polycyclic aromatic hydrocar-
bons, UNIFAC, modified UNIFAC, mobile order
theory.

Thermodynamic Background
A pure solid solute (2) partly dissolves in a liquid solvent (1). The
equation of solid-liquid phase equilibrium for the solute is given by:

*Author to whom correspondence may be addressed. E-mail address: hkhansen@ It is given elsewhere (Prausnitz, 1969) that the
kemi.dhr.dk fugacity ratio of the pure solute can be derived to:

1168 The Canadian Journal of Chemical Engineering, Volume 78, December 2000
In($) pure 2 = 2(+ - + !$ (1 - +)
$. - r.x.
The second and third terms on the righthand side are - CfiXi
addressed to differences in the heat capacity and are often small i
and of opposite sign. Therefore, they can be neglected. The last
term on the righthand side accounts for the pressure effect, and
vanishes at low pressures. Equation (2) reduces hereby to:

For the liquid phase we can write:

The residual part is given by:

where x , is the mole fraction of the dute, is the ctivity

1,

coefficiek of the solute and fLpure is the fLgacity of pure,

subcooled liquid ( 2 ) a t the system temperature. where
When assuming that there are no solubility of component 1 r 1
in the solid phase, we can combine Equations (3) and (4) and
obtain the normally used equation for solid-liquid equilibrium:

y2 is normally different from unity. Group contributions

methods such as UNIFAC and Modified UNIFAC (Dortmund)
are tools to estimate the activity coefficient knowing only the
functional groups of the solute and solvents, and their interac-
tion parameters.

UNIFAC Model
The activity coefficient using UNIFAC is considered to consist of i n
two parts: a combinatorial contribution, due to differences in
molecular size and shape, and a residual contribution, due to
differences in intermolecula forces of attraction between the
functional groups:

In(yi) = In(yf) + In(yP) (6)

Modified UNIFAC (Dortmund) Model
The combinatorial part is given by: The Modified UNIFAC (Dortmund) model was developed in
order to have a better estimate of the activity coefficient in
larger temperature ranges than the original UNIFAC model, and
(7) a t the same time be able to predict vapour-liquid equilibria,
enthalpies of mixing, activity coefficients a t infinite solution,
liquid-liquid equilibria, and excess heat capacities. The
where modified UNIFAC (Dortmund) differs from UNIFAC only in the
following terms:

(19 )
z = 10 (9)

The Canadian Journal of Chemical Engineering, Volume 78, December 2000 1169
where Results and Discussion
,314 Table 1 gives the calulated solubilities of anthracene using
UNIFAC and modified UNIFAC (Dortmund) models in a series of
organic solvents together with the experimental values found
by Roy et al. (1999). Looking a t the estimated solubility data
one must take into account that many of these particular
For the residual part the interaction parameters are given by: systems are highly nonideal, and that the experimental solubil-
ity data cover over a 1000-fold range in mole fraction. If one
a, + bnmT+ cnmT2 could predict the value to within a factor of two or three that
would be adequate in many applications where only a rough
value is needed. Furthermore, when predicting absolute solubil-
ities, there is some uncertainty associated with assumed that the
The rest of the terms from UNIFAC is the same in Modified enthalpy of fusion is independent of temperature. This could
UNIFAC (Dortmund). lead to errors in the case of anthracene and other high melting
point solutes. Therefore it can generally be said that the two
Mobile Order Theory UNIFAC models give solubilities that are within the acceptable.
The Mobile Order Theory (Huyskens and Siegel, 1988; Siegel This is in accordance with Cmehling et al. (1978) where the
et al., 1990; Huyskens, 1992) assumes that all molecular groups solubility of anthracene was determined in 12 solvents using
perpetually move, and that neighbours of a given kind of exter- "old" UNIFAC interaction parameters. Modified UNIFAC
nal atom in a molecule constantly change identity. The Mobile (Dortmund) does not predict the solubility well in the very polar
Order Theory gives the volume fraction saturation solubility, solvent acetonitrile. The ideal solubility (y2 = 1) for anthracene
@pt, for a crystalline solute, A, as: is 0.01049 for all solvents, and it can be seen that only in
systems containing some of the chloronated solvents ideal
solubility can be assumed to exist.
Looking closer on the calculated values, UNIFAC gives good
estimates for anthracene in alcohols, esters and ethers - all
polar solvents. The solubilities of anthracene in alkanes or
aromatic hydrocarbons do not seem to be as good as wanted.
In these systems the functional groups are only -CH,- and ACH
(Fredenslund e t al., 1977), and therefore UNIFAC does not
contribute with sufficient parameters to predict the interaction
well. In the systems with polar groups more interaction parame-
ters are taken into account, and therefore a better estimate of
where y2 is found.
Table 2 shows the calculated values for the solubility of
Ksolvent @solvent fluoranthene in various organic solvents together with the
experimental values and values estimated following the Mobile
Vsolvent
= Order Theory taken from Hernindez and Acree (1998). The
(solvent
1+ Kso/vent@so/vent
Vsolvent I (23)
picture is the same as with anthracene: UNIFAC gives reason-
able estimates for polar solvents like alcohols, ketones, ethers
and esters. But in systems with alkanes or aromatic hydrocar-
bons UNIFAC fails to give exact activity coefficients. Using the
Modified UNIFAC the solubilities of fluoranthene in alkanes is
and calculated satisfactorily, but in benzene and toluene the calcu-
lated values are much lower than the experimental ones. For
lower alcohols the prediction is poor, but the prediction gets
better with increasing the number of carbon atoms in the
alcohol. The Mobile Order Theory, on the other hand, gives
precise solubilities for systems with alkanes and aromatics,
If complexation does occur between the crystalline solute whereas the prediction of the solubilities in polar solvents is
and the solvent, then: poor. The values calculated by the Mobile Order Theory
is based on the modified solubility parameter gA, which is
optimized on hexane, heptane and octane. This gives the
reason why the prediction of solubilities in alkanes is very exact
using this model. At present time, one can predict PAH solubil-
ities in more solvents with the UNIFAC and Modified UNIFAC
methods than with Mobile Order Theory. The modified solubil-
ity parameters needed in the Mobile Order theory calculations
are available for only a very limited number of organic solvents.
Also, modified solubility parameters are available for only a
small number of the polycyclic aromatic hydrocarbon solutes
such as fluoranthene. The ideal solubility for fluoranthene is

1170 The Canadian Journal of Chemical Engineering, Volume 78, December 2000
 Table 1. Solubilities of anthracene in various solvents at 25°C.

Error
Error Mobile Mobile
Error Mod. Mod. Order Order
Exp. Value UNIFAC UNIFAC UNIFAC UNIFAC Theory Theory
Solvent x,exp XZmk % XZCalC YO XZmk %

Hexane 0.001 290 0.002380 84.5 0,001 585 22.9 0.001 424 10.4
Heptane 0.001 571 0.002407 53.2 0.001 789 13.9 0.001 519 -3.3
Octane 0.001 850 0.002438 31.8 0.002385 28.9 0.001 729 -6.5
Decane 0.002345 0.002556 9.0 0.002690 14.7 0.0021 13 -9.9
Hexadecane 0.00380 0.002920 -23.1 0.003309 12.9 0.00301 5 -20.7
2,2,4-Trimethyl pentane 0.001 087 0.00221 6 103.9 0.001 902 75.0 0.001 179 8.5
Cyclohexane 0.001574 0.003780 140.1 0.001 259 -20.0 0.001 746 10.9
Methylcyclohexane 0.001 65 0.003480 110.9 0.001 52 -7.9 0.001 919 16.3
Cyclooctane 0.002258 0.003328 47.4 0.001 131 -49.9 0.002451 8.5
Benzene 0.00741 8 0.01 234 66.3 0.0051 16 -31 .O 0.01016 37.0
Toluene 0.00736 0.01 050 42.7 0.0081 76 11.1 0.008090 9.9
m-Xylene 0.007956 0.00781 1 -1.8 0.001 703 -84.5 0.005930 -25.5
p-Xylene 0.00733 0.00781 1 6.6 0.001 703 -76.8 0.006149 -1 6.1
Methanol 0.000243 0.00041 17 69.4 0.0001 95 -1 9.8 0.000492 102.5
Ethanol 0.000460 0.0002837 -38.3 0.0005 69 23.7 0.000762 65.7
1-Propanot 0.000591 0.0004703 -20.4 0.000508 -1 4.0 0.000957 61.9
2-Propanol 0.000411 0.0004745 15.4 0.00051 0 24.1 0,001104 168.6
1-6utanol 0.000801 0.0006638 -1 7.1 0.000792 -1.1 0.001 212 51.3
2-Butanol 0.000585 0.0006685 14.3 0.000434 -25.8 0.000965 65.5
2-Methyl-1-propanol 0.000470 0.0003453 -26.5 0.000473 0.6 0.000778 65.5
1-Pentanol 0.001 097 0.0008576 -21.8 0.000904 -1 7.6 0.001 316 20.0
1-Hexanol 0.001483 0.001 057 -28.7 0.00091 8 -38.1 0.001 266 -1 4.6
1-Heptanol 0.001 869 0.001 248 -33.2 0.000996 -46.7 0.001439 -23.0
1-0ctanol 0.0021 60 0.001 434 -33.6 0.001 524 -29.4 0.001 598 -26.0
Acetonea 0.0031 0.003095 0.2 0.00220 -29.0
2-Butanoneb 0.00477 0.004454 -6.6 0.0031 24 -34.5
Diethyl etheP 0.0029 0.00451 8 55.7 0.00401 2 38.3
Dibutyl ether 0.00354 0.002899 -1 8.1 0.002905 -1 7.9 0.006433 81.7
1,4-Dioxane 0.008381 0.008332 -0.6 0.007889 -5.9 0.01 128 34.6
Ethylacetate 0.00484 0.005086 5.1 0.005392 11.4 0.01 088 124.8
Butylacetate 0.00661 0.005808 -1 2.1 0.0031 15 -52.9 0.01 026 55.2
Diethyladipate 0.01033 0.006220 -39.7 0.007533 -27.1 0.0081 18 -21.4
Dichloromethane 0.009387 0.02225 137.0 0.006327 -32.6 0.01 316 40.2
Trichloromethane 0.01 084 0.02084 92.2 0.0081 27 -25.0 0.01 024 -5.5
Tetrachloromethane 0.00464 0.006433 38.6 0.002243 -51.7 0.005843 25.9
1-Chlorobutane 0.00586 0.006435 9.8 0.003310 -43.5 0.00591 0 0.9
1,4-Dichlorobutane 0.01053 0.008537 -1 8.9 0.007320 -30.5 0.01 055 0.2
1-Chlorooctane 0.007780 0.004677 -39.9 0.003721 -52.2 0.007628 -1.8
Chlorocyclohexane 0.006353 0.006986 10.0 0.006621 4.2 0.00871 3 37.1
Chlorobenzene 0.009962 0.009659 -3.0 0.001 580 -84.1 0.01 049 5.3
Ethylene glycol 0.000071 5 0.00001 91 -73.3 0.0000528 -26.2 0.0000693 -3.1
Acetonitrile 0.000830 0.001447 74.3 0.002693 224 0.008202 888
N,N-Dimethylformamide 0.007839 0.003335 -57.4 0.001427 -81.8 0.009891 26.2
Overall abs. deviation 40.2 36.4 55.0
Overall abs. deviation 37.8 31.9 33.6
Without worst result

Error (%) = 100 [XZcoIc- X Z e x ~ ] / X Z e X pExperimental

. data from Roy et al. (1 999). (I Fohl et al. (1966), 20°C. Cepeda et al. (1996).

0.1 912, which differs from all experimental values found by
Hernandez and Acree (1 998).
Table 3 lists the solubilities of pyrene in a series of organic
solvents calculated by UNIFAC and Modified UNIFAC
(Dortmund). Experimental values and values calculated by the
Mobile Order Theory by Powell et al. (1994) are given too.
Again it is seen that UNIFAC generally gives good estimates for
the solubility of pyrene in polar solvents, but deviation from the
experimental values is found for solvents like alkanes and
aromatic hydrocarbons. It must be noted that the UNIFAC
description (Fredenslund et al., 1977) of pyrene is chosen to be
10 ACH and 6 AC, which is exactly the same as the fluoranthene
description. Only differences in the enthalpy and temperature
of fusion between the two solutes account for differences in the

The Canadian Journal of Chemical Engineering, Volume 78, December 2000 1171
 Table 2. Solubilities of fluoranthene in various solvents at 25°C.

Error
Error Mobile Mobile
Error Modified Modified Order Order
Exp. Value UNIFAC UNIFAC UNIFAC UNIFAC Theory Theory
Solvent X,exp Xpk % x,co'c 96 X2mlc %

n-Hexane 0.01476 0.03270 121.5 0.01 367 -7.4 0.01686 14.2

n-Heptane 0.01 870 0.03296 76.2 0.01 623 -1 3.2 0.01 797 -3.9
n-Octane 0.02260 0.03325 47.1 0.01 779 -21.3 0.02080 -8.0
n-Nonane 0.02642 0.03362 27.2 0.01 94 -26.6 0.02466 -6.7
Decane 0.0301 5 0.03408 11.5 0.0269 -1 0.8 0.02599 -1 3.8
Hexadecane 0.05046 0.03954 -21.6 0.0450 -1 0.8 0.03800 -25.7
Cyclohexane 0.01 807 0.06865 279.9 0.01 313 -27.3 0.02246 24.3
Methylcyclohexane 0.021 79 0.0691 5 217.3 0.01 557 -28.5 0.02460 12.9
Cyclooctane 0.0301 1 0.05280 73.4 0.0229 -23.9 0.03327 10.5
2,2,4-Trimethylpentane 0.01 162 0.02928 152.0 0.01 804 55.2 0.01 297 11.6
t-8utylcyclohexane 0.02482 0.04346 75.1 0.03290 32.5 0.03429 38.2
Benzene 0.1 21 1 0.2045 68.9 0.01 052 -91.3 0.1 814 49.8
Toluene 0.1 160 0.1 818 56.7 0.01 897 -83.6 0.1465 26.3
Methanol 0.00267 0.004822 80.6 0.0071 35 167.0 0.00745 179.0
Ethanol 0.00544 0.002673 -50.8 0.00991 5 82.3 0.01 081 98.7
1-Propano1 0.00670 0.004776 -28.7 0.0051 5 -23.1 0.01 323 97.5
2-Propanol 0.00475 0.004831 1.7 0.00566 19.2 0.01 582 233.1
1-8utanol 0.00996 0.007070 -29.0 0.00712 28.5 0.01 680 68.9
2-8utanol 0.00702 0.007136 1.7 0.00799 13.8 0.01253 78.5
2-Methyl-1-propano1 0.00495 0.007136 44.6 0.021 80 340.4 0.00955 92.9
1-Pentan01 0.01446 0.009436 -34.7 0.01 188 -1 7.8 0.01 779 23.0
2-Pentanol 0.01 021 0.00951 0 -6.9 0.01 396 36.7
3-Methyl-1-butanol 0.00862 0.00951 0 10.3 0.00629 -27.0
2-Methyl-2-butanol 0,00970 0.007805 -1 9.5 0.00875 -9.8
Cyclopentanol 0.01 772 0.01477 -1 6.6 0.01 663 -6.1
1-Hexanol 0.01 986 0.01 182 -40.5 0.01 612 -1 8.8 0.01 631 -1 7.9
2-Methyl-1-pentan01 0.01 172 0.01 189 1.4 0.01 123 -4.2
4-Methyl-2-pentanol 0.00948 0.01 197 26.3 0.0091 4 -3.6
1-Heptanol 0.02524 0.01 418 -43.8 0.01 843 -27.0 0.01 860 -26.3
1-0ctanol 0.03125 0.01 652 -47.1 0.0251 2 -1 9.6 0.02068 -33.8
2-Ethyl-1-hexanol 0.01 782 0.01660 -6.8 0.01 55 -1 3.0
Dibutylether 0.051 77 0.06310 21.9 0.0562 8.6 0.1058 104.4
Ethyl acetate 0.08589 0.08484 -1.2 0.0794 -7.6 0.1 991 131.8
Butyl acetate 0.1106 0.08 7 79 -20.6 0.0843 -23.8 0.1 827 65.2
Tetrachloromethane 0.081 57 0.1159 42.0 0.081 36 -0.3
Trichlorornethane 0.1 426 0.2344 64.3 0.1 260 -1 1.6
1,2-DichIoroethane 0.1268 0.1603 26.4 0.0998 -21.3
1-Chlorobutane 0.1 133 0.06469 -42.9 0.0634 -44.0 0.1 058 -6.6
1-Chlorohexane 0.1 315 0.08051 -38.8 0.1 261 -4.1
1-Chlorooctane 0.1 367 0.05376 -66.7 a0692 -49.4 0.1291 -5.5
Chlorocyclohexane 0.1 159 0.1 225 5.7 0.0901 2 -22.2 0.1 540 32.9
Ethylene glycol 0.000750 0.0001 73 -76.9 0.000520 -30.7
N,N-Dimethylacetamide 0.2370 0.1 602 -32.4 0.1 621 -31.6
N,N-Dimethylforrnarnide 0.1 797 0.05247 -70.8 0.1 320 -26.5
Acetonitrile 0.01 315 0.02314 76.0 0.00461 7 -64.9 0.1971 1399.2
Overall abs. deviation 51.2 36.4 94.9
Overall abs. deviation 46.0 29.5 51.4
Without worst result

Error (94) = 100 [X,colc - X2wq/X2exP. Experimental data from Hernhndez et al. (1998).

estimated activity coefficients. The Modified UNIFAC predicts
the solubility satisfactorily for all pyrene containing systems
investigated in this work except for acetone. For the alcohols
again Modified UNIFAC gives better solubilities for higher
number of carbon atoms in the molecule. The Mobile Order
Theory on the other hand gives good estimates for solubilities
in alkanes and aromatic solvents, but poor prediction in polar
solvents. In chloronated solvents the solubility of pyrene can be
said to be in the order of the experimental value for all three
models. The ideal solubility of pyrene is 0.1 31 2, which is far
from the experimental found solubilities (Powell et al., 1994) for
all solvents except 1,4-dichIorobutane.

1172 The Canadian Journal of Chemical Engineering, Volume 78, December 2000
 Table 3. Solubilities of pyrene in various solvents a t 26°C.
~~

Error
Error Mobile Mobile
Error Modified Mod. Order Order
Exp. Value UNIFAC UNIFAC UNIFAC UNIFAC Theory Theory
Solvent X,.XP XZCOIC % XZCOIC % xzcarc %

Hexane 0.00852 0.01 921 125.5 0.00664 -22.1 0.00998 17.1

Heptane 0.01 101 0.02051 86.3 0.00740 -32.8 0.01068 -3.0
Octane 0.01 379 0.021 93 59.0 0.00823 -40.3 0.01 243 -9.9
Cyclohexane 0.01 089 0.04290 293.9 0.00927 -14.9 0.01 323 21.5
Methyl cyclohexane 0.01 300 0.03739 187.6 0.001 52 -88.3 0.01 460 12.3
2,2,4 Trimethylpentane 0.00721 0.01 928 124.2 0.00721 0.0 0.00769 6.7
Cyclooctane 0.01 956 0.03405 74.1 0.00991 49.8 0.01 987 1.6
t-Butylcyclohexane 0.01 590 0.04364 174.5 0.0214 34.6 0.02070 30.2
Benzene 0.0631 6 0.1416 124.2 0.0321 3 49.1 0.1 221 7 93.4
Toluene 0.06785 0.1223 80.2 0.02785 -59.0 0.09409 38.7
m-Xylene 0.07055 0.09222 30.7 0.0465 -34.1 0.06323 -10.4
pXylene 0.06831 0.09222 35.0 0.041 3 -39.5 0.0661 2 -3.2
Methanol 0.001 49 0.003214 115.7 0.000973 -34.7 0.00455 205.4
Ethanol 0.0031 7 0.001 81 7 -42.7 0.00535 68.8 0.00655 106.6
1-Propano1 0.00426 0.003233 -24.1 0.00576 35.2 0.00802 88.3
2-Propanol 0.00290 0.003269 12.7 0.00462 59.3 0.00964 232.4
1-Butanol 0.00622 0.004774 -23.2 0.00587 -5.6 0.01 021 64.1
2-Butanol 0.00433 0.00481 6 11.2 0.00433 0.0 0.00756 74.6
2-Methyl-1-propano1 0.00319 0.00481 6 51.o 0.00432 35.4 0.00573 79.6
1-Pentanol 0.00926 0.006360 -31.3 0.0061 0 -34.1 0.01 081 16.8
1-0ctanol 0.02097 0.01 111 47. 0 0.00845 -59.7 0.01 259 -40.0
Acetone 0.0361 2 0.0341 3 -5.5 0.001 09 -97.0 0.14379 298.1
Dibutyl ether 0.02980 0.041 02 37.6 0.03841 28.9 0.0665 123.2
1,CDioxane 0.03520 0.09747 176.9 0.0453 28.7 0.14314 306.6
Ethyl acetate 0.04251 0.051 80 21.8 0.0294 -30.8 0.1 3864 226.1
Butyl acetate 0.05932 0.05860 -1.2 0.04530 -23.6 0.1 2378 108.7
Carbon tetrachloride 0.04229 0.071 63 69.4 0.0431 0 1.9 0.06472 53.0
1,2-DichIoroethane 0.08746 0.1 022 16.8 0.0692 -20.9 0.1 4649 67.5
1-Chlorobutane 0.06094 0.06952 14.1 0.0553 -9.3 0.06470 6.2
1,4-DichIorobutane 0.1 097 0.08820 -1 8.7 0.0794 -27.6 0.1 286 17.2
Overall abs. deviation 70.6 35.2 78.7
Overall abs. deviation
Without worst result 62.9 33.3 70.8
~

Error (%) = 100 [XZCok - X,exP]/X2exP. Experimental data from Powell et al. (1994).

Both the UNIFAC and Modified UNIFAC methods predict Modified UNIFAC (Dortmund) to estimate the activity coeffi-
activity coefficients, not saturation solubilities per se. Solubilities cient of the solute.
are related t o activity coefficients through Equation (5). Any Based on the calculations of the solubilities of the three PAHs
errors resulting from neclect of the heat capacity terms in it can generally be said that the UNIFAC models give good
Equation (2) would affect the UNIFAC and Modified UNIFAC predicted solubilities of PAH’s in polar organic solvents like
methods more than Mobile Order Theory. In using Mobile alcohols, ketones, ethers and esters.
Order Theory, the numerical value of the solute’s modified The Mobile Order Theory predicts well the solubilities of
solubility parameter is calculated assuming that the heat capac- PAH’s in alkanes and aromatic hydrocarbons.
ity terms in Equation (2) are neglected. This eliminates from the UNIFAC and the Mobile Order Theory supplement each
Mobile Order theory predictions some of the errors/uncertain- other well, which means that one can choose the right model
ties associated with assuming that the heat capacities are depending on the solvent one is using.
negligible.
Generally, the calculations presented in this work indicate Nomenclature
that the three models supplement each other well, each with U activity
their optimal type of solvents. a,, UNIFAC group interaction parameter between groups n and
m, (K)
b,, Modified UNIFAC interaction parameters between groups n
Conclusions and rn
The solubility of anthracene, fluoranthene and pyrene in various Modified UNIFAC interaction parameters between groups n
Cnm
organic solvents has been calculated using UNIFAC and and m, (K-’)

The Canadian Journal of Chemical Engineering, Volume 78, December 2000 1173
 CP heat capacity, (Jmol-’.K-’) Fohl, I., M. Fris and M. Smutek, ”Kristallisationsgleichgewichte ill.
f fugacity, (Pa) Loslichkeiten von Anthracen und Carbazol in einigen reinen
Fi auxiliary property for component i, (surface fraction/mole Losungsmitteln”, Collect. Czech. Chem. Commun. 31, 401 5-4031
fraction) (1966).
enthalpy of mixing, (jmol-’) Fredenslund, Aa., R.L. Jones and J.M. Prausnitz, “Croup Contribution
,,:
Hm
solute - solvent equilibrium constant
see Equation (8)
Estimation of Activity Coefficient in Nonideal Liquid Mixtures”, AlChE
J. 21, 1086-1099 (1975).
P pressure, (Pa) Fredenslund, Aa., J. Gmehling and P. Rasmussen, “Vapor-Liquid
dimensionless van der Waals surface area Equilibria using UNIFAC”, Elsevier, Amsterdam, The Netherlands
9
van der Waals surface area of group 1977.
Qk Cmehling, J.G., T.F. Anderson and J.M. Prausnitz, “Solid-Liquid
r dimensionless van der Waals volume
see Equation (23) Equilibria using UNIFAC”, Ind. Eng. Chem. Fundam. 17, 269-273
gas constant = 8.31 4 kJ mol-’ .K-’ (1978).
absolute temperature, (K) Cmehling, J., J. Lohmann, A. Jakob, J.D. Li and R. Joh, ”A Modified
energy parameter, (jmol-1) UNIFAC (Dortmund) Model. 3. Revision and Extension”, Ind. Eng.
auxiliary property for component i (volume fraction/mole Chem. Res. 37,4876-4882(1998).
fraction) (Modified UNIFAC) Hansen, H.K., P. Rasmussen, Aa. Fredenslund, M. Schiller and J.
Cmehling, “Vapor-Liquid Equilibria by UNIFAC Group Contribution.
Vi molar volume (Mobile Order Theory)
V; empirically modified Vivalue (Modified UNIFAC)
5. Revision and Extension”, Ind. Eng. Chem. Res. 30, 2352-2355
(1991).
X liquid-phasemole fraction
Hernandez, C.E. and W.E. Acree Jr., “Solubility of Fluoranthene in
X liquid-phase group fraction
Organic Nonelectrolyte Solvents. Comparison of Observed versus
Z coordination number (= 10)
Predicted Values based upon Mobile Order Theory”, Can. J. Chem.
76, 1312-1316 (1998).
Greek Symbols Huyskens, P.L. and G.C. Siegel, ”Fundamental Questions about
r group residual activity coefficient Entropy. 3. A Kind of Mobile Order in Liquids - Preferential Contacts
Y liquid-phaseactivity coefficient Between Molecular Croups”, Bull. SOC. Chim. Belg. 97, 821-824
6 modified solubility parameter (1 988).
e area fraction Huyskens, P.L., “Mobile and Static Molecular Disorder in Liquids”, 1.
V number of groups Mol. Struct. 8,223-246 (1992).
4J segment fraction Powell, J.R., D. Voisinet, A. Salazar and W.E. Acree Jr., “Solubility of
@Asot volume fraction saturation solubility Pyrene in Organic Nonelectrolyte Solvents - Comparison of
w see Equation (21) Observed versus Predicted Values based upon Mobile Order Theory”,
Phys. Chem. Liq. 28, 269-276 (1994).
Prausnitz, J.M., ”Molecular Thermodynamics of Fluid-Phase Equilibria”,
Subscripts Prentice-Hall, Englewood Cliffs, NJ(1969).
1 solvent
Roy, L.E., C.E. Hernandez and W.E. Acree Jr., “Thermodynamics of
2 solute
Mobile Order Theory. Part 3. Comparison of Experimental and
A solute Predicted Solubilities for Fluoranthene and Pyrene”, Polycyclic
f fusion Aromatic Compounds 13, 205-21 9 (1 999).
i, j component Siegel, C.C., P.L. Huyskens and C. Vanderheyden, “Competition
k molecular group between Solute-Solvent and Solvent-Solvent Hydrogen-bonds.
m melting in Tm; otherwise refers to molecular group Pyridines in Alcohols and in Water”, Ver. Bunsen-Ces. Phys. 94,
n molecular group 549-553 (1 990).
Weidlich, U. and J. Cmehling, “A Modified UNIFAC Model. 1. Prediction
Superscripts of VLE, hE,and Y’‘, Ind. Eng. Chem. Res. 26, 1372-1 381 (1 987).
C combinatorial
calc calculated value
exp experimental value
(11 molecule i
L liquid phase
R residual
s solid phase
sat saturated

References
Acree Jr., W.E., internal communication (1999).
Cepeda, E.A. and M. Diaz, “Solubility of Anthracene and Anthraquinone
in Acetonitrile, Methyl Ethyl Ketone, lsopropyl Alcohol and Their Manuscript received September 27, 1999;revised manuscript received
Mixtures”, Fluid Phase Equil. 121, 267-272 (1 996). May 23, 2000;accepted for publication July 13, 2000.

1174 The Canadian Journal of Chemical Engineering, Volume 78, December 2000