Envlron. Sci. Technol. 1986, 2 0 , 527-528
CORRESPONDENCE aromatic chlorine interaction parameter, so it is tentative. A report on solubility prediction using UNIFAC will be forthcoming along with optimized interaction parameters for aromatic chlorine groups in water.
Comment on “Henry’s Law Constants for the Polychlorinated Biphenyls” SIR: I recommend your UNIFAC calculations be reevaluated because of two statements ( I ) : (1) “UNIFAC predicted the y (activity coefficient) for biphenyl to be greater than that of decachlorobiphenyl” and (2) “UNIFAC yielded very inaccurate y’s for all the compounds.” I calculate log y for biphenyl to be 6.025 and 15.117 for decachlorobiphenyl, 9 orders of magnitude larger, not smaller. While UNIFAC-calculated y’s may not be accurate, they are not as inaccurate as the second quote indicates. Actual PCB solubilities (2)differ from solubilities calculated from UNIFAC activity coefficients (Figure 1; solid line indicates a perfect prediction), but I would not call that very inaccurate for all compounds, especially when one tries to a priori predict solubilities over 7 orders of magnitude. The data indicate a systematic error. If UNIFAC’s aromatic chlorine/water interaction parameter changes from 920.4 to 92.04 K, calculated solubilities are nearly identical with actual solubilities (Figure 2)UNIFAC performs excellently! This interaction parameter has not been optimized and did not consider the water/
SOLUBILITY mo//m’
-8
:/
+
1
- 12
0 UNIFAC
l o g C,
Figure 1. UNIFAC-predicted solubility.
./
SOLUBILITY mol/m
-87
I
-8
,
I
I
-6 UNIFAC
, -4
,
, -2
,
, 0
l o g C,
Figure 2. UNIFAC-predicted solubility using new aromatic chlorine/ water interaction parameter.
0013-936X/86/0920-0527$01.50/0
Literature Cited (1) Burkhard, L. P.; Armstrong, D. E.; Andren, A. W. Enuiron. Sei. Technol. 1985, 19, 590. (2) Miller, M. M.; Wasik, S. P.; Huang, G.-L.; Shiu, W.-Y.; Mackay, D. Environ. Sei. Technol. 1985, 19, 522.
Wllliam Brian Arbuckle Department of Civil Engineering The Univeristy of Akron Akron, Ohio 44325
S I R Professor Arbuckle suggests that we reevaluate our UNIFAC procedure for estimating activity coefficients for PCBs (I). In addition, the author provides an alternate aromatic chlorine/water interaction parameter for improving the fit between experimental and UNIFAC-derived parameters. There are two basic data sets for UNIFAC functional group interaction parameters (am,J, Le., those of Magnussen et al. (2) derived from binary and ternary liquidliquid equilibrium data and those of Gmehling et al. (3) designed to model vapor-liquid equilibria. In our work ( I ) , we used the former parameters following the suggestion of Lyman et al. ( 4 ) ,who stated that these would be preferable for solubility calculations. We have now recalculated activity coefficients for selected PCB congeners as suggested by Arbuckle using both data sets. The results from these calculations (as well as experimentally derived values) are presented in Table I (columns 3 and 4). We agree with Arbuckle that results using liquid-liquid derived parameters are unsatisfactory. However, activity coefficients calculated using vapor pressure parameters may still differ from experimental values by 4 orders of magnitude. We consider this error to be unacceptable. We also calculated activity coefficients by adjusting the aromatic chlorine/water interaction parameter from 920.4 to 92.04 K (as suggested by Arbuckle). The results are displayed in column 5 of Table I. We do not observe the close agreement indicated by Professor Arbuckle in his Figure 2. For example, the activity coefficient estimate for decachlorobiphenyl using this parameter still differs from the experimental value by approximately 2 orders of magnitude. From our experimental data (5-111, we have obtained a preliminary value for the water/aromatic chlorine interaction parameter of 526 K. The results using this parameter (in conjunction with the adjustment suggested by Arbuckle) are depicted in column 6 of Table I. Agreement between experimental and UNIFAC-derived activity coefficients is significantly improved. First, it is suggested that these parameters be subjected to additional optimization (12). Second, it must be emphasized that these parameters were derived by using only polychlorinated biphenyls. The application of these pa-
0 1986 American Chemical Society
Environ. Sci. Technol., Vol. 20, No. 5, 1986 527