Environ. Sci. Technol. 1984, 18, 55-57
Adsorptive Displacement from Activated Carbon: Recovery of 4,4'-Dichlorobiphenyl Tiren Gu and Milton Manes" Chemistry Department, Kent State University, Kent, Ohio 44242
Adsorptive displacement, which has been proposed for determining multiple trace refractory adsorbates on activated carbon, was applied to 4,4/-dichlorobiphenyl(DCB) (a model compound for polychlorinated biphenyls). The adsorption of DCB was first determined in cyclohexane and in toluene (a weakly and a strongly displacing solvent) and then in each solvent with added (concentrated) naphthalene as a displacer; all loadings were in the milligram per gram range. Cyclohexane gave no detectable extraction, and toluene showed a significantly nonlinear isotherm. By contrast, linear isotherms were observed in both solvents with added naphthalene; this linearity makes it possible to analyze for trace DCB in these systems. The adsorptive displacement method was tested by shaking preloaded carbon samples (0.77-3.16 mg/g) with naphthalene-toluene (100 g/L), followed by determination of DCB in the equilibrium solutions. The average deviation between known and determined loadings was *5 % .
Introduction The immediate objective of this study was to determine the applicability of the recently developed adsorptive displacement methods (1,2) to the determination of small amounts of polychlorinated biphenyls (PCB's) on activated carbon. The model compound used for this study was 4,4'-dichlorobiphenyl, which is available as a relatively pure compound; the choice of this model eliminated the analytical problems associated with the study of complex commercial PCB mixtures. Although adsorptive displacement for trace analysis has been described earlier, we here consider the principles in somewhat further detail. Earlier studies by Greenbank ( 4 ) and Greenbank and Manes (3,5)had led to the expectation that the adsorption isotherms of multiple trace components should become linear and independent in the presence of a dominant component, provided only that the dominant component is not much more weakly displacing than any of the trace components (the criterion for displacement strength is the adsorption potential density, or adsorption energy per unit volume, a property that correlates strongly with refractive index). The GreenbankManes studies demonstrated linearity for single trace components over a wide concentration range but were largely limited to binary solutions in water. Gu and Manes ( I , 2) observed the independence of multiple trace components, first in water solution and then in organic solvents. These articles contain the suggestion that the linearity and independence of trace adsorption isotherms under their stated conditions may well be a quite widespread phenomenon. The importance of linear isotherms to the analysis of trace adsorbates is twofold: (a) in principle (although the process may be slow) one can achieve complete extraction of an adsorbate that exhibits a linear isotherm in the trace region, and (b) if the slope of the linear isotherm (under given conditions) is known, then one can, in principle, determine the amount of a trace component originally on the carbon by equilibrating it with a fixed concentration of some dominant component and determining the equilibrium concentration in the solution. 0013-936X/84/09 18-0055$01.50/0
The calculations for adsorptive displacement are quite straightforward. First, we consider the determination of the isotherm slope, (or Henry coefficient),H. Assume, for example, m grams of carbon shaken with V liters of solution containing an initial adsorbate concentration of co, which goes to c at equilibrium. Then the mass adsorbed, x (in milligrams), is equal to V(co- c), where V is in liters and co and c are in milligrams per liter. For a linear isotherm x / m = Hc (1) where H is the Henry coefficient, in liters per gram. Substituting x = V(co- c ) , we can write (V/m)(cO - c ) = Hc (2) or H = ( V / m ) ( c o / c- I )
(3)
If we denote the fraction of total adsorbate in the liquid phase (c/co) as R , then H = ( V / m ) ( l / R - 1) = ( V / m ) ( l- R ) / R (4) For R = 0.5, H = V / m . The Henry coefficient as given here is therefore the number of liters required for 50% recovery of total original adsorbate into the solution. The adsorbate can obviously be assumed to originate either on the carbon or in the solution at the beginning of the experiment, i.e., xo = Vco. Once H is determined for a system, one can readily determine x o l m as Vco/m,which is equal to ( V / m ) ( c / R ) . Solving for 1 / R from eq 4 gives 1
U
-I = - 11 + 1 R V/m
R=
-
V I/.." m
V/m
+H
(5)
whence ( x o / m )= c(H + V / m )
(6)
Let us consider an example. Suppose that a calibrating run shows the equilibrium concentration of trace component A to be 5.0 mg/L at a loading of 1.0 mg/g (i.e., that the Henry coefficient is 0.2 L/g), in the presence of 10.0 g/L of a dominant component or displacer (e.g., PNP in water). Assume now that a 0.5-g sample of a carbon is shaken with 50 cm3of 10 g/L PNP solution and that the concentration of A at equilibrium is 15 mg/L. Then the loading on the carbon is 15 mg/L X 0.2 L/g or 3 mg/g, or 1.5 mg on the sample, and the amount in solution is 15 mg/L X 0.05 L = 0.75 mg. The total original adsorbate on the carbon is 1.5 + 0.75 = 2.25 mg, and the original loading was 2.25 mg/0.5 g = 4.50 mg/g. The calculation is done a bit more rapidly by using eq 6. Here V l m = 0.1 L/g, c = 15 mg/L, and x o / m = X(0.2 + 0.10) or 4.5 mg/g. One can do the same calculation for multiple trace components, given the individual Henry coefficients and equilibrium concentrations, and one can determine all of these Henry coefficients in a single calibration run. Although one can in principle determine multiple components at any equilibrium concentrations, the method loses sensitivity if the Henry coefficients are too large (Le., at low fractional recoveries). The condition of substantial
0 1983 American Chemical Society
Environ. Scl. Technol., Vol. 18, No. 1, 1984 55
recoveries requires at least significant solubility in the solvent, since no displacer can raise the equilibrium concentration of a trace component above its saturation concentration. However, very high solubility should not be necessary to achieve good recoveries, given a good displacer. In this study we have determined the adsorption of 4,4'-dichlorobiphenyl (DCB), first in cyclohexane and in toluene without any displacer and then from these solvents with naphthalene as a displacer. The results illustrate the effect of the displacer in a strongly displacing solvent (toluene) and in a relatively weak displacing solvent (cyclohexane). Experimental Section The carbon was a Filtrasorb-400 carbon (special CAL, 200-325 mesh) that was provided by the Calgon Corp.; it was dried overnight at 110 "C before use. The shaker bath was the same as in previous studies. The 4,4'-dichlorobiphenyl (DCB) came from the supplier at a specified 98% purity, and was used as received. Equilibrium took place in 125-mL screw-capped Erlenmeyer flasks containing 1.0-g samples of carbon in 25 cm3 of solution. Solutions were shaken for at least 36 h, following the procedure that was used earlier with water solutions. In all likelihood this longer shaking time was not necessary in the organic solvents. Following equilibration the carbon was allowed to settle out and the supernatant solutions were centrifuged clear and analyzed by direct injection of 10-L samples on a Tenax column (1/4 in. X 6 ft) at 300 "C by using a flame ionization detector. Because of its higher volatility, the naphthalene came out sufficiently ahead of the DCP to give good separation; this was an important reason for the choice of naphthalene as the displacer. The analytical determination of PCB on a contaminated carbon was simulated by first preloading carbon samples with measured amounts of DCB from cyclohexane solution, from which earlier experiments had shown it to be completely adsorbed. The cyclohexane was largely (but not completely) removed from the carbon by filtration and oven drying at 60 "C in a preheated oven (disconnected from the power source for safety) after which the carbon samples were equilibrated with naphthalene-toluene (100 g/L) and the original loading determined by using the previously described calculation method. (Small amounts of residual cyclohexane on the carbon were not considered significant.) The solubility of DCB in cyclohexane and in toluene at 25 "C was determined by addition of solvent to a measured amount of solute in a shaker bath until solution was complete. The solubilitieswere 2.2% by weight in cyclohexane and 10.8% in toluene. Results and Discussion The data for DCB from cyclohexane solution, with and without naphthalene displacer, are given in Table I. The data for naphthalene (40 g/L)-cyclohexane are plotted in Figure 1. Table I1 gives the data for DCB from toluene and from 40 and 100 g/L naphthalene in toluene; the data are plotted in Figure 2. Finally, Table I11 gives the results of the simulated analysis of the preloaded carbon samples. We now consider the data in order of presentation. Table I illustrates the powerful effect of the naphthalene displacer in a relatively weakly displacing solvent such as cyclohexane. Whereas at these low loadings (of the order of mg/g) the DCB is all adsorbed by the carbon from pure cyclohexane, the fractional recovery with naphthalene displacer is approximately 20% (by eq 5, with V/m at 56
Environ. Sci. Technol., Vol. 18, No. 1, 1984
Table I. Adsorption of 4,4'-Dichlorobiphenyl(DCB) from Cyclohexane with or without Naphthalene as a Dominant Component onto Special CAL at 25 O c a equilibrium Henry concn, xlm, coefficient, mg/L mg/g H,L/g Without Naphthalene 0 3.16 0 2.33 0 1.58 0 0.79 (Naphthalene) = 4 0 g/L 3.14 0.54 0.144 7.79 1.07 0.137 12.1 1.60 0.132 17.3 2.10 0.121 20.4 2.65 0.130 a For these experiments ( V / m= 0.025 L/g), fractional recovery = 0.025/(0.025 + H ) .
-
Figure 1. Adsorption of 4,4'-dichlorobiphenyI onto Special CAL from naphthalene (40 g/L)-cyclohexane solution at 25 'C.
0.025 L/g), which would suffice for a reasonably accurate estimate of the DCB. Figure 1shows the linearity of the adsorption isotherm of DCB in the presence of naphthalene. Table I1 shows, first of all, that the value of H and therefore the recovery of DCB with toluene alone are considerably better (R = 0.7) than with cyclohexanenaphthalene, concomitant with a reduction of H by a factor of about 15. The improvement over cyclohexane-naphthalene may be at least in part ascribed to the higher solubility (about fivefold) in toluene. Whereas toluene is itself a rather powerful displacer, it is not quite powerful enough to impose linearity on DCB by itself; Figure 2 shows decided curvature in the isotherm, which is somewhat better seen from the tabulated data. By contrast, both naphthalene-toluene systems show good linearity. At 40 g/L the most striking effect of the naphthalene addition is in the linearity rather than in the recovery (0.78), and the increase in naphthalene concentration from 40 to 100 g/L lowers the value of H significantly. The effect is not large, however, and the somewhat improved recovery (0.78 to 0.85) is not critical. Note, however, that increasing recoveries are accompanied by decreased sensitivity of the accuracy of the method to the accuracy of H, quite obviously, as the fraction adsorbed approaches zero (or as H
