J. Phys. Chem. 1983,87,3334-3335
3334
Application of the Polanyi Adsorption Potential Theory to Adsorption from Solution onto Activated Carbon. 13. Multiple Organic Liquids from Water at Comparable Loadings Tlren Gu and MlHon Manes" Chemistry Department, Kent State Univefsify, Kent, Ohio 44242 (Received: November 8, 1982; I n Final Form: January 28, 1983)
Experimental data are presented for the adsorption of ternary and quaternary organic liquids from water solution onto activated carbon, in which all of the adsorbates are at comparable loadings, i.e., where no single adsorbate predominates. The ternary mixture comprised diethyl ether (EE), ethyl acetate (EA),and 1,2-dichloroethane (DCE),and the quaternary mixture contained the same components plus 1-pentanol (PEN). The results, which supplement earlier experiments with one dominant component, are additional evidence that, for multiple organic liquids that do not exhibit a wide range of refractive indices, the simple calculation of multiple-component adsorption from the individual isotherms is not limited by total loading, relative loadings, or number of components.
Introduction The immediate objective of this work has been to investigate the adsorption of multicomponent liquid mixtures (three and four components), where the loadings of all of the adsorbate components are approximately equal, i.e., where no single component predominates over the others. This work is a supplement to the more extensive results of Greenbank and Greenbank and Manes,'s2 and the subsequent work of Gu and Manes: on the adsorption of multicomponent liquid adsorbates from water solution. Most of this earlier work dealt with the systematic variation of the concentration of one or more trace components in the presence of a dominant component. For adsorbates in which the dominant component was of comparable displacing power with the trace components (displacing power being equivalent to adsorption energy per unit volume, and correlating strongly with refractive index), the adsorption isotherms of all of the trace components were linear and independent of each other, and the slopes of these linear isotherms could be estimated by use of a multicomponent Polanyi-based model, with the assumption that the organic adsorbate behaves as a uniform ideal solution. These systems may be thought of as especially (and perhaps excessively) simple, both because the adsorption of the trace components is linear and because the trace components do not interfere significantly with the adsorption of the dominant component. By contrast, when all components are present in comparable quantities, all compete with each other for the available adsorption surface (or "adsorption space"); since this may be thought of as introducing some further complexities into the model, such systems are particularly interesting to study. We have therefore determined the adsorption of a triple mixture and a quadruple mixture, both at different total loadings but all at comparable loadings of the individual components. The triple mixture comprised diethyl ether (EE), ethyl acetate (EA), and 1,2-dichloroethane (DCE); the quadruple mixture contained the same components and 1-pentanol (PEN). These systems should fit quite well into the Polanyibased (uniform adsorbate) for multiple liquids, and no complications were expected, either from the number of components or from their mutual interference. (1) Greenbank, Mick. Ph.D. Dissertation, Chemistry Department, Kent State University, Kent, OH 44242, May 1981. (2) Greenbank, M.; Manes, M. J. Phys. Chem. 1981,85, 3050. (3) Gu, Tiren; Manes, M. J. Phys. Chem. 1982, 86, 4221.
Theoretical Section The problem is to calculate the individual equilibrium concentrations from the adsorbate loadings (which are readily available from the experimental conditions), assuming that the adsorbate behaves as a uniform ideal solution.'I2 Assume that each isotherm gives the equilibrium concentration, ci, as a function of the adsorbate volume, Vi. Now suppose the adsorbate consists of ni moles (or Vi cm3) of each adsorbate per 100 g of carbon. To calculate the equilibrium concentration, ci, in a mixture, one determines the concentrations, ci*, of each component that would be in equilibrium at a loading of V , of that component (as a single adsorbate), where
v,= vi
(1)
i.e., where V , is the total adsorbate volume. For example, given a loading of 1cm3/100 g each of EE, EA, and DCE, for a total of 3 cm3,the c* value for EE is the concentration that is read from its single-component isotherm at an adsorbate volume of 3 cm3/100 g, and the c* values for EA and DCE are similarly determined from the same volume of 3 cm3 and the respective isotherms. Then for each component ci = x.c.* 1 1 (2) where xi is the adsorbate mole fraction of each component. The test of the model is the comparison of the calculated and experimental values of the ci at given loadings of the individual components. Incidentally, this calculation holds for any values of the mole fractions and illustrates the simplicity of the Polanyi-based model for multicomponent liquid adsorbates. (Although the Polanyi-based model is not the only model that can be applied here, it is the simplest, and we have no reason to believe that any alternative model is better.) Experimental Section The carbon (CAL 2131, 200-320 mesh) was the same carbon that was used extensively in preceding studies.2 Although this batch is approaching depletion, it was used because adsorption isotherm data for single components and from some binary mixtures were available from earlier work. The adsorbates were the same as in earlier work. The equilibrium experiments were conventional shaker-bath experiments of 25 "C, using 125-mL screw-capped Erlenmeyer flasks, with 1.00-g carbon samples, 25.0 mL of solution, and 48-h shaking time; the shaking time was
0022-3654/83/2087-3334$01.50/00 1983 American Chemlcal Society
The Journal of Physical Chemistty, Vol. 87, No. 17, 1983 3335
Adsorption from Solution onto Activated Carbon
TABLE I: Multicomponent Adsorption from Water onto Activated Carbona ~~
loading, cm3/ 100 g
~
mole fraction
g/L exptl
predicted
A. Triple Mixtures 0.76 0.276 0.507,0.510 1.07 0.297 0.597,0.682 0.21 0.427 0.239,0.226 0.020 0.291 0.0284 0.031 0.317 0.0306 0.018 0.392 0.0218 0.0034 0.296 0.0048, 0.0024 0.0039 0.315 0.0044,0.0045 0,0052 0.389 0.0046,0.0037 B. Quadruple Mixtures 0.0041 1. EE 1.14 0.235 0.0032,0.0047 .0.0070 1.15 0.254 0.0071,0.0086 EA 0.0052 1.14 0.313 0.0038, 0.0046 DCE 1.00 0.198 0.00025,0.00018 0.00022 PEN 0.042 2. EE 2.64 0.190 0.041,0.032 0.071 2.63 0.204 0.091,0.068 EA 0.037 2.72 0.262 0.026,0.023 DCE 0.021 4.94 0.344 0,019,0.017 PEN a EE = diethyl ether; EA = ethyl acetate; DCE = 1,2Equilibrium concendichloroethane; PEN = l-pentanol. tration.
1. EE EA DCE 2. EE EA DCE 3.EE EA DCE
8.26 8.35 9.59 2.42 2.77 2.47 0.99 1.01 1.00
thought to be necessary at the low loadings and was previously found to be suffi~ient.~ Following settling and clearing by centrifugation, the supernatant liquid was analyzed by gas-solid chromatography, using direct aqueous injection of 10-pL samples, and a Tenax or Porapak column (6 f t X 1/4 in.) at uniform temperatures ranging from 100 to 130 “C,depending on the system, the higher temperature being used for pentanol-containing systems. The systems were chosen for ease of analysis. The solutions were prepared at known concentrations of each adsorbate. Since most of the adsorbate in solution went onto the carbon, the correction for adsorbate in solution was relatively small, and the loadings are quite accurate. The error in determining the equilibrium concentrations by gas chromatography should not be over & 5 % ; the largest errors come, as usual, from reproducibility of individual runs.
Results and Discussion The results are given in Table I, which gives the equilibrium loadings, the mole fractions, and the experimental
and predicted equilibrium concentrations. Duplicate data indicate duplicate experiments, which were carried out for most of the runs. The average percent deviation of the theoretical predictions from the means of the experimental concentrations is *20% (as calculated from the average logarithmic deviations of the individual concentration estimates), and the average deviation of duplicate runs from their means is approximately &lo%. The largest discrepancies are for EE and EA, which appear to be well outside experimental error; they are apparently due to deficiencies in the model. The overall agreement with theory may be considered as quite good, and certainly good enough for many practical applications. Indeed, one could expect the same sort of deviations from ideality for the individual component solubilities in bulk mixtures. It is worth noting that the mutual interference between the adsorption of the individual components is here quite significant; the increase in equilibrium concentration over the concentrations estimated for the individual components alone (i.e., at their given loadings, with the assumption of no other adsorbate) ranges from about 4-fold to about 10-fold. Incidentally, this should be true only when the adsorption isotherms are nonlinear; if all the components have linear isotherms, then the Polanyi model (and possibly other models as well) predicts no evidence of adsorptive interference. In such c a e s (which we would expect to find for weakly adsorbed liquids at trace concentrations) the adsorption isotherms of the individual components become independent; one can verify this point by going through the calculations outlined in the Theoretical Section with assumed linear isotherms (with any slopes) for each component. In conclusion, our results with three and four components at comparable loadings, together with the preceding work with dominant components in binary, ternary, and quaternary adsorbate systems, support the idea that, for partially water-miscible liquid adsorbates that do not differ radically in refractive index,2the Polanyi-based “uniform adsorbate” model is not limited by total loading, relative loadings, or number of components. It is reasonable to expect that the number of components one can handle experimentally is limited only by analytical capability.
Acknowledgment. This article is based upon work supported by the National Science Foundation under Grant No. CME 7909247, and by Calgon Corporation. Registry No. C, 7440-44-0; EE, 60-29-7; EA, 141-78-6; DCE, 107-06-2; PEN, 71-41-0.