J. CHEM. SOC. FARADAY TRANS., 1994, 90(4), 649-652 Adsorption of Binary Mixtures of Heptane and Alkanols by Activated Carbon Amelia M. Gonqalves da Silva," Virgilio A. M. Soares and Jorge C. G. Calado Centro de Quimica Estrutural, Complexo I, lnstituto Superior Tecnico , 1096 Lisbon, Portugal The adsorption isotherm of heptan+hexanol on activated carbon at 298 K is presented and compared with our earlier work on heptane-ethanol and heptane-butanol mixtures under similar conditions. The data show that the preferred adsorption shifts from the alkane to the alkanol as the length of the alkanol chain, relative to that of the alkane chain, increases. The surface mole fraction, interfacial tension differences and surface activity coeffi- cients have been estimated based on an approximate theoretical model (surface phase model).These results show that the minimum thickness of the layers required to achieve thermodynamic consistency between the surface phase model and the experimental data increases as the length of the alkanol chain decreases. All the systems exhibit smaller deviations from ideality in the adsorbed phases than in the bulk. This work is part of a larger study of the adsorption of binary mixtures of alkanes and alkanols on activated carbon with the purpose of examining the applicability of thermody- namic methods of analysis, and of testing the adsorption models and their interpretation in terms of molecular behav- iour. Previously, we reported the adsorption data for the system (heptane-ethanol)/activated carbon and their interpretation by simple models.' This study was complemented with infor- mation obtained from the adsorption on the same adsorbent of the pure components from the gas phase.In another paper it was shown that the three binary mix- tures involving butanol, heptane and dodecane in contact with activated carbon behave, within the experimental preci- sion, in a mutually consistent fashion, from a thermodynamic viewpoint.2 The surface phase model was tested on all the systems and it was found that only the nearly ideal system (dodecane-heptane) shows a good agreement between theory and experiment; the remaining systems, all involving alka- nols, display strongly non-ideal behaviour and are not satis- factorily described by that simple model.This paper reports measurements on the system heptane- hexanol on activated carbon at 298 K; a comparison is made with our earlier work on mixtures of heptane-alkanol (ethanol and butanol). Experimental Adsorption isotherms were determined by the conventional immersion method, using a differential refractometer (R 401, from Waters Associates) to analyse the equilibrium composi- tion of the bulk solutions. The procedure used has already been described.2 Materials Commercial activated carbon, 'pro analysis' grade from Merck, was used without purification. Hexanol ( 2990/,) and heptane ( 299.5%) were puris. grade from Fluka, butanol ( 299.5%) and ethanol ( 2990/,) were pro analysis grade from Merck and all were used without further purification.Their purity was further checked by adding suit- able quantities of adsorbent to each liquid (in the adsorption cells) and verifying that no significant change in the refractive index of the liquids occurred after contact with the adsorbent. Physical characterization of the activated carbon was made by adsorption of nitrogen at 77 K and pure heptane and pure ethanol at 298 K, as reported previously.' The analysis of these adsorption isotherms by the BET method and the a, plot of the nitrogen isotherm revealed the existence of narrow pores of different width, including ultramicropores in which the accessibility of ethanol and heptane is restricted. The spe- cific surface areas, us (BET) obtained from nitrogen, heptane and ethanol adsorption were 860, 646 and 697 m2 g-', respectively. Results The specific surface excess of the preferentially adsorbed com- ponent, 2, can be expressed in the following form:3 n;(")fm= noAxifm (1) where no is the total amount of components 2 and 1 in the system and Ax; = xi -xi is the variation of the bulk mole fraction when the solution, of initial concentration, xi, is equilibrated with a mass m of the solid.The specific surface excess isotherm as a function of xi for the system heptane(1)-hexanol(2) at 298 K is shown in Fig. 1. The isotherm data are given in Table 1. The experi- mental data follow a type IV isotherm in the Schay and Nagy cla~sification.~The hexanol(2) is the component preferentially adsorbed for xi values up to 0.6 and heptane(1) is preferentially adsorbed at higher values of xt .The measure- ment of the adsorption at different times of contact between solution and adsorbent for similar systems showed that the adsorption equilibrium was reached in 24 h.' Moreover, the -0.2L 1 I 1 I 0 0.2 0.4 0.6 0.8 1.0 x: Fig. 1 Specific surface excess isotherm of [heptane(l)-hexanol(2)]/ activated carbon at 298 K. Symbols represent experimental points; (-) calculated by eqn. (2) and eqn. (3); (---) calculated by eqn. (2) using K = 1.59. 650 Table 1 Values of (n;(")/rn)for heptane(1)-hexanol(2) on activated carbon at 298 K 0.0180 0.449 0.5027 0.125 0.0287 0.533 0.5641 0.045 0.0325 0.510 0.6532 -0.032 0.0565 0.584 0.7639 -0.088 0.0914 0.59 1 0.8041 -0.106 0.1463 0.539 0.8468 -0.104 0.2106 0.500 0.9026 -0.097 0.298 1 0.380 0.9336 -0.068 0.4001 0.244 0.9710 -0.030 0.4550 0.190 points in Fig.1 for the various concentrations were obtained with different contact times, ranging from 24 to 72 h; the very low scatter of the experimental points is a further confirma- tion of the true equilibrium nature of the data. Discussion Comparison with Other Alkane-Alkanol Systems One particularly interesting feature of the alkane-alkanol mixtures adsorbed by activated carbon is that the preferential adsorption shifts from the alkane to the alkanol as the length of the alkanol chain, relative to that of the alkane chain, increases.This is illustrated in Fig. 2 for three mixtures with heptane : heptane is preferentially adsorbed from mixtures with ethanol and butanol in the composition range up to a mole fraction of heptane of 0.6,while hexanol is preferentially adsorbed from mixtures with heptane up to a mole fraction of hexanol of 0.6. These results follow a similar trend to that obtained on graphon by Everett.' At the same temperature (298 K), heptane is preferentially adsorbed from mixtures with ethanol and butanol over the whole composition range, while hexanol and octanol are preferentially adsorbed from the mixtures with heptane. Everett suggested that the preferential adsorption depends on the ratio of the number of C and 0 atoms in the two molecules which can interact simulta-neously with the surface.These results suggest that the long chain of the alkanol lies roughly parallel to the surface. This orientation is mostly determined by the dispersion forces between the long hydrocarbon chains and the surface and is probably enhanced by the hydrogen bonding between the OH groups. 2.0 I I I I (a1 ---C v -0.5 -J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 The systems presented in Fig. 2 also show that the adsorp- tion (selectivity) increases with the difference in the chain lengths of the two components, a fact which is consistent with the view that the adsorption arises mainly from dispersion forces between alkyl groups and the carbon surface. The S shape of the adsorption isotherms exhibited by heptane-alkanol mixtures is clearly related to the surface polarity of the activated carbon, since the adsorption iso- therms of the same mixtures on graphon (with a non-polar surface) are U-shaped.Surface Phase Model The surface phase model6 provides a relation between the specific surface excess and the composition of the bulk liquid where K is the equilibrium constant of the phase-exchange reaction between a binary solution of components 1 and 2 in contact with a solid adsorbent, ns is the number of moles in the adsorbed phase, and yi,7; are the activity coefficients of components 1 and 2 in the bulk solution. Eqn. (2) assumes an ideal adsorbed phase of equal size molecules. K is related to the interfacial tension difference, as follows K = exp[-(at -af)u,/RT] (3) where a:, a; are the interfacial tension between the pure liquids, € and 2, and the activated carbon, a, is the molar area occupied by component 1 in the interface, a, = uJns, with usbeing the specific surface area of the solid.In order to apply eqn. (2) to the experimental results we need to evaluate (a; -a:)uJRT by integration of the Gibbs equation from x;=otox;= 1: Since there are no reliable data available for the system (heptane-hexanol)/activated carbon the bulk activity coeffi- cients were estimated using the UNIFAC method.' The curve of (n;(")/m)/(xix:7;) vs. xi 7; shown in Fig. 3 was determined using both the experimental points and inter- polated ones. Integration procedures, making use of the Newton Cotes formulae have been described before.' The amount of substance in the adsorbed phase, ns = 1.81 k0.06 mmol g-', was determined through the best 4 I 1 I I -3-I 8El, -2-8 -f .8 -"8 8 .-ii--------W 8 -1 -'8 ..I I I I J-2 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 fit of eqn. (2) and eqn. (3) to the experimental isotherm. The result of the integration of eqn. (4), (a; -at)uJRT = -1.06 0.02 mmol g-' was then used in eqn. (3). The agreement between eqn. (2) (Fig. 1, solid line) and the experimental data (symbols) is quite good. The K value can also be estimated from the azeotropic 2point, since at the azeotropic composition x; = xi, nu(")= 0. Assuming an ideal adsorbed phase of equally sized molecules, K becomes equal to yi/yi.The best fit of eqn. (2) with K = 1.59 is represented in Fig. 1 by the broken line. The agreement between the model and the experimental results is worse than before, and this can probably be attributed to the approximate estimation of the bulk activity coefficient data. Surface Mole Fraction Assuming the surface phase model, Everett6 has derived the following equation to calculate the surface composition of an adsorbed phase of z layers of molecules on a plane, smooth, completely covered homogeneous surface : where a: and a! are the areas occupied by the molecules in a monolayer. The adsorption results, for the system (heptane- hexanol)/activated carbon, are compatible with a monolayer because, for z = 1, x; d 1 and xi increases with xi over the whole composition range.This test was applied to the present system using molecular areas derived from vapour adsorp- tion : a:(heptane) = 0.568 nm2 molecule-' (experimental value determined by Clint' for close-packed molecules with the major axis parallel to the surface of the graphon) and ai(hexano1) = 0.414 nm2 molecule- (estimated value). The latter value was estimated from data for other alkanols, since no experimental data could be found in the literature for hexanol. Given the fact that the OH group in alkanols allows for different orientations at the surface, we have adopted for each alkanol molecule the average value, calculated from a set of experimental values on several adsorbents;' these average values were then plotted as a function of the number of the atoms of carbon in the chain; the a! for hexanol is then obtained through interpolation for n = 6.The a, value used was 646 m2g-obtained by heptane vapour adsorption. This test has been applied before to the systems heptane- ethanol' and heptane-butanol on activated carbon' and it was found then that the minimum thicknesses of the surface layer which obeys the above conditions (xi < 1 and x; increasing with xi over the whole range of composition) are z = 3 and z = 2, respectively. For all three heptane-alkanol systems, we can conclude that the minimum thickness of the surface layer (zmin)decreases when the chain length of the alkanol increases. Interfacial Tension Difference The value of (a -a:)aJRT, where D is the surface tension of a solution of mole fraction xi in contact with the solid, derives from eqn.(4) where the integration limits are now xi = 1 and xi . The combination of eqn. (4) with eqn. (2) pro-vides a relation between the interfacial tension and the com- position of the bulk liquid (a -a*)a = -ln{xiyi + xiy: exp[(o; -of)a,/nsRRT]) (6)n'RT Comparison of eqn. (4) with eqn. (6) in Fig. 4 shows that there is good agreement between the experimental results and 65 1 I I I (a) ' 1.0 I--U (Dh +N I b Y 1.0 1 r -lS *-2.0 0 0.2 0.4 0.6 0.8 1.0 x: 1.5 1.o c I 0 0.5 0 -0.5 -1 .o -1.5 I I 1 I J-2.0 I 0 0.2 0.4 0.6 0.8 1.0 x: Fig.4 [(a -at)aJRT] as a function of xi, (m) calculated from eqn. (4) and (V)from eqn. (6), for: (a) heptane(l)-hexanol(2);(b)heptane(2)-butanol(1); (c) heptane(l)+thanol(2) on activated carbon the theory, except for the system heptane-thanol. This is probably due to the fact that eqn. (6) is based on the assump- tion that the molecules have equal sizes, which can hardly be true in the case of the heptane-ethanol system. Surface Activity Coefficient The activity coefficients in the adsorbed phase, y;, can be evaluated through the following equation where xi is given by eqn. (5), with z = 1 for heptane-hexanol, T = 2 for heptane-butanol, and z = 3 for heptane-ethanol and (a -at)a$RT obtained from eqn.(4). The activity coeff- cients at the interface yt are plotted us. xi in Fig. 5. The curves of 7: us. xi are also shown in the same figure. The present data together with those obtained for the other two systems seem 652 1 I I 8 6 x 4 2 0 0.2 0.4 0.6 0.8 1.0 4 12 I I IB' I\ ~ 0 0.2 0.4 0.6 0.8 1.0 4 Fig. 5 Activity coefficients in bulk phase (---) and in adsorbed phase (-) as a function of bulk composition: A heptane(1)- hexanol(2); B heptane(2)-ethanol(l) on activated carbon. (a)y:, (b) y:,(4I4and (4Ys2. to indicate that the adsorbed phases of the mixtures of alkanes and alkanols on activated carbons exhibit smaller deviations from ideality than the bulk phases. Similar conclu- sions were reached by Kiselev and Khopina," and Nagy and J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Schay'' from results obtained with strongly non-ideal bulk phases. Our study confirms that adsorption from solution can often be discussed satisfactorily by assuming that the adsorbed phase behaves like an ideal monolayer in equi- librium with a non-ideal bulk pha~e.'~.'~This model is acceptable when the molecules of both components are of similar size, as assumed in eqn. (2), (heptane-hexanol, Fig. 1 and 4).The discrepancies between the theoretical and experi- mental results increase with the difference in the size of the component molecules. References 1 A. M. Gongalves da Silva, V. A. M. Soares, J. C. G. Calado and M. Brotas de Carvalho, J. Chem. SOC.,Faraday Trans., 1991,87, 3799. 2 A. M. Gongalves da Silva, V. A. M. Soares and J. C. G. Calado, J. Chem. SOC., Faraday Trans., 1991,87,755. 3 Manual ofSymbols and Terminology, IUPAC, Phys. Chem. Div., Appendix 11, part 1; D. H. Everett, Pure Appl. Chem., 1972, 31, 579. 4 G. Foti, L. G. Nagy and G. Schay, Acta Chim. Hung., 1974,80, 25. 5 D. H. Everett, Progr. Colloid Polym. Sci., 1978,65, 103. 6 D. H. Everett, Pure Appl. Chem., 1986,58,967. 7 A. Fredenslund and P. Rasmussen, Fluid Phase Equilib., 1985, 24,115. 8 J. H. Clint, Faraday Trans. I, 1972,2239. 9 A. L. McClellan and H. F. Harnsberger, J. Colloid Interface Sci., 1967,23, 577. 10 A. V. Kiselev and V. V. Khopina, Trans. Faraday SOC., 1969,65, 1936. 11 L. G. Nagy and G. Schay, Acta Chim. Acad. Sci. Hung., 1963,39, 365. 12 D. H. Everett, Trans. Faraday Soc., 1964, 60, 1803; 1965, 61, 2478. 13 S. Sircar and A. L. Myers, J. Phys. Chem., 1970,74,2828. Paper 3/06096I; Received 12th October, 1993