Horvath, H., Charlson, R. J., J. Am. Znd. Hygiene ASSOC., in press, 1969. Horvath, H., Noll, K. E., Atmospheric Environ., in press, 1969. Huschke, R. E., “Glossary of Meteorology,” American Meteorological Society, 1959. Junge, C. E., Ber. Deut. Wetterd. 35, 261-77 (1952). Koschmieder, H., Beitr. Phys. Freien Atm. 12, 33-53, 171-81 (1924). Liu, B.Y. H., Whitby, K. T., J . ColloidSci. 26, 161-5 (1968). Lundgren, D. A., Cooper, D. W., J. Air Pollution Control ASSOC. 19, 243-7 (1969). Middleton, W. E. K., “Vision through the Atmosphere,’’ pp. 60-82, University of Toronto Press, Toronto, 1952. Noll, K. E., Mueller, P. K., Imada, M., Atmospheric Environ. 2, 465-75 (1968). Peterson, C. M., Paulus, H. J., Paper 67-133,Air Pollution Control Assoc. Meeting, Cleveland, Ohio, June 1967. Pilat, M. J., Charlson, R. J., J. Rech. Atmosphkrique 2, 165-70 (1966). Pueschel, R. F., Noll, K. E., J. Appl. Meteorol. 6, 1045-62 (1 967).
Radke, L. F., Hobbs, P. V., J. Atmospheric Sci. 26, 281-8 (1 969). Robinson, E., in “Air Pollution,” A. C. Stem, ed., Vol. 1, 2nd ed., Chap. 11, pp. 349-99,Academic Press, New York, 1968. Ruppersberg, G. H., Jahrbuch der WGL, 230-36, FFM Bericht Nr. 29 (1959). State of California, “California Standards for Ambient Air Quality and Motor Vehicle Exhaust,” Dept. of Public Health, 74-5 (1960). U.S. Dept. of Health, Education and Welfare, “Air Quality Criterion for Particulate Matter,” Chap. 3, 1969.
Received for review September 16,1968.Accepted June 30,1969. Symposium on Colloid and Surface Chemistry in Air and Water Pollution, Division of Colloid and Surface Chemistry, 156th meeting, ACS, Atlantic City, N.J., September 1968. Research sponsored in part by the National Air Pollution Control Administration under Grants No. AP 00336-05 and 06.
Sorption of Phenol and Nitrophenol by Active Carbon Vernon L. Snoeyink,’ Walter J. Weber, Jr., and Harry B. Mark, Jr. Departments of Civil Engineering and Chemistry, University of Michigan, Ann Arbor, Mich. 48104
rn Equilibrium measurements of the sorption of phenol and p-nitrophenol from aqueous solution by active carbon suggest a heterogeneity of active surface sites with respect to energy of adsorption. Desorption studies show the presence of significant hysteresis effects when long equilibration periods are involved, although these effects are much smaller when adsorption-desorption equilibria are attained more rapidly. Differences in surface properties for different carbons is suggested by more extensive sorption of phenol at lower surface coverages on a coconut carbon than on a coal carbon of similar surface area. Further, again for low surface coverages and the same coconut carbon, p-nitrophenol is sorbed more extensively than phenol. At higher surface coverages the sorption is apparently less specific, and the sorption isotherms tend to converge. Studies at various pH levels indicate that the capacity of active carbon for adsorption of the anionic forms of both phenol and p-nitrophenol is less than for the corresponding neutral species. There is no marked effect of pH on the sorption of the neutral form ofp-nitrophenol in the pH range from 2.0 to 6.5.The capacity for the neutral phenol molecule decreases significantly with decreasing pH in this same range, however, suggesting that the hydrated proton competes effectively with phenol for active surface sites.
A
t least one fact has emerged quite clearly from recent efforts to characterize the processes of adsorption on active carbon in aqueous systems; namely, the structural and surface complexities of this material result in a diversity of sorptive reaction mechanisms for organic compounds of interest in water and wastewater treatment applications. Phenol and various substituted phenols comprise a group of comPresent address, Department of Civil Engineering, University of Illinois, Urbana, Ill. 61801 918 Environmental Science & Technology
pounds of particular concern for both water and wastewater treatment. The present report describes an investigation of the sorptive reactions of phenol and p-nitrophenol on active carbon. As noted, the structural characteristics and surface properties of an active carbon are essential determinants of the nature of the sorptive behavior of the carbon. Active carbon is generally considered to be comprised of rigid clusters Qf microcrystallites, each microcrystallite consisting of a stack of graphitic planes. The diameter of the planes, as well as the height of stacking, is normally less than 100 A. Each carbon atom within a particular plane is joined to three adjacent carbon atoms by u bonds, with the fourth electron of the atom participating in a x bond. It is likely that part of the carbon within a microcrystallite is highly disordered, thus deviating from the ideal graphite structure. Carbon atoms at the edges of the graphitic planes have highly reactive “free” valences (free radical sites). It is probably these free radical carbon atoms, in conjunction with van der Waals forces, which serve to bind the microcrystallite into a rigid unit. The porosity of active carbon results from the “burn-out” of intermicrocrystallite material and planes of the microcrystallite by oxidizing gases during the activation process. The extensive intraparticle surface which results is very heterogeneous in nature, consisting of basal planes and edges of the microcrystallites. A diversity of functional groups undoubtedly forms on the edges of the microcrystallites in commercial carbons because of the high reactivity of the free valences and the variety of substances used in the preparation of such carbons. Not all of these groups have been characterized, but some of the oxygen-containing functional groups which have been identified are proving very important in various applications of active carbon. Extensive discussions of the structural and surface characteristics of carbons have been given by Boehm (1966),Garten and Weiss (1957a,1957b), Snoeyink and Weber (1967),and Mattson, Mark, et a(. (1969). The presence of oxygen-containing functional groups on the surface of carbon markedly affects the adsorption of certain
compounds. Several investigators (Beebe, Biscoe, et al., 1947; Gasser and Kipling, 1960; Wright, 1967) have observed significant differences in the nature and degree of adsorption on two different carbons, Graphon and Spheron, for a variety of organic materials. It was suggested that the differences in the adsorption characteristics were due mainly to large differences in the amounts of chemisorbed oxygen on the two carbons. Recent infrared internal reflectance studies (Mattson, Mark, and Weber 1969, and Mattson, Mark, et al., 1969) of the surface of active carbon in the absence and presence of sorbed p-nitrophenol have shown that carbonyltype functional groups on the carbon surface interact with the sorbed p-nitrophenol. A comparative investigation of the sorption of phenol and various nitrophenol compounds has indicated that the formation of an acceptor-donor-type charge transfer complex between the phenols and surface carbonyl functional groups is responsible for the adsorption (Mattson, Mark, et al., 1969). The importance of the nature of surface “oxides” on carbon in aqueous systems has been shown by Coughlin and Ezra (1968) and Coughlin, Ezra, and Tan (1968). Extensive oxidation of the surface of an active carbon with ammonium persulfate to increase the quantity of acidic oxygen on the surface resulted in a reduction of the capacity of the carbon for adsorption of phenol by a factor of eight on a weight basis and by a factor of four on a surface area basis. The capacity of the carbon for nitrobenzene was reduced by a factor of four on a weight basis and by about two on a surface area basis by similar treatment. However, the original capacity was partially restored by treatment of the surfaces with a reducing agent. Bartell and Miller (1924), Frumkin (1930), Garten and Weiss (1957a, 1957b), and Snoeyink and Weber (1968) have studied the sorption of a series of strong acids on active carbon in considerable detail. The data obtained from these studies can be interpreted in terms of either physical or chemical sorption of the acids. It is generally agreed by these several investigators, however, that oxygen functional groups play a significant role in the mechanism of the sorption of strong acids. The present work is an extension of the study of the sorption of strong acids on carbon carried out by Snoeyink and Weber (1968). The sorption characteristics of phenol and p-nitrophenol are studied as a function of pH, nature of the strongacid anion or conjugate base, and the nature and concentration of inorganic salts. The purpose of this study is to further delineate the effect of solution constituents and the nature of the carbon surface on the sorption process. Experimental The active carbon used for the majority of these studies was the same 273-micron size coconut-shell carbon (Columbia Activated Carbon, LC Grade, National Carbon Co.) used by Snoeyink and Weber (1968) for the strong-acid sorption studies. This carbon was prepared for the studies by mechanical grinding and sieving to obtain a relatively narrow range of particles passing a No. 50 U.S. Standard Sieve and being retained on a No. 60 sieve. After sieving, the carbon was washed with distilled water to remove dust and fines, and dried to constant weight at 105’ C. The inorganic content of the carbon was determined to be 0 , 7 z by weight, which is rather low for a commercial carbon. Other beneficial properties of this carbon include the fact that it is very resistant to attrition in rapidly stirred reactors. A coal-base carbon (Pittsburgh Activated Carbon, Type SGL, Pittsburgh Activated Carbon Co.), also prepared in the manner described above, was used for a small portion of the present study. The ash content of this carbon was 8 Unless
z.
otherwise indicated, the carbon used for a particular study was the coconut-shell carbon. The solutes, phenol and p-nitrophenol (PNP), were selected because they can be studied both as neutral and negative species, and can be analyzed easily in aqueous solution (Weber, 1966). Phenol, in addition, is a pollutant encountered in many industrial waste discharges. The phenol was obtained in crystalline form (Mallinckrodt Analytical Reagent grade). Solution concentrations were measured by ultraviolet spectrophotometry at the wave length of maximum absorption, 270 mp. The molar absorptivity at this wave length was experimentally determined as 1540 liters per mole-cm. Since the pK, for phenol is 9.89 (Handbook of Chemistry and Physics, 1967), all concentrations were measured at a pH below 6, thereby assuring that essentially only the neutral species was present. Glass-distilled water was used for all experiments and for preparation of all stock solutions. The solubility limit of the neutral phenol species was determined to be about 0.8 mole per liter in the pH range 2 to 4. The anion solubility was not determined, but it is very soluble. The PNP was obtained in powdered form (Matheson, Coleman and Bell White Label grade). Spectrophotometric analyses for concentration were performed at a wave length of 226 mp, the molar absorptivity at this wave length being 6580 liters per mole-cm. This molar absorptivity applies only to the neutral species; thus, because the pK, of PNP is 7.15 (Handbook of Chemistry and Physics, 1967), all analyses were performed at a pH less than 4 to ensure that essentially all species were present in the neutral form. The solubility of the neutral species of PNP is approximately 0.1 mole per liter, as determined for the pH range 2 to 4. Glass-distilled water was again used for all stock and working solutions. Equilibrium studies were performed by first preparing three liters of solution at a desired pH, salt concentration, solute concentration, and temperature. Then, 100-ml. portions of this solution were pipetted into 125-ml. wide-mouth bottles, each containing a carefully measured quantity of carbon. The usual procedure was to fill 20 bottles in this manner, along with four bottles containing no carbon to serve as blanks. The weights of carbon added to the bottles were carefully selected to yield as wide a range of residual equilibrium solute concentrations, C,,,as possible for a given initial concentration. The bottles were then shaken continuously for a period of three to four weeks, to allow for equilibration of the sorption reaction. Initial tests carried out in rapidly stirred reactors had indicated that equilibrium was reached after 25 to 30 days. Approximately 95 of the amount of solute adsorbed at equilibrium was taken up within the first 10 days. These initial systems were equilibrated at a solution concentration of approximately 1 mmol per liter. The times required to reach equilibrium are similar to those measured by Weber and Morris (1964) for porous granular carbon. After equilibration, the concentration of residual solute in each experimental solution was analyzed, and the amount of solute sorbed, S, in moles per gram of carbon, was then plotted against C,,to yield the adsorption isotherm. The temperature at which these studies were carried out was 25 O C . ,unless noted otherwise. The final pH of each system was also measured and compared with the initial pH. For systems initially at pH 4, a slight increase was noted, in keeping with sorption of strong acids by the carbon (Snoeyink and Weber, 1968). However, a systematic change in pH was also noted for PNP systems having initial pH values of 6 and 8. The pH values of those Volume 3, Number 10, October 1969 919
initially at pH 6 increased to about 6.4 to 7.0, while for those systems initially at pH 8, the increase was to about 9.0 to 10.0. All of the data for this study have been plotted according to final pH rather than the initial pH (Figure 7 and related discussion). Sorption Equilibria
Figure 1 shows a log-log plot of equilibrium data for the sorption of phenol from distilled water solutions. The data for the coconut-shell carbon were tested for conformance to the Langmuir adsorption equation, but with little success. Small sections of data could be described by the Langmuir equation by recalculating the Langmuir constants for each sector, but no set of constants proved adequate for the entire range of the data. Weber and Morris (1964) and Coughlin and Ezra (1968) found similar data for the “high” and “low” concentrations for phenol adsorption on active carbon and carbon black, respectively, but intermediate concentrations were not examined. Step-type behavior has been observed by Giles, MacEwan, et al. (1960), and by Mattson, Mark, et al. (1969) for different active carbons. The data in Figure 1, however, indicate that the increase in capacity with concentration is gradual, and that a distinct step in the sorption isotherm between the “low” and “high” concentration regions apparently does not exist for the present experimental systems. The type of sorption behavior suggested by this isotherm is consistent with what might be anticipated for a heterogeneous surface, such as that of a microcrystallite active carbon. It is also possible that a rearrangement of the sorbed molecules occurs as surface coverage increases. If such a rearrangement is gradual for a heterogeneous surface, it would not be expected to produce a “stepped” isotherm. Although differential capacitance studies at the double layer have shown that surface reorientation involves a sharp “phase change” in most cases (Frumkin and Damaskin, 1964; Parry and Parsons, 1969; Hansen, Kelsh, and Grantham, 1963), there are examples of cases in which gradual rearrangement over a large change in concentration have been observed (McCoy and Mark, 1969; MacDonald, 1969). Larger portions of the data-though not all-could be described with the empirical Freundlich adsorption equation. The Freundlich equation has the form
s = FC,,N where S is the amount of solute adsorbed on the carbon at
equilibrium, C,, is the equilibrium concentration of solute in solution, and F and N are constants ( N < 1). The constants F and N can readily be determined from a log-log plot of the experimental sorption data if the data exhibit a reasonably linear trace when so plotted. The Freundlich equations corresponding to the various sectors of the experimental isotherms are noted in Figure 1. Equilibrium data for sorption of PNP from a distilled water solution are presented in Figure 2. The pH of this system was lowered to 4.0 with HCI to ensure that essentially all of the PNP was present as the neutral species. No additional salts were added, however. The Freundlich equation approximations to the data are also shown in Figure 2. Data for sorption of the phenol ar,d PNP on the coal-base carbon are also shown in Figures 1 and 2, respectively. The relatively small difference in the capacities of the coal-base and coconut-shell carbons for PNP (Figure 2) can be at least partially explained on the basis of different surface areas, but the differences noted for phenol suggest the influence of factors other than just surface area differences. The coconut-shell carbon used in these studies had a surface area of about 1200 square meters per gram (Coughlin and Ezra, 1968), while that of the coal-base carbon was about 1000 square meters per gram (Pittsburgh Activated Carbon Co., Circular No. AC7556). The capacity of the coconut carbon for sorption of solute might therefore be expected to be somewhat greater than that of the coal-base carbon. This is the case for sorption of the PNP, as illustrated in Figure 2. Conversely, as evidenced in Figure 1, the quantity of phenol sorbed on the coal-base carbon is significantly greater than that taken up by the coconut carbon in the lower concentration range. The nature of the carbon surface relative to the type of sorbate is a very important consideration in the sorption of particular solutes from solution. Nonetheless, the precise reasons for the capacity differences with respect to the two solutes are not known; need for further studies is indicated. The experimental adsorption isotherms for phenol do appear to converge for the two different carbons at higher concentrations, suggesting that the nature of the surface becomes of less importance at higher surface coverages. It is informative when comparing isotherms for sorption of different solutes to do so by plotting the number of nioles of sorbate removed per gram of carbon, S, against the reduced equilibrium concentration, C.,JCs, after Hansen and Craig (1954), where C,is the maximum solution concentration of the solute determined by solubility considerations. This reduced
-
COCONUT-SHELL CAROON C O A L - B A ? I ChRBON
Caq (MOLES/LITERl
Figure 1. Equilibrium capacities and Freundlich equations for sorption of phenol
4
i
*L
I
-
COCONUT- SHELL C A R M U COAL-BASL CARDON
i
zt 10-41 10-6
I 2
I
3
I
I
I
4 5 6
I
I
I
8 0-5
2
I
3
1
4
1
1
5 6
1
8 10-4
I 2
I
3
I
I
I
4 5 6
I
8 10-3
C q (MOLES/ LITER)
Figure 2. Equilibrium capacities and Freundlich equations for p-nitrophenol
concentration has the effect of eliminating differences in equilibrium sorption characteristics due to solubility. Differences in equilibrium sorption curves plotted in this manner should then be due to differences in the nature of sorbatesurface bonding, sorbate-sorbate interactions, or to differences in the types of sites at which sorption occurs. The quantities of phenol and PNP sorbed on the experimental coconut-shell carbon are given as functions of C,,IC, in Figure 3. The experimental data are not reproduced; rather, the curves of best fit given in Figures 1 and 2 have been normalized for presentation in Figure 3. More PNP than phenol is sorbed on the range, (1 to 10) X coconut carbon in the lower C&, At high values of C&,, the sorption capacities are the same for each solute; the solute molecules are evidently packed into the surface in such a way that the larger molecular volume of PNP does not affect the number of molecules sorbed. The rather marked differences in the low concentration regions of the isotherms for sorption of phenol and PNP on coconut carbon noted in the reduced concentration plot given in Figure 3 were not observed for the coal-base active carbon. When plotted on a reduced concentration scale, the isotherms shown in Figures 1 and 2 for adsorption of phenol and PNP on the coal-based carbon are nearly parallel, and closely spaced over the experimental range of concentrations. Within the limits of experimental error, they can be considered as
I
I
10-
10-6
10-5
essentially identical when corrected for solubility, the same as for the higher concentration regions and surface coverages for the coconut carbon isotherms given in Figure 3. It would thus appear that somewhat different surface effects are involved for the two carbons in the lower regions of surface coverage. Sorption Reuersibility
Experiments were performed to determine the degree of reversibility of the sorption reaction for phenol on active carbon. In one experiment, 5 grams of carbon were brought to mole per liter solution of phenol equilibrium with a 5 X in distilled water. In another, 1 gram was equilibrated with a 4 X mole per liter solution. After equilibration, the carbon, which was normally kept in suspension with motordriven polyethylene stirrers, was periodically allowed to settle to the bottom of the reaction vessel and the solution concentration was measured. A volume of 2.5 liters of the solution was then decanted and replaced with distilled water. The carbon was allowed to re-equilibrate with the new solution, after which the new surface concentrations were determined and plotted as a function of the residual concentration of solute in solution. Initially, periods of 10 to 12 days were allowed between decantations. During this time, frequent analyses were made to determine the time required for the carbon to re-equilibrate with the new solution. Very little
I
I
I 10'~
I
I
1
I
I
10-3
10-2
10-1
Ceq'C,
Figure 3. Sorption capacities as a function of reduced concentration Volume 3, Number 10, October 1969 921
change occurred after the third or fourth day. Thereafter, solution concentration was measured and the solution decanted every three to five days. The results of these tests are , given in Figure 4; the equilibrium sorption curve of best fit is reproduced from Figure 1 for comparison. The data in Figure 4 indicate a definite hysteresis effect in the desorption of the phenol. Indeed, only about 50% of the phenol was desorbed in each case, although the solution conM in one case and nearly centration was reduced to about 10-0 M in the other. Within the detectable limits of the analytical methods employed for determination of phenol, almost no perceptible amount of solute was released by the carbon during the three final decantation procedures, which took place over a two-week period. The possibility does exist that either the sorption curve or the desorption curve or both do not represent equilibrium conditions. However, the times allowed for equilibration, along with the numerous analyses for changes in solution concentrations, tend to discount these possibilities. A similar type of behavior has been observed for adsorption of gases by porous solids (Adamson, 1967). This phenomenon is generally attributed to the filling of capillary pores, some of which may be “ink bottle” pores having small access channels leading into larger cavities. The existence of an analogous situation for dilute aqueous systems does not seem logical. Some attempts have been made to explain adsorption from solution in terms of capillary condensation (Hansen and Hansen, 1954) but, as noted by Adamson (1967), this type of reasoning can be applied only to systems in which the solute is near its saturation concentration, and not to dilute solutions. The solutions used for the present studies were very dilute, and it would not appear that a capillary condensation argument would apply. It is possible that an irreversible chemical reaction that results in the breakdown of phenol occurs after the phenol is sorbed from solution by the active carbon. This processprobably an oxidation reaction-might proceed very slowly in homogeneous solution phase, but be markedly catalyzed at the surface of the active carbon. If the oxidation product(s) are strongly adsorbed--as might be expected, in view of the fact that active carbons are commonly employed to purify organic, solutions of oxidation products-a pronounced apparent hysteresis would result. Some qualitative evidence to support this hypothesis has been obtained from studies of adsorption-desorption phenomena using different experimental techniques. When a rapid technique for adsorbing and desorbing p-nitrophenol on Nuchar C-1000 (West Virginia
Pulp and Paper Co., Covington, W. Va.) was used, Mattson, Mark, et af. (1969) observed only a slight hysteresis effect. In the course of the present studies these experiments were repeated, using the slow equilibration technique described above, and a pronounced hysteresis of the same order as that shown in Figure 4 was observed. There is also some evidence from these continuing studies that dinitrophenol undergoes rapid oxidation in the presence of active carbon. However, these few observations are far from conclusive proof of an irreversible chemical breakdown mechanism to explain the observed hysteresis effect, and a study to examine this phenomenon in more detail is now in progress. Efects of Temperature
Equilibrium studies of phenol sorption on carbon were performed at temperatures of 11” and 37’ C , in addition to the studies at 25” C . discussed previously in connection with Figure 1. The results are given in Figure 5 . Similar results were obtained for PNP at the same temperatures. The influence of temperature on these particular systems is not very great. The observed temperature-dependence is undoubtedly the net effect of temperature on the sorbatesorbent bond, the solvent-sorbent bonds, the sorbate-sorbate interactions, and the solvent-sorbate interactions. The solventsorbent bond is important because sorption of a solute molecule probably involves concomitant displacement of solvent molecules from the surface. As such, the total process is very complex, and the effect of temperature on equilibrium capacity is difficult to interpret. Because of the observed hysteresis and the consequent question regarding the reversibility of the sorption reactionat least for the phenol system-no attempts were made to calculate heats of sorption. From a practical standpoint, however, the order of magnitude of the temperature effect is of interest. Effect of p H The effects of pH on sorption from solution must be considered to result from the combined effects of pH on the nature of the sorbate molecule, the sorbent surface, and the inorganic acids, bases, and salts present (used to adjust initial pH). For example, the sorption isotherm for the neutral PNP molecule will be different than that for the anion, at least because the anion is more soluble in the solution phase, and probably because it will be sorbed by a different mechanism. Furthermore, repulsive forces between the anionic PNP species
5 1
Figure 4. Hysteresis effects in the desorption of phenol 922
Environmental Science & Technology
C#q (MOLES/LITER)
Figure 5. Isotherms for sorption of phenol at 11°C. and 37°C.
and the sorbent can also be significant if the surface has a net negative charge, as can sorbate-sorbate repulsion after sorption occurs. Conversely, the effect of pH on the adsorbent may be manifest in an alteration of the chemical nature of surface sites-for example, protolysis of an acid functional group (Bartell and Miller, 1924; Frumkin, 1930-r in changes in the surface charge by adsorption of any of the species of the system. For the particular carbon used in this study, it has previously been shown that significant quantities of strong acid will react with the surface (Snoeyink and Weber, 1968). In an attempt to separate and explain these effects, equilibrium measurements of sorption for both phenol and PNP on carbon from solutions of different pH were made. Isotherms for sorption of phenol at pH values of 2.0, 5.6, 7.5, and 10.6 are given in Figure 6; HC1 and NaOH were used for pH adjustments in these experiments, but no other reagents were added. A summary of this plot, plus additional data points at pH 4.0 and 10.0, are given in Table I. The pKa of phenol is 9.89, so that the principal adsorbing species above this pH is most likely anionic, and a reduction in the extent of sorption is not unexpected. Both the repulsive forces between the sorbate anion and the carbon surface, and between the sorbed species themselves, would tend to cause such a decrease. However, an explanation for the observed reduction in quantity sorbed with decreasing pH below pH 7.5 is less obvious. The solubility of phenol was determined
2.0
*)
0
Table I. Effect of pH on the Sorption of Phenol on Active Carbon S(Moles/Gram) X 108 (ceQ = 7 x 10-4~) PH 2.0 1.67 4.0 1.88 5.6 1.93 7.5 2.15 10.0 1.75 10.6 1.50
experimentally to be about 0.8 mole per liter in the pH range 2.0 to 4.0, and it may reasonably be assumed that the solubility does not decrease markedly between pH 4.0 and 7.5. Ap-
parently, the nature of the carbon surface is being altered in some manner by the acid used for pH adjustment. The decrease in capacity for phenol with decreasing pH can be explained in part by the competitive effect of the increased quantity of acid sorbed with decreasing pH. From the results of recent internal reflectance studies (Mattson, Mark, et ai., 1969), it would appear that sorption of phenol on active carbon surfaces occurs at carbonyl oxygen sites. It is not unreasonable to expect that protons would interact competitively with carbonyl
-
1.5-
w
IA c , ~(MOLESILITER)
I
E
-
7.5 5.6 10.6
2.0
-
104
Figure 6. Isotherms €or phenol at different pH Volume 3, Number 10, October 1969 923
oxygen to desorb phenol as pH decreases. Also, as the chloride ion has been shown to chemisorb to a slight extent on even negatively charged surfaces, such as platinum, mercury, and carbon electrodes (Grahame, 1947; Gierst, 1959), the observed decrease in phenol capacity may result from the increased chloride adsorption on addition of HCl. The lack of significant anion effects, discussed below, tends to support proton interaction with carbonyl groups as the principal cause of the decrease in phenol capacity with increased proton concentration, or decreased pH. More work is planned in this area. Similar studies were made to determine the effects of pH on sorption of PNP, the results of which are presented in Figure 7. As noted previously, pH values for the experimental systems originally at 6.0 increased to the range 6.4 to 7.0 at equilibrium and the systems originally at 8.0 increased in pH to the range from about 9.0 to 10.0. This pH increase is easily explained in terms of the structure of the double layer. The surface adsorption of PNP anions results in a net increase in the negative surface charge of the carbon. There then must be a corresponding increase in positive ions in the diffuse double layer to electrically balance the surface charge. Thus, hydrogen ions and sodium ions are sorbed from bulk solution into the diffuse layer, causing an increase in solution pH. The model of the structure of the electrical double layer upon which this interpretation and hypothesis are based has been extensively discussed in the electrochemical literature (Grahame, 1947; Gierst, 1959; Delahay, 1965; Mohilner, 1966), and is well accepted. The hypothesis differs somewhat from that put forth by Steenberg (1944) and Garten and Weiss (1957a, 1957b) in their interpretations of acid adsorption and salt effects. If one accepts the hypothesis of adsorption of the proton in the compact or primary layer (at the surface) as suggested by Steenberg (1944) and by Garten and Weiss (1957a, 1957b), then acid adsorption should not be expected to be strongly affected by the nature of the anion or conjugate base of the acid. Adsorption of the anion in this case would result simply from the required establishment of an electrostatic balance of total surface charge within the diffuse double layer (Mohilner, 1966). The chemical nature of the anion would thus be a minor factor, relative to the establishment of a charge balance in determining the extent of sorption of the anion in the diffuse layer. However, all experiments on acid sorption show stronger anion effects than would be expected from simple hydration effects on penetration of the diffuse double layer (Steenberg, 1944; Garten and Weiss, 1957a, 1957b; Snoeyink and Weber, 1968). This stronger effect can
be explained by chemisorption of the anion in the compact double layer, with the majority of the proton sorption occurring in the diffuse layer to accomplish charge balance. This interpretation is supported by the fact that the nature of the cation in several salt-effect studies has been observed to have little influence on acid sorption (Steenberg, 1944; Garten and Weiss, 1957a, 1957b; Snoeyink and Weber, 1968). The primary interest for this particular study is the relationship of capacity to equilibrium pH. Thus, in Figure 7,only the capacities for which the equilibrium pH was 6.5 + 0.2 in one case and 10.0 i 0.2 in the other case were plotted. At pH 4, the equilibrium pH differed very little from the initial pH. The solid lines shown in Figure 7 represent the curves of best fit, as determined from a log-log plot of the data. The slight differences shown between the isotherms for pH 2.0 and pH 6.5 are relatively insignificant. The equilibrium capacities did not decrease in this case with decreasing pH below that corresponding to the pK,. This is consistent with the competitive adsorption effects of the proton and/or chloride ion discussed above. Studies of the comparative sorption of various phenols on carbon have shown that nitrophenols are more strongly sorbed through an acceptor-donor complex reaction than is phenol (Mattson, Mark ef af., 1969). Thus, competitive effects of H30+ and/or C1- should be less important in the case of the PNP. The difference in capacity applies only to the lower regions of surface coverage. The capacity of carbon for PNP decreases with increasing pH above that corresponding to the pK, of 7.15, as for phenol. Differences in solubility between the neutral and anionic species would certainly contribute to this decrease in capacity with increased pH. However, the magnitude of the difference was greater than could be reasonably accounted for in terms of solubility differences. Rather, the effect of increased pH on the development of repulsive forces between the anion and the surface, or between the sorbed anions themselves, appears to be more significant. The negative surface charge of the carbon may be increased by physical sorption of OH- ions on the surface, or by ionization of very weak acidic functional groups on the surface, or both. Repulsive forces are discussed in greater detail below in relation to the effect of inorganic salts on sorption. The temperature-dependence of sorption as a function of pH was determined for both phenol and PNP. Very little effect was noted for either species at PH values above the pK,. In addition, the magnitude of the dependence for the neutral PNP species does not vary significantly from that shown in
3.2-
2.4
-
I
I
1
2
I
3
clS
I
4
I
5
6.5
2.0 10.0
I
S
(MOLEWLITER) x 104
Figure 7. Isotherms for p-nitrophenol at different pH 924 Environmental Science & Technology
I
7
I
:O-J
I
I
l
l
I
I
I
I
I
1
I
2
3
4
s
7
,0'4
2
3
4
s
7
,0'3
C q (MOLES/LITERI
Figure 8. Equilibrium capacitiesfor phenol at pH 2 and temperatures of 11 O and 37°C.
Figure 5 for phenol. The temperature-dependence of phenol sorption at pH 2 is quite different from that at pH 5.6, however, in keeping with the effect of pH on capacity of the coconut-shell carbon for the neutral species. The data for pH 2 at temperatures of 11O and 37" C. are shown in Figure 8. The systems used for this study contained 1 X 10+ mole per liter NaCI, but at the values of pH studied in this set of experiments, there was no apparent effect of the NaCl at this concentration level on the capacity of carbon for either phenol or PNP. These data, along with the data on the effects of NaCl concentration at pH 2, seem to indicate that proton sorption at the carbonyl sites produces the predominant effect on capacity. The temperature effect for phenol sorption at pH 2 also varies more extensively with surface concentration than was observed at pH 5.6. The greater effect of temperature on phenol sorption at pH 2 is possibly related to the effect of temperature on the quantity of acid sorbed. Preliminary investigations indicate that the reaction of acid with coconut-shell carbon at pH values lower than 3.7 is endothermic. Because phenol sorption is apparently exothermic, an increase in temperature would result in a decrease in the sorption of phenol. In addition, however, a concomitant increase in temperature would lead to increased acid sorption by the carbon surface, the presence of which would further decrease the amount of phenol which could be adsorbed. The acid, then, would tend to magnify the effect of tem-
z
perature, as was observed. Additional complications arise from the differences in the dependence of solubility of the phenol on temperature at different pH levels. Salt Effects Equilibrium measurements were made for sorption of both phenol and PNP from solutions at pH 2.0 with NaCl concentrations of zero, 1 X 10-2 mole per liter, and 1.0 mole per liter. No significant differences in capacity were noted for any of these systems for the different salt concentrations. Studies of PNP sorption at pH 10.0, however, indicated a significant salt effect. This effect is illustrated in Figure 9. The data for the 1 X mole per liter NaCl solution are not shown, because it is essentially the same as that for the solution containing no NaC1. A significant increase in capacity at lower surface coverage can be observed for the 1.OM NaCl system. This is expected, as the high salt concentration may increase the degree of ion pairing of the cation with the PNP anion, which would have the effect of increasing adsorption capacity. The ion pair would behave more like the acid form of the PNP. No change would be expected in the amount of sorbate taken up at high equilibrium concentration, since new sorption sites are not being created; this is verified by the fact that the isotherms converge on approximately the same surface coverage at higher concentrations. Another possible explanation is that the salt is acting to reduce the repulsive
I
0, v)
IWNoCl
O !!
10-
No CI
d v)
5
0
2
S
3
c , ~(MOLESILITER)
a 104
Figure 9. Isotherms for the pnitrophenol anion in the presence and absence of NaCl Volume 3, Number 10, October 1969 925
forces between the sorbed anions, thus allowing more molecules on the surface at lower solution concentrations. One would expect that the ultimate capacity of the carbon would also increase, but apparently this is not the case, since both isotherms appear to have nearly the same ultimate capacity. Finally, equilibrium measurements were made for sorption of both phenol and PNP from solutions of pH 2.0 in the concentration range of (1 to 10) X IO-’ mole per liter, using HC104, HaP04, &Sod, and HNOI for pH adjustment, instead of HCl. The results differed very little from those reported for the cases in which HCl was used, indicating that anion adsorption has little effect on the sorption sites of phenol. Conclusions The characteristics of the sorption of phenol and PNP from aqueous solution by active carbon suggest a heterogeneity of surface sites with respect to the energy of adsorption. This is partially indicated by the fact that the equilibrium data do not conform to the Langmuir sorption equation. Also, PNP sorbs more extensively than phenol at low solution concentrations on the coconut-shell carbon. At higher concentrations, where the driving forces for adsorption are greater, the two isotherms converged on approximately the same surface coverage. Phenol sorption on coconut-shell carbon is affected by strong acids. These acids partially reduce the capacity of the carbon for phenol, possibly as a result of proton competition with the latter for some of the sorption sites. This effect was not observed for PNP, however, indicating that PNP is more strongly bound than phenol. The nature of sorption sites can vary significantly between different carbons, too. For example, phenol sorbs more extensively on a coal-base carbon than on a coconut-shell carbon at low concentrations, although the two carbons tested have very similar total surface areas. The pH-dependence of the capacity of active carbon for sorption of phenol and PNP appears significant. The equilibrium sorption of phenol is reduced by 30 to 50% over the concentration range of (1 to 10) X lo-‘ mole per liter as pH increases from 7.5 to 10.6. Similarly, the capacity for PNP decreases by approximately 67 % for the same concentration range as the pH increases from 4.0 to 10.0. This decrease in capacity can be at least partially attributed to an increase in repulsive forces between neighboring anionic species of the sorbate and to increased repulsive forces with the negatively charged carbon surface. These repulsive forces increase with increasing pH. Further evidence for these forces is afforded by the observation that sorption of the PNP anion increases in the presence of NaCl in a concentration of 1.0 mole per liter. This effect may result from increased ion pairing of the sodium ion with the PNP anion. Another aspect of the effect of NaCl on the sorption of organic anions is that it could act to reduce the mutual repulsive forces of the sorbate species. Studies to determine more completely the sorptive behavior and characteristics of active carbon, the types of functional groups present on the surface, and the best means, if any, of altering the surface to produce a more efficient adsorber for a given purpose are continuing. The present report is addressed to this over-all objective.
926 Environmental Science & Technology
Literature Cited
Adamson, A. W., “Physical Chemistry of Surfaces,” pp. 400-408, 548, 638, Interscience, New York, 1967. Bartell, F. E., Miller, E. J., J. Phys. Chem. 28,992(1924). Beebe, R. A,, Biscoe, J., Smith, W. R., Wendell, C. B., J . Am. Chem. SOC.69,95 (1947). Boehm, H. P., in “Advances in Catalysis,” D. D. Eley, H. Pines, P. B. Weisz, Eds., Vol. 16, p. 179, Academic Press, New York. 1966. Coughlin, R: W., Ezra, F. S., ENVIRON. SCI.TECHNOL. 2 (4), 291-97 (1968).
Coughlin,k. W., Ezra, F. S., Tan, R. N., J. Colloid Interface Sci. 28, 386-396 (1968). Delahay, P., “Double Layer and Electrode Kinetics,” Interscience, New York, 1965. Frumkin, A., Kolloid-2. 51,123 (1930). Frumkin, A. N., Damaskin, B. B., “Modern Aspects of Electrochemistry,” J. O’M. Bockris, B. E. Conway, Eds., Vol. 3, p. 149, Butterworths, London, 1964. Garten, V. A., Weiss, D. E., Australian J. Chem. 10, 309 (1957a).
Garten, V. A., Weiss, D. E., Rev. Pure Appl. Chem. 7, 69 (1957b).
Gasser, C. G., Kipling, J. J., Proc. 4th Conf. Carbon, Buffalo, 1959, p. 55, Pergamon, New York, 1960. Gierst, L., “Transactions of Symposium on Electrode Processes,” p. 109, Wiley, New York, 1959. Giles, C. H., MacEwan, T. H., Nakhwa, S. N., Smith, D., J. Chem. SOC.1960, p. 3973. Grahame, D. C., Chem. Rev. 41,441 (1947). “Handbook of Chemistry and Physics,” 48th ed., Chemical Rubber Co., Cleveland, Ohio, 1967. Hansen, R. D., Hansen, R. S . , J. Colloid Sci. 9, 1 (1954). Hansen. R. S.. Kelsh. D. J.. Grantham. D. H.. J. Phvs. Chem. 67, 2316 (1963).
’
Hansen. R. S.. Craig. R. P.. Zbid.. 58. 211 (1954). MacDonald, H. C., %., Ph.D. thesis, ’University‘of Michigan, Ann Arbor, Mich., 1969. Mattson, J. S., Mark, H. B., Jr., Malbin, M. N., Weber, W. J., Jr., Crittenden, J. L., Preprints of Papers, J. Colloid Interface Sci., p. 250, Academic Press, New York, 1969. Mattson, J. S., Mark, H. B., Jr., Weber, W. J., Jr., Anal. Chem. 41, 355 (1969). McCoy, L. R., Mark, H. B., Jr., J. Phys. Chem. 73, in press, 1969.
Mohilner, D. M., “Electroanalytical Chemistry,” A. J. Bard, Ed., Vol. 1, p. 241, Marcel Dekker, New York, 1966. Parry, J. M., Parsons, R., J. Electrochem. SOC. 113, 992 (1966).
Pittsburah Activated Carbon Co.. Circular No. AC7-556. PittsbGrgh, Pa. Snoeyink, V. L., Weber, W. J., Jr., ENVIRON. Scr. TECHNOL. 1 (3), 228-34 (1967). Snoeyink, V. L., Weber, W. J., Jr., 79,112-134 (1968). Steenbern. B.. Ph.D. thesis. Stockholm Universitv. 1944. Weber, W . J., Jr., Proc. 3rd Intern. Conf. Witer Pollution Res. 1-12, Munich, Germany, 1966. Weber, W. J., Jr., Morris, J. C., J. Sanit. Eng. Diu. 90 (SA3), 79 (1964).
Wright, E. H. M., J. Colloid Interface Sci. 24, 180-84 (1967).
Received for review November 4, 1968. Accepted June 20,1969. Presented at the Division of Colloid and Surface Chemistry, 156th Meeting, ACS, Atlantic City, N. J., September 1968. The research reported herein has been supported in part by Research Grant WP-00706 from the Federal Water Pollution Control Administration, U.S. Department of the Interior, and in part by a National Science Foundation TraineeshipNumber GE 4802 to Vernon L. Snoeyink.
