1122
Ind. Eng. Chem. Res. 1992, 31, 1122-1130
Moore, R. B.; Martin, C. R. Procedure for Preparing Solution-Cast Perfluorosulfonate Ionomer Films and Membranes. Anal. Chern. 1986,58,2570. Moore, R. B.;Martin, C. R. Chemical and Morphological Properties of Solution-Cast Peffluorosulfonate Ioners. Macromolecules 1988, 21, 1334. Noble, R. D. Analysis of Facilitated Transport with Fixed Site Carrier Membranes. J . Membr. Sci. 1990,50, 207. Noble, R. D. Facilitated Transport Mechanism in Fixed Site Carrier Membranes. J . Membr. Sci. 1991a,60,297. Noble, R. D. Facilitated Transport with Fixed Site Carrier Membranes. J. Chem. Soc., Faraday Trans. 1991b,13, 2089. Noble, R. D.; Way, J. D., Eds. Liquid Membranes: Theory and Applications; ACS Symposium Series No. 347;American Chemical Society: Washington, DC, 1987. Noble, R. D.; Way, J. D.; Powers, L. A. Effect of External Mass Transfer Resistance on Facilitated Transport. Ind. Eng. Chem. Fundam. 1986,25, 450.
Noble, R. D.; Koval, C. A.; Pellegrino, J. J. Overview of Facilitated Transport Membrane Systems. Chem. Eng. R o g . 1989,85,58. Pellegrino, J. J.; Nassimbene, R.; Noble, R. D. Facilitated Transport of C02 through Highly Swollen Ion-Exchange Membranes: The Effect of Hot Glycerine Pretreatment. Gas Sep. Purif. 1988,2, 126. Schultz, J. S. In Synthetic Membranes: Science, Engineering and Applications; Bungay, P. M., Ed.; Reidel: Dordrecht, The Netherlands, 1986;p 523. Teramoto, M.; Mabuyama, H.; Yamashiro, T.; Katayama, Y. Separation of Ethylene from Ethane by Supported Liquid Membranes Containing Silver Nitrate as a Carrier. J. Chem. Eng. Jpn. 1986, 19, 419. Way, J. D.; Noble, R. D.; Flynn, T. M.; Sloan, E. D. Liquid Membrane Transport: A Survey. J. Membr. Sci. 1982,12,239. Received for review August 22, 1991 Accepted December 2, 1991
Coadsorption of Hydrocarbons and Water on BPL Activated Carbon Edgar N.Rudisill, John J. Hacskaylo,+and M. Douglas Levan* Department of Chemical Engineering, University of Virginia, Charlottesville, Virginia 22903-2442
A novel experimental apparatus is used to measure hexane, acetone, water, hexane/water, and acetone/water isotherms on BPL activated carbon. Isotherms for pure components are measured from 25 to 125 "Cwith hydrocarbon pressures varied by a factor of lo7,from low values to saturation. Emphasis is placed on measuring the hysteresis loop in the water and hydrocarbon/water systems. For adsorption of pure water, the location and size of the hysteresis loop is found to be temperature dependent, with the loop shifting toward higher relative pressures and narrowing as temperature is increased. For the hydrocarbon/water mixtures, the hydrocarbon loading is held constant and water isotherms are measured at 25 and 100 OC. The partial pressure of the hydrocarbon is found to increase dramatically at constant hydrocarbon loading as the water loading is increased. Also, hysteresis is observed in the hydrocarbon partial pressure depending on whether the system is on the adsorption or desorption branch of the water isotherms. Behavior is complex and is interpreted qualitatively. Introduction There is a need for a better understanding of the adsorptive behavior of water vapor and mixed hydrocarbonlwater vapors on activated carbons. Water is present in humid, hydrocarbon-laden air fed to many separation and purification processes and is present in the form of steam for the steam regeneration of activated carbon beds. At present, there are very few equilibrium data available for such systems. An understanding of how water affects the adsorption of other components is largely lacking. Hysteresis in the adsorption isotherms of pure water can be expected. The temperature dependence of the hysteresis, the effect of hydrocarbon adsorption on the hysteresis, and the effect of water adsorption and its hysteresis on hydrocarbon adsorption are not known. Unlike most hydrocarbon adsorbates, water does not interact strongly with carbonaceous solids. This results in relatively small amounts adsorbed at low relative pressures, but the isotherm typically rises sharply in some region prior to saturation. Water isotherms on carbons are usually of the type 5 shape for porous solids and type 3 shape for nonporous solids in the Brunauer classification system (Gregg and Sing, 1982). The thermodynamic properties of water are dominated by hydrogen-bonding forces which give behavior during adsorption on carbons
* Author t o whom correspondence should be addressed. 'Present address: Dow Chemical Company, Houston, TX.
entirely different from that of nonpolar hydrocarbons, for which the strong interaction with carbons is due to the dispersive forces between the solid and the hydrocarbon chain or chain segments. Due to the polarity of water molecules, the amount adsorbed on a surface becomes sensitive to the number of high-energy sites available on the solid. Studies have shown that the amount of water adsorbed on nonpolar carbons can be correlated directly to the concentrationof chemisorbed oxygen atoms (Walker and Janov, 1968). The exact mechanism 'for adsorption of water vapor on microporous solids, such as the BPL activated carbon used in this work, is unclear. It is believed that adsorption occurs through a cooperative mechanism involving the hydrogen bonding of the water molecules (Gregg and Sing, 1982). Theories for the sharp rise in the water isotherm prior to saturation on microporous carbons include coalescence of clusters of molecules nucleated around highenergy sites such as chemisorbed oxygen atoms (Voll and Boehm, 1971; Tsunoda, 1990; Barton et al., 1991) and formation of a water monolayer, as would occur on a nonporous solid, with a small increase in dispersive forces (Dubinin, 1980). A characteristic that is apparent in almost all water isotherms on microporous carbons is the hysteresis loop, with the desorption branch often attributed to the evaporation of a condensed phase over a narrow pressure range. Dubinin (1980) and co-workers have established that the size and extent of the hysteresis region can be related to
Q888-5885/92/2631-1122$03.QQ/Q0 1992 American Chemical Society
Ind. Eng. Chem. Res., Vol. 31, No. 4,1992 1123 the porosity of the carbon. They found that the shape of the water isotherm is similar for nonporous and porous solids. Only the presence of the hysteresis region distinguishes the porous solid isotherm. In carbons where the activation resulted in less than 8% burnout, no hysteresis region was observed in the water isotherm. As the amount of burnout was increased further, the hysteresis region began and then spread to include most of the isotherm. Other theories for the hysteresis phenomenon suggest the existence of a metastable state (Evans and Marconi, 1986; Peterson et al., 1986; Peterson and Gubbins, 1987; Heffelfinger et al., 1988). Several investigators (e.g., Kadlec and Dubinin, 1969; Burgess and Everett, 1970)have discussed the uncertainties that exist in trying to determine the location of the lower point where hysteresis begins (the closure point). Dubinin (1980) and co-workers also investigated the energy of adsorption of water molecules in porous solids. For hydrocarbon adsorbates, the dispersive interactions with the solid are enhanced in small pores due to the superposition of the potentials of opposite pore walls; this phenomenon gives the sharp rise in the amount adsorbed at low pressures seen in type 1 isotherms. It was found that significant enhancement did not occur with water molecules. The dispersive interactions between the solid and the water are so weak that even in microporous solids no significant increases in adsorptive energy were found. Dubinin concluded by stating that the "nature of interactions in adsorption of water is the same as in the condensation of its vapors in the bulk phase." This is an important conclusion as it significantly differentiates the adsorptive behavior of water from many other common adsorbates on carbonaceous adsorbents. To complicate the analogy between bulk liquid and adsorbed water, Gregg and Sing (1982) point out that the volume of water adsorbed a t saturation, calculated using the Gurvitsch rule, is usually less than that of other adsorbates; this indicates that adsorbed water may be present in a form less dense than bulk water. The measurements of adsorption of pure water vapor on activated carbons have been limited to room temperature. Experimental investigations include consideration of the mechanism of water adsorption (Barton et al., 1984), studies of the effect of carbon oxidation on water adsorption and water hysteresis (Dubinin, 1980; Barton et al., 1984,1991; Matsumura et al., 1985), estimation of pore sizes based on the water isotherm (Tsunoda, 1990),studies of heats of adsorption (Dubinin, 1980), and ways of correlating water isotherms (D'Arcy and Watt, 1970; Evans, 1987; Dubinin and Serpinsky, 1981; Mahle and Friday, 1989; Barton et al., 1991). A theoretical study of the effect of temperature on hysteresis for a condensible adsorbate has been conducted by Ball and Evans (1989). Two models were used in their calculations. The first attributes hysteresis to a metastable fluid state. The second treats the adsorbent as an interconnected system of variably dimensioned pores with hysteresis caused by pore blocking due to different mechanisms of adsorption and desorption and different saturation pressures arising from different pore diameters. Independent of the model, Ball and Evans conclude that the hysteresis loop will become smaller with increasing temperature and will disappear at a capillary "critical" temperature that is significantly lower than the critical temperature of the bulk fluid. The temperature dependence of a hysteresis loop appears to have been measured only for the cryogenic systems xenon on Vycor glass (Burgess, 1971; Nuttal, 1974) and
carbon dioxide on silica gel (Dubinin et al., 1958). In these studies, the hysteresis loop was found to shrink in size and eventually vanish as the temperature was increased. For the adsorption of hydrocarbon/water mixtures, although one would expect different water-rich and hydrocarbon-rich phases to form in macropores for waterimmiscible adsorbates, what happens in micropores where the forces are high is unclear. In models for adsorption of hydrocarbon/water mixtures, it is sometimes assumed that an immiscible adsorbate (like hexane) would act independent of the presence of water at moderate to high loadings or a t moderate to low relative water pressures since the forces for the adsorption of hydrocarbon are much stronger (Manes, 1983). Another assumption sometimes made is that water adsorption would be a function of the pore volume occupied by the hexane (Manes, 1983; Doong and Yang, 1988). In this model, a two-phase adsorbate is assumed. The amount of water adsorbed is predicted to be zero when the volume of organic adsorbate is greater than the volume of water that would be adsorbed from a pure vapor at the existing water pressure. A variation on this theory considers water adsorption independent of the adsorption of the organic (Grant et al., 1983). Another theory treats the equilibrium of water and organics in terms of a capillary condensed phase in small pores and surface adsorption in larger pores (Okazaki et al., 1978). In this model, the effect of the organic on water adsorption is attributed to a change in the contact angle of the assumed capillary condensed phase. Measurements of adsorption of hydrocarbon/water mixtures on activated carbons have been confined to relatively small variations in partial pressures at room temperature. Emphasis has been on the effects of humidity on mixture adsorption (Okazaki et al., 1978; Freeman and Reucroft, 1979; Scamehorn, 1979; Ripperger and Germerdonk, 1983). Data have been collected for the adsorption branch of the water isotherm, but very few data are available for the desorption branch. Systems studied include acetone/water, methanol/water, toluene/water, and benzene/ water. In this paper, we are interested in examining the effects of temperature and partial pressure on the shape and location of the water hysteresis loop and on the interactions between water and organic adsorbates. We report experimental results for the temperature dependence of the hysteresis loop for water adsorbed on BPL activated carbon, a primarily microporous adsorbent. Adsorption and desorption branches are measured at temperatures from 25 to 125 "C using a novel apparatus. Data on the water hysteresis loop have not been previously reported for temperatures well above room temperature. We also report results for water coadsorbed with hexane and with acetone on BPL carbon at 25 and 100 "C. The data are gathered with the hydrocarbon loading held constant while the loading of the water is varied at a fixed temperature. We believe that this is the first presentation of such data for hydrocarbon/water systems. Hexane and acetone were chosen because of their respective water immiscibilityand water miscibility as bulk liquids. Experiments Apparatus. A diagram of the apparatus is shown in Figure 1. It is similar to the volumetricapparatus of Kaul (1987),the major differences being that ours is contained in an environmental chamber (Thermotron S8 with programmable temperature controller) to allow measurement of isotherms of condensible vapors at elevated temperatures and pressures up to saturation, and that ours oper-
1124 Ind. Eng. Chem. Res., Vol. 31,No. 4,1992
U
Injection Port
Flow Restriction
Environmental Chamber
Figure 1. Schematic diagram of apparatus.
ates with loadings chosen rather than with total pressure fixed. The experiments are based on recirculating a gas through a closed loop using a magnetic pump (Ruska 2330-805). The loop is of small, known volume and additional volume can be added using tanksfor measurement of hysteresis as described below. The loop contains a small bed of a known amount of clean activated carbon. Adsorbate is injected into the loop as liquid through an injection port using a gastight syringe; the amount injected is determined by weighing the syringe before and after injection. Gas concentrations are measured using a gas sampling valve, a gas chromatograph with flame ionization and thermal conductivity detectors, and a digital integrator. To determine when equilibrium has been reached, gas samples are taken and analyzed until concentrations reach constant values. A pressure transducer is used to record system pressures and for leak testing. Some features of our apparatus are that a wide range of measurements is possible, the recirculation allows for rapid achievement of equilibrium compared to static gravimetric methods, and the technique works equally well for pure components and gas mixtures. Quantities adsorbed are determined by material balance. It should be emphasized that, in calculating these, there is not a subtraction of two numbers of approximately equal magnitude and thus a source for significant errors. The quantity adsorbed is given by the quantity injected minus the measured amount remaining in the vapor phase. For measurement of the pure hydrocarbon isotherms, the loop volume is minimized so that a very large fraction of the injected material is adsorbed; thus, the injection essentially fixes the adsorbed-phaseloading and the vapor phase is sampled to determine the small, residual amount that remains unadsorbed. For the water-containing systems, where the measurement of hysteresis is a concern, the volume of the loop is increased such that roughly half of the water is adsorbed, as described below. Materials. The activated carbon used in these experiments is type BPL (Calgon Carbon Corp., lot no. 4814-5) in 6 X 16 mesh form. The adsorbates used were n-hexane (ACS certified 99 mol 9% pure), acetone (ACS certified 99.5 mol 9% pure), and water (distilled and deionized). Operating Procedure. The apparatus was operated in two different modes-one for pure hydrocarbons and one for water and hydrocarbon/water mixtures. For measuring isotherms for pure hexane and pure acetone the
total volume of the loop was kept small (76cm3) and the amount of carbon was large (-4 g). The advantage in this is that essentially d of the hydrocarbon will reside in the adsorbed phase. The experiment is begun by injecting a small amount of hydrocarbon into the loop. The gas phase is sampled until equilibrium is reached at 125 "C. The temperature is then lowered to 100 "C. Again, the vapor phase is sampled until equilibrium is reached, and the temperature is reduced another 25 "C. This continues until 25 "C is reached, at which point five equilibrium points have been measured. For all five pointa, the carbon loading will be about the same because of the large fraction of the hydrocarbon that is adsorbed. Changes in the small, residual amount remaining in the gas phase are still appreciable, however, giving significant variation in the gas-phase concentration. Then, more hydrocarbon is injected and the system is reheated to 125 "C. In this way, equilibria are measured from trace concentrations to near saturation over the entire range of temperatures. To measure isotherms for pure water and hydrocarbon/water mixtures, the apparatus must be able to measure hysteresis. For these measurementa the amount of carbon and the volume of the loop are chosen such that a change in temperature can change the equilibrium loading of the carbon significantly. We specified these such that approximately half of the water injected would be in the adsorbed phase at saturation of the vapor phase. Points on the adsorption branch were m e a s d by heating the apparatus and then slowly cooling to the isotherm temperature of interest. The heating desorbs material, dropping the adsorbed-phase concentration to a value below that of the adsorption branch of the desired isotherm. The slow cooling then allows material to readsorb, with the final state lying on the adsorption branch. We found that by heating the apparatus to a temperature 10 or 15 "C above the desired isotherm temperature and then cooling at a rate of 2 or 3 "C/h was sufficient to insure that the measured point would lie on the adsorption branch. Desorption points were measured in a similar way by approaching the temperature of interest from a lower temperature. We typically cooled the apparatus to 15 or 20 "C below the isotherm temperature and then warmed the system at a rate of 3 "C/h. In the hydrocarbon/water experiments, the loading of the hydrocarbon was held essentially constant while increasing the loading of the water and measuring hysteresis. This was possible since the hydrocarbon is much more strongly adsorbed and the amount of hydrocarbon in the loop remains constant. Even for relatively large changes in the hydrocarbon vapor concentration, the amount of material that desorbs is small compared to the amount on the carbon. Nevertheless, additional injections of hydrocarbon were occasionally made in order to maintain a particular loading more accurately. Isosteric mixture measurements using a conventional 'flow-through" type apparatus would be nearly impossible since the relation between the hydrocarbon partial pressure and the water loading would not be known a priori. Carbon was thoroughly regenerated before use by passing hot dry nitrogen at a flow rate of 2-4 L/min through the bed until long after the weight of the sample no longer changed. The regeneration temperature was 250 "C for the pure component hexane and acetone isotherms, which were measured on the same carbon sample, and 175 OC for water and the mixtures. Equilibration times for measurement of adsorption data ranged from a few hours to 2 days,the longest being for pure hexane at low loadings. We occasionally waited several days after equilibrium was
Ind. Eng. Chem. Res., Vol. 31, No. 4,1992 1126
5-
0
25T
+
50°C
ph
(MPa)
Figure 2. Isotherms for n-hexane adsorbed on BPL activated carbon.
'1
Q
25°C
* 50°C
1 i.
I
h
M
3
: E
v
Po (MPa) Figure 3. Isotherma for acetone adsorbed on BPL activated carbon.
reached to verify absolutely that no changes were occurring. Quantities adsorbed and partial pressures were reproducible to within about 2%.
Results and Discussion Adsorption properties of activated carbons have been reported to change after long-term exposure to humid air (Adams et al., 1988). We observed no changes in adsorption properties over the course of our experiments. If such changes were taking place, they were on such a large time scale that they were not detected. It should be noted that since our carbon was exposed to nitrogen rather than air, a change in the extent of oxidation of the carbon is not expected. Also, although some investigatorshave reported irreversible changes in the carbon during adsorption of organic vapors (Bailey et al., 1971),we observed no such changes in our experiments. Isotherms for Pure Hexane and Acetone. Our isotherms for pure hexane and pure acetone at 25,50,75,100, and 125 OC are shown in Figures 2 and 3 over approximately seven decades of pressure. The isotherms are of the type 1 classification, being strongly concave downward when plotted on rectangular coordinates. Note that the saturation pressures, as indicated by the vertical lines, are reached for all temperatures; at saturation, enough liquid has been injected into the system to exceed the effective capacity of the adsorbent. A close inspection of Figures 2 and 3 shows that at low loadings data points line up in horizontal rows across the five isotherms; in other words,
each of the isotherms has a data point with about the same adsorbed-phase concentration. Points along each of the horizontal rows were measured for the same injection. At high loadings, it is apparent that the higher temperatures give a lower loading for a given injection because of the higher vapor-phase concentration. Also, the data indicate a small amount of hysteresis for acetone at P/P,, > 0.2, corresponding to adsorption at high loadings in mesopores (Eissmann, 1991). Isotherms for Pure Water. Isotherms for water vapor at 25,50,75,100,and 125 "C are shown in Figure 4. In all figures involvingwater adsorption, filled circles are data for the adsorption branch and open circles are for the desorption branch. The isotherms at 25 "C are similar to those reported by other investigators for BPL carbon at room temperature (Freeman and Reucroft, 1979;Barton et al., 1984; Mahle and Friday, 1989; Tsunoda, 1990; Hassan et al., 1991). Comparing the adsorption branches, we see that at low temperatures the isotherms turn up sharply at relative pressures of about 0.5, while at high temperatures the sharp increase in adsorption is delayed until a relative pressure of 0.6. In addition, the slope of the isotherms becomes less steep for increasing temperature. The isotherms at higher temperatures exhibit a "tail" as P,/P', approaches unity, corresponding to adsorption in the macropore structure (Gregg and Sing, 1982). It is very probable that the isotherms at lower temperatures also have a tail at relative pressures close to unity. Comparing the hysteresis loops, we find that the lower points for the beginning of the loops are at points where the isotherm turns up sharply. The loops close at points close to saturation. This pattern varies slightly for the different temperatures. Most notably, we see that the size and location of the hysteresis loop is strongly dependent on the temperature. The hysteresis loop becomes smaller for increasing temperature and shifts to a higher relative pressure. Similar observations have been made by other investigatorsfor the temperature dependence of hysteresis for other systems under cryogenic conditions (Amberg et al., 1957;Miles, 1964;Bailey, 1965;Burgess, 1971;Nuttal, 1974) as discussed by Ball and Evans (1989). Although hysteresis in the water/BPL carbon system did not vanish over the range of temperatures examined, the trends observed in the data suggest that the capillary "critical" temperature of Ball and Evans (1989)does exist for this system. Ball and Evans also found that the single-pore and network models for hysteresis showed different temperature dependencies. The single-pore model allows for a shrinking of the width of the hysteresis loop with increasing temperature, whereas the network model shows the width not to change significantlywhile the loop shrinks by covering less of the range of the amount adsorbed. Figure 4 shows that as the temperature increases the width of the hysteresis loop shrinks while the vertical span remains about the same. This would indicate that the water/BPL carbon results show the most similarity to the hysteresis behavior predicted by the single-pore model. The theory of Ball and Evans suggests that single-pore behavior would be dominate in systems with a narrow range of pore distributions. The shrinking of the hytereais loop in the single-pore model is attributed to the approach to the capillary critical temperature in individual pores. The pore shape in activated carbons is not consistent and is thought to be primarily "slitlike" (Dubinin, 1980). Considering the complexity and heterogeneity of the carbon structure, we would expect to see the type of
1126 Ind. Eng. Chem. Res., Vol. 31, No. 4,1992
251
25
20
2 1 : 0
15
F?
101
5L
00 0
0.2
0.4
0.6
0.8
1
0.2
0.6
0.4
0.8
1
p; 30,
25
I
,
I
IC
I
251
k'
20 h
M &
h 2 153
v
$2
0
0.2
0.6
0.4
0.8
10 -
0
1
0.2
0.4
0.6
0.8
1
P:.
P;,
0
0.2
0.4
0.6
0.8
1
P;. Figure 4. Isotherms for water adsorbed on BPL activated carbon. P, = Pw/Pw.Filled circles denote adsorption, and open circles denote desorption. (a) 25 O C , P, = 3.169 kPa. (b) 50 "C, P, = 12.34 kPa. (c) 75 O C , P, = 38.55 kPa. (d) 100 "C, P, = 101.33 kPa. (e) 125 "C, P, = 232.18 kPa.
hysteresis predicted by a network model. The mechanism for water adsorption is unlike that used in the single-pore model of Ball and Evans. They assume a cylindrical pore with adsorbate-adsorbate potentials given by hard spheres with a Yukawa attractive tail and adsorbate-adsorbent potentials given as a hard core potential with a van der Waals attractive tailintegrated along the axis of the pore. Both of these potential forms attribute intermolecular interactions, in general, to dispersive forces. Such forces are not thought to be dominant in the
adsorption of water on carbonaceous adsorbents. The physical system Ball and Evans compare predictions to is xenon adsorbed on Vycor glass (Nuttal, 1974; Burgess, 1971). Since the adsorption of water is thought to be closely linked to the presence of high-energy sites on the solid surface, it is probable that the adsorptionmechanism for water is so geometrically specific that a model like the single-pore model is not sufficient to give good predictions of water behavior. Although there is currently much work being done investigatingsingle-pore hysteresis for simple
Ind. Eng. Chem. Res., Vol. 31, No. 4,1992 1127
3 0 1 *O
7
2o
7
25
o
0.2
0.4
0.6
0.8
1
p;, Figure'& Water isotherms at 25 'C for hexane loadings of nh 0 (top), 0.497,1.032,and 1.993 (bottom) mol/kg. P, = PJP,. Filled circles denote adsorption, and open circles denote desorption.
solid geometries, it would seem the complexity and inhomogeneities found in the interactions of water with industrial carbons currently limit the fundamental understanding of the mechanisms for the adsorption and the hysteresis within them. Coadsorption of Water and Hexane. Experiments were conducted for hexane loadings of 0.497 f 0.001,1.032 f 0.007,and 1.993 f 0.007 mol of hexane/kg of carbon (nominally 0.5, 1, and 2 mol/kg) at 25 "C. Water concentrations were varied from low values to saturation for each of these loadings. h u l t a are shown in Figure 5 along with the pure water isotherm for 25 "C. Testa confirmed that the same equilibria were reached independent of the order of introduction of the adsorbed components, There are two important things to notice in Figure 5. First, the hysteresis loop decreases dramatically in size with increasing hexane loading. Second, and most interesting, the apparent totalpore volume filled by the hexane and water is significantly less than the total pore volume available for either pure component. To illustrate this point, consider the following. Pure component loadings near saturation at 25 "C from Figures 2 and 4 are approximately 4 mol/kg for hexane and 22 mol/kg for water. Basii adsorbate densities on these loadings and assuming that the same pore volume is accessible to both components, for the hexane loading of 0.497 mol/kg, 87% of the pore space is available for water adsorption. However, from Figure 5, the water loading at saturation is only about 16 mol/kg, indicating that water fills only 79% of the available pore space. Similarly, for the hexane loadings of 1.032 and 1.993 mol/kg, at saturation water fills only 55% and 21% of the remaining pore space, respectively. To ensure that the low water loading in the presence of adsorbed hexane was not due to pore blockage by hexane, the following experiment was performed. Starting with clean carbon, water was injected into the system such that the pure water equilibrium was n, = 3.38 mol/kg and P, = 0.41. Then, under isothermal conditions, hexane was added to the system until nh = 2.04 mol/kg. When equilibrium for the hexane/water mixture was reached, the water loading and relative pressure were n, = 1.56 mol/kg and P, = 0.79,which agrees with the mixture experiment data for nh = 1.993 mol/kg. Thus, even under isothermal conditions,when the water is already adsorbed, the hexane drives off the water and reaches approximatelythe same equilibrium as with the hysteresis measurement. The low water loadings, therefore, reflect a lower average adsorbed-phase density for water when coadsorbed with
0
I
I
I
3
6
9
I
I
1
2
1
5
nw (mol/kg)
2o
15
7 I C
0
0.5
1
1.5
2
2.5
3
nu,(mol/kg) Figure 6. Effect of water loading on hexane partial pressure at 25 'C. p h = Ph/Ph. Filled circles denote adsorption, and open circlee denote desorption. P h = 20.0 kPa. (a) ?ah = 0.497 mol/&. (b)nh = 1.032 mol/kg. (c) nh = 1.993 mol/kg.
hexane and possibly for hexane when coadsorbed with water. Figure 6 shows the dependence of the hexane partial pressure at constant hexane loading on the amount of water adsorbed for the data shown in Figure 5. Here, we see the hexane partial pressure is a strong function of the water loading. For the hexane loadings of 0.497 and 1.032 mol/kg the dependence is almost exponential with the partial pressure increasing by over an order of magnitude. An interesting observation is that there are two different
1128 Ind. Eng. Chem. Res., Vol. 31, No. 4, 1992
25
E
b
A
15
v
2
/
t
n u
1
//
lot 5c
n
0
0.2
0.4
0.6
0.8
1
0
0.2
0.4
0.6
0.8
1
denote adsorption, and open circles denote desorption.
p,: Figure 8. Water isotherms at 25 "C for acetone loadings of n, = 0 (top), 1.92, and 3.76 (bottom) mol/kg. P, = P,/P,. Filled circles denote adsorption, and open circles denote desorption.
equilibrium hexane partial pressures for the same water loading. Thus, the hexane partial pressure depends on the water adsorption path. The hysteresis of the hexane is in the same direction as that for water; the partial pressures of both components are higher on the adsorption branch. Figure 7 shows water isotherms at 100 "C for hexane loadings of 0,0.97f 0.03,and 2.00 f 0.07 mol/kg (nominally 0, 1, and 2 mol/kg). The data indicate behavior similar to that for the hexane/water mixtures at 25 "C with the exception of the amount of available pore volume f i e d by the water at saturation. Here, we find that water fills approximately all of the available pore volume for n h = 0.97 mol/kg and 70% of the available pore space for nh = 2.00 mol/kg. For these experiments, the effect of water loading on the hexane partial pressure was very similar to that shown in Figure 6 for 25 "C. The low water loadings found when water is coadsorbed with hexane could be caused not only by the repulsion between water and hexane molecules but also by the hexane adsorbing in such a way as to shield at least partially the oxide sites from the water molecules, thereby changing the forces apparent to the water molecules. This would alter the nature of water adsorption, preventing clusters from forming and coalescence from occurring around these high-energy sites. In Figure 6 we found that the presence of adsorbed water with hexane resulted in a higher hexane partial pressure. This finding is at odds with theories in which the hydrocarbon equilibrium is completely independent of the water loading. Our result indicates that there are not two independent adsorbed phases, but rather that the two components are competing for much of the same surface. As water adsorbs, this competition requires that the hexane partial pressure increase to maintain the same hexane loading. The equilibrium of the coadsorbed hexane is strongly dependent on the water loading, even for moderate hexane loadings. The somewhat different character of the results for hexane/water mixtures at 100 "C may be due to the much higher vapor-phase concentrations of water at 100 OC (approximately 2 orders of magnitude over that at 25 "C). This may enable water to compete more effectively for the available high-energy sites. Coadeorption of Water and Acetone. The experimente with water and acetone coadsorbed on BPL carbon are motivated by the polar nature of acetone. Here, a hydrocarbon is considered that has favorable interactions
with water in bulk form (Le., it is miscible) yet is still strongly adsorbed on BPL carbon. As with the hexane/ water experiments, the amount of acetone adsorbed was held approximately constant while the loading of water was increased. Because of the miscibility of water and acetone, the vapor phase will reach saturation prior to the point at which the partial pressure of water reaches the pure component vapor pressure. Water isotherms at 25 "C were measured for acetone loadings of 1.92 f 0.09 and 3.76f 0.24 mol/kg (nominally 2 and 4 mol/kg). Figure 8 shows the pure water isotherm at 25 "C and the water isotherms for the two acetone loadings. The most striking feature of these data is that the total pore volume is much closer to being fiied by the acetone/water adsorbed mixture than it was by the hexane/water adsorbed mixture. Approximately 73% and 47% of the pore space is available for water adsorption for the acetone loadings of 1.92 and 3.76 mol/kg, respectively. From the water isotherms in Figure 8, the water loadings near saturation are such that 90% of the available pore volume is filled for the lower acetone loading, while for the higher loading the pore volume is essentially filled. In addition, there appears to be less water hysteresis than for hexane/water coadsorption. Figure 8 shows that about the same amount of water adsorbs at low pressures even with acetone filling part of the pore space; this indicates the existence of a mechanism for enhancement of water adsorption at low water relative pressures. Figure 9 shows the dependence of the equilibrium acetone partial pressure on the amount of water adsorbed. Although the dependence is seen to be much weaker than that observed for hexane, it is not negligible at higher water loadings. Also, the acetone partial pressure exhibits hysteresis with the higher acetone partial pressure occurring on the adsorption branch of the water isotherm. Water isotherms at 100 "C for acetone loadings of 0,1.89 f 0.07, and 3.94 f 0.15 mol/kg (nominally 0, 2, and 4 mol/kg) are shown in Figure 10. Here, the water fiis all of the pore volume not occupied by the acetone as the vapor approaches saturation. The effect of the water loading on the acetone partial pressure at 100 "Cresembles that shown in Figure 9 for 25 "C. The d e r extent of hysteresis for components that are miscible with water compared to components that are immiscible with water has been observed by other investigators. On the basis of limited data,Okazaki et al. (1978) observed lese hysteresis for acetone/water mixtures than for benzene/water mixtures. These researchers attributed
e l Figure 7. Water isotherms a t 100 "C for hexane loadings of nh = 0 (top), 0.97, and 2.00 (bottom) mol/kg. P, = P,/P,. Filled circles
Ind. Eng. Chem. Res., Vol. 31, No. 4, 1992 1129
l 0 I - - - - - l P X
hp
Figure 9. Effect on water loading on acetone partial pressure at 25 OC. P,= PJP,. Filled circles denote adsorption, and open circles denote desorption. = 30.696 Wa. (a) n, = 1.92 mol/kg. (b) n, = 3.76 mol/kg.
Conclusions A recirculating constant volume apparatus was used to measure both pure component and mixture isotherms over a wide range of conditions. Some of the novel features of this apparatus are that vapol-solid adsorption equilibria can be measured over a wide range of temperatures and pressures, even for adsorbates of moderate volatility, and mixture data can be collected in isostere form; i.e., the loading of one component can be held constant while the loading of the other is varied. Isotherms for water on BPL carbon have been measured over a range of practical temperatures. The hysteresis loop, which is characteristic of water isotherms on porous materials, was found to shrink and change location with increasing temperature. These observations are similar to those found by other investigators for the temperature dependence of hysteresis for other systems under cryogenic conditions. The data agree with the prediction of Ball and Evans (1989) of a capillary critical temperature but do not agree with the trends they found for the disappearance of the hysteresis loop for a heterogeneous solid. This discrepancy is attributed to a fundamental difference between the mechanism for water adsorption on carbonaceous adsorbents and the mechanism used in their model. Water isotherms at constant hydrocarbon loading were measured for hexane/water and acetone/water systems at 25 and 100 "C. For both hexane and acetone, the partial pressure of the hydrocarbon was found to increase significantly at constant hydrocarbon loading as the water loading was increased. Furthermore, the hydrocarbon partial pressure was found to exhibit hysteresis depending on whether the system was on the adsorption or desorption branch of the water isotherm. Results at 25 OC indicate that water does not fill the available pore volume left by the adsorbed hexane as the water vapor approaches saturation. At high hexane loadings it was found that only a small percentage of the available pore space was filled by the water before saturation. At 100 "C the water fills the available pore space much more effectively in hexane/ water mixtures than at 25 "C. In contrast, water was found to fill almost all of the available pore volume as the water vapor approached saturation in the acetone mixture experiments. The data presented in this study provide new insights into water and hydrocarbon/water mixture adsorptive behavior. The degree of complexity in the mixture results indicates the difficulties to be encountered in developing equilibria models for systems of hydrocarbons and water adsorbed on BPL and other activated carbons.
7
30 I 25
J
i-
Acknowledgment
0
0.2
0.4
0.6
0.8
1
p :
Figure 10. Water isotherms at 100 O C for acetone loadings of n, = 0 (top), 1.89, and 3.94 (bottom) mol/kg. P, = P,/P,. Filled circles
We gratefully acknowledge financial support from the U.S. Army CRDEC (Contract DAAA15-90-C-0012),the National Science Foundation (Grant CBT-8417673), and the donors of the Petroleum Research Fund, administered by the American Chemical Society. R&stw NO.H&(CHJ,CH, 110-51-3;H&C(O)CH3,67-64-1; HzO, 7732-18-5; C, 7440-44-0.
denote adsorption, and open circles denote desorption.
some water adsorption when in the presence of an adsorbed, miscible organic to dissolution. This and the observation that acetone/water mixtures appear to fill the pore volume are due largely to the polarity of the acetone. Since acetone has a favorable interaction with water, there is enhancement for the adsorption of water molecules and this reduces the magnitude of the hysteresis.
Literature Cited Adams, L. B.; Hall, C. R.; Holmes, R. J.; Newton, R. A. An Examination of How Exposure to Humid Air Can Result in Changes in the Adsorption Properties of Activated Carbons. Carbon 1988,26, 451. Amberg, C. H.; Everett, D. H.; Ruiter, L. H.; Smith, F. W.The Thermodynamics of Adsorption and Adsorption Hysteresis. In Proceedings of the Znternutwnul Conference on Surface Actioity
1130
I n d . E n g . C h e m . Res 1992,31, 1130-1135
(2nd); Schulman, J. H., Ed.; Butterworths: London, 1957; pp 3-16. Bailey, A. Ph.D. Dissertation, University of Bristol, 1965. Bailey, A.; Cadenhead, D. A.; Davies, D. H.; Everett, D. H.; Miles, A. J. Low Pressure Hysteresis in the Adsorption of Organic Vapours by Porous Carbons. Trans. Faraday Sac. 1971, 67, 231. Ball, P. C.; Evans, R. Temperature Dependence of Gas Adsorption on a Mesoporous Solid: Capillary Criticality and Hysteresis. Langmuir 1989,5,714. Barton, S. S.; Evans, M. J. B.; Holland, J.; Koresh, J. E. Water and Cyclohexane Vapour Adsorption on Oxidized Porous Carbon. Carbon 1984,22,265. Barton, S. S.; Evans, M. J. B.; MacDonald, J. J. The Adsorption of Water Vapor by Porous Carbon. Carbon 1991,29,1099. Burgess, C. G. Ph.D. Dissertation, University of Bristol, 1971. Burgess, C. G. V.; Everett, D. H. The Lower Closure Point in Adsorption Hysteresis of the Capillary Condensation Type. J . Colloid Interface Sci. 1970,33,611. D'Arcy, R. L.; Watt, I. C. Analysis of Sorption Isotherms on NonHomogeneous Sorbents. Trans. Faraday Sac. 1970,66, 1236. Doong, S.J.; Yang, R. T. Adsorption of Mixtures of Water Vapor and Hydrocarbons by Activated Carbon Beds: Thermodynamic Model for Adsorption Equilibrium and Adsorber Dynamics. AIChE Symp. Ser. 1988,83 (No. 259), 87. Dubinin, M. M. Water Vapor Adsorption and the Microporous Structures of Carbonaceous Adsorbents. Carbon 1980,18,355. Dubinin, M. M.; Serpinsky, V. V. Isotherm Equation for Water VaDor Adsomtion bv MicroDorous Carbonaceous Adsorbents. Carbon 1981,'19, 40i. Dubinin, M. M.; Bering, B. P.; Serpinsky, V. V.; Vasil'ev, B. N. The ProDerties of Substances in the Adsorbed State: Studies of Gas Adsorption over a Wide Temperature and Pressure Range. In Surface Phenomena in Chemistry and Biology; Danielli, J. F., Pankhurst, K. G. A,, Riddiford, A. C., Eds.; Pergamon Press: London, 1958;p 172. Eissmann, R. N. Personal communication, 1991. Evans, M. J. B. The Adsorption of Water by Oxidised Microporous Carbon. Carbon 1987,25,81. Evans, R.; Marconi, U. M. B. Fluids in Narrow Pores: Adsorption, Capillary Condensation, and Critical Pointa. J . Chem. Phys. 1986, 84, 2376. Freeman, G. B.; Reucroft, P. J. Adsorption of HCN and H 2 0 Vapor Mixtures by Activated and Impregnated Carbons. Carbon 1979, 17,313. Grant, R. J.; Joyce, R. S.; Urbanic, J. E. The Effect of Relative Humidity on the Adsorption of Water-Immiscible Organic Vapors on Activated Carbon. In Fundamentals of Adsorption; Myers, A. L., Belfort, G., Eds.; Engineering Foundation: New York, 1983; pp 219-227.
Gregg, S. J.; Sing, K. S. W. Adsorption, Surface Area and Porosity; Academic Press: New York, 1982. Haasan, N. M.; Ghosh, T. K.; Hines, A. L.; Loyalka, S. K. Adsorption of Water Vapor on BPL Activated Carbon. Carbon 1991,29,681. Heffelfinger, G. S.;van Swol, F.; Gubbins, K. E. Adsorption Hysteresis in Narrow Pores. J . Chem. Phys. 1988,89,5202. Kadlec, 0.;Dubinin, M. M. Comments on the Limits of Applicability of the Mechanism of Capillary Condensation. J . Colloid Interface Sci. 1969,31,479. Kaul, B. K. Modern Version of Volumetric Apparatus for Measuring Gas-Solid Equilibrium Data. Ind. Eng. Chem. Res. 1987,26,928. Mahle, J. J.; Friday, D. K. Water Adsorption Equilibria on Microporous Carbons Correlated Using a Modification to the Sircar Isotherm. Carbon 1989,27,835. Manes, M. Estimation of the Effects of Humidity on the Adsorption onto Activated Carbons of Vapors of Water-Immiscible Organic Liquids. In Fundamentals of Adsorption; Myers, A. L., Belfort, G., Eds.; Engineering Foundation: New York, 1983;pp 335-344. Matsumura, Y.; Yamabe, K.; Takahashi, H. The Effects of Hydrophilic Structures of Active Carbon on the Adsorption of Benzene and Methanol Vapors. Carbon 1985,23,263. Miles, A. J. Ph.D. Dissertation, University of Bristol, 1964. Nuttal, S. Ph.D. Dissertation, University of Bristol, 1974. Okazaki, M.; Tamon, H.; Toei, R. Prediction of Binary Adsorption Equilibria of Solvent and Water Vapor on Activated Carbon. J. Chem. Eng. Jpn. 1978,11,209. Peterson, B. K.; Gubbins, K. E. Phase Transitions in a Cylindrical Pore: Grand Canonical Monte Carlo, Mean Field Theory and the Kelvin Equation. Mol. Phys. 1987,62,215. Peterson, B. K.; Walton, J. P. R. B.; Gubbins, K. E. Phase Transitions in Narrow Pores: Metastable States, Critical Pointa, and Adsorption Hysteresis. In Fundamentals of Adsorption; Liapis, A. I., Ed.; Engineering Foundation: New York, 1986,pp 463-471. Ripperger, S.; Germerdonk, R. Binary Adsorption Equilibria of Organic Compounds and Water on Active Carbon. Ger. Chem. Eng. 1983,6,249. Scamehorn, J. F. Removal of Vinyl Chloride from Gaseous Streams by Adsorption on Activated Carbon. Ind. Eng. Chem. Process Des. Dev. 1979, 18, 210. Tsunoda, R. Adsorption of Water Vapor on Active Carbons: Estimation of Pore Width. J . Colloid Interface Sci. 1990,137,563. Voll, M.; Boehm, H. P. Basische Oberfllchenoxide auf Kohlenstoff-111. Aktiver Wasserstoff und Polare Adsorptionszentren. Carbon 1971,9,473. Walker, P. L.; Janov, J. Hydrophilic Oxygen Complexes on Activated Graphon. J . Colloid Interface Sci. 1968,28,449. Received for review August 1, 1991 Accepted December 23, 1991
Ca(OH),/Fly Ash Sorbents for SO2 Removal Ch'un-Sung Ho and Shin-Min Shih* Department of Chemical Engineering, National Taiwan University, Taipei, Taiwan, R.O.C.
The reactivity of Ca(OH)2/flyash sorbent with SO2has been studied by using a fiied-bed differential reactor under the conditions simulating the bag filters of the spray-drying flue gas desulfurization. The source of fly ash and the sorbent preparation conditions affect the reactivity of the sorbent. The reactivity of the sorbent was found to be closely related to the content of the calcium silicate hydrate formed in the sorbent preparation. The sorbent has a much higher utilization of Ca(OH), than that of pure Ca(OH), sorbent, and in some range of Ca(OH)2content the sorbent also has a higher SO2 capture capacity per unit weight of sorbent than that of pure lime. The fly ash from the Shin-Da plant of the Taiwan Power Company produced the best sorbent of all fly ashes in this study. The higher ratio of fly ash/Ca(OH),, the higher slurrying temperature, the longer slurrying time, and the smaller particles of fly ash enhance the utilization of Ca(OH)2,but the water/solid ratio has an optimal value. The relative humidity in the reactor has a significant effect on the reactivity of Ca(OH),/fly ash sorbents, but the effect of the sulfation temperature is subtle. Introduction Spray-drying flue gas desulfurization (spray-drying FGD)is one of the effective processes for removing SO2
* Author to whom correspondence should be addressed.
from the flue gas (Miller, 1986). Compared to other proceases, this process needs less space and is easier to retrofit, and it produces dry solid product. There are many coal-fired power plants that adopt this process to remove
sop
0888-5885/92/2631-ll30$03.00/00 1992 American Chemical Society