Ind. Eng. Chem. Res. 1993,32, 2047-2052
2047
Adsorption of Aromatic Hydrocarbons in NaX Zeolite. 1. Equilibrium Douglas M. Ruthven' Department of Chemical Engineering, University of New Brunswick, Fredericton, New Brunswick, Canada E3B 5A3
Bal K. Kaul Exxon Research and Engineering, Florham Park, New Jersey 07923
Equilibrium isotherms and isobars have been measured for a series of aromatic hydrocarbons (benzene, toluene, xylene, mesitylene, tetramethylbenzene, naphthalene, hexamethylbenzene, dimethylnaphthalene, and anthracene) on well-defined crystals of NaX zeolite. Henry constants, energies of adsorption, and complete isotherms are reported. The isotherms for the higher molecular weight species approach the ideal Langmuir form with a saturation capacity close to 1 molecule/cage. Heats of sorption increase linearly with carbon number, and there appears to be little difference in either the isotherms or the heats of sorption for different aromatic sorbates of the same carbon number. Adsorption of linear paraffins (up to C d has been studied in detail on several common zeolite adsorbents (Stach, 1977; Stach et al., 1983),but except for the lower homologs, the adsorption of aromatic molecules has received much less attention. Such information is, however, needed both in order to understand catalytic behavior and to delineatethe possibilitiesfor adsorptiveseparations. In this paper we report the results of a detailed experimental study of adsorption equilibrium for a series of aromatic hydrocarbons on NaX zeolite crystals. Kinetic data for the same systems are reported in part 2. The higher molecular weight aromatics are very strongly adsorbed and intracrystalline diffusion is quite slow, even at temperatures as high as 300 "C. As a result, it is difficult to make reliable and accurate measurements of the adsorption equilibrium isotherms for these species, and these difficulties are compounded by the very low equilibrium vapor pressure (at room temperature) which restricts the accessible pressure range. Such considerations probably explain the scarcity of prior studies of these systems.
dimensionless Henry's law constant ( K ) is derived directly from the first moment of the response peak:
Results and Discussion Measurements with different crystal size fractions yielded consistent results at higher temperatures, but with the heavier sorbates at the lower temperatures significant discrepancies were observed, with the apparent capacity being lower for the larger crystals. These discrepancies are attributed to slow diffusion, since, for the larger sorbates in the larger crystals, more than a week was sometimes required to approach equilibrium. For the larger sorbates the detailed analysis, presented below, is therefore based only on data for the smallest (2 pm) crystals. Conformity of the experimental isotherms to the ideal Langmuir model
Experimental Section The larger zeolite crystals used in this study were synthesized in this laboratory by Charnell's method (Charnell,1971). The smallestcrystals (-2pm) were from a commercial sample provided by Union Carbide. The Si/Al ratios for all samples, determined by atomic absorption spectroscopy, were in the range 1.2-1.3. Equilibrium isotherms were measured by the usual gravimetric method using a Cahn vacuum microbalance system fitted with a Barocel electronic manometer and a digital data storage/retrievalsystem which records readings of pressure and sample weight at predetermined time intervals. Measurements were made with several different crystal size fractions, primarily to establish the nature of the controlling mass-transfer resistance. Prior to the measurements the samples were outgassed overnight at 375 OC, 106 Torr. The experimental isotherms (Figures 1-6) are plotted as weight percent based on the weight of dehydrated zeolite crystals. Measurementsof Henry constants for some of the lighter species were made also by the chromatographic method (see,for example, Ruthven (1984)). Retention times were measured over a range of flow conditions using a short column packed with unaggregated zeolite crystals. The 0888-5885/93/2632-2047$o4.00/0
where bq, = K (the Henry constant), may be tested by plotting l/q vs llp: (3)
In general it was found that the isotherms for the higher molecular weight sorbates conform closely to this simple model, especially at higher temperatures, and for theae systems, the saturation limit (qs) is essentially independent of temperature. No doubt this is a reflection of the wellknown tendency of large molecules to average out the effects of adsorbent heterogeneities. Henry's law constants were derived either from the slopes of the Langmuir plots or from virial plots of log(q/p) vs q (Barrer and Davies, 1970). Adsorption energies were then found from the temperature dependence of the Henry constant, which is assumed to follow the normal van t'Hoff form:
where AUo =
AH0
+ RT.
0 1993 American Chemical Society
2048 Ind. Eng. Chem. Res., Vol. 32, No. 9, 1993
The vapor pressure of the higher molecular weight sorbates was too low to allow accurate measurement of the isotherms. It was however, possible to measure the isobars at the (room temperature) saturation vapor pressure. For a system that conforms to the Langmuir model (eq 2) a plot of log(l/q - l/qJ vs 1 / T at constant sorbate pressure should be linear with slope -AH/R. As the isotherms for the heavier sorbates appear to conform to the Langmuir model (with qs N 12.5%) one may use eqs 3 and 4 to estimate the Henry constants and the sorption energies. Adsorption equilibrium for the lower molecular weight aromatics in NaX has been studied in considerable detail (see, for example, Barrer et al. (19751, Barthomeuf and Ha (1974), Ruthven and Doetsch (1976), Goddard and Ruthven (1984),andRuthven and Goddard (1986)). These data are not reviewed here, although Henry constants and sorption energies are included in Table I in order to show the general trends with molecular weight. Langmuir parameters, for the isotherms that conform to eq 2 (higher molecular weight sorbates at the higher temperatures) are summarized in Table 11. Mesitylene ( M = 120). The isotherms and representative Langmuir plots are shown in Figure 1. The Langmuir plots show pronounced curvature, and the saturation limit (by extrapolation to l / p 0) varies with temperature. Clearly the Langmuir model does not provide an adequate representation of the isotherms for this sorbate. The Henry constants derived from the chromatographic and gravimetric measurements show good agreement. Naphthalene (M = 128). The naphthalene isotherm and corresponding Langmuir plots are shown in Figure 2. The Langmuir plots are quite linear but the saturation limit varies with temperature. However, at temperatures above 573 K the saturation limit appears to approach a constant value (- 12.5 wt % ). 1,2,3,5-Tetramethylbenzene ( M = 134). The isotherms and corresponding Langmuir plots are shown in Figure 3. For this sorbate kinetic limitations are clearly evident with the large (50 pm) crystals at the lower temperatures. However at higher temperatures there is good agreement between the isotherms for the two crystal sizes. The Langmuir model appears to provide an accurate representation at all temperatures, and the saturation limit is essentially constant. 1,3,5-Triethylbenzene ( M = 156). As with tetramethylbenzene, diffusion is slow and reliable isotherms could be obtained only with the smaller crystals (Figure 4). The Langmuir model appears to provide a good representation of the data, and the saturation limit is temperature independent. Dimethylnaphthalenes (M = 156). Adsorption of several of the isomeric dimethylnaphthalenes was studied both chromatographically and gravimetrically. Only one full isotherms (for 1,3-dimethylnaphthalene(DMN)a t 580 K) was obtained; this is shown in Figure 5. The isotherms conform to the Langmuir model and the saturation capacity -10 wt % corresponds to about 1.1 molecules/ cage. The chromatographically measured Henry constants show little difference between the isomers, and the values are in reasonable agreement with gravimetric values from the isobar and from the limited isotherm data (Figure 7 and Table I). Hexamethylbenzene and Anthracene. For these sorbates the vapor pressures at ambient temperature are so low that only the isobars (at saturation vapor pressure, SVP) could be measured with any confidence (Figure 6a).
-
Table I. Summary of Henry's Law Constants for NaX -AUo sorbate T(K) KO KO (kcaUmol) 409 2.1 x 10' b cyclohexane 437 9.1 x 1096 5X10-9 12.5 458 4.6 X 109 488 2.1 x 109 b 437 8X106b benzene 458 2.8 x 106 b 488 9.2 X 10' 512 3.5 x 10' b 7 x 1 0 4 18.0 517 2.9 X 10' 540 1.35 X 10' 559 8x103' 581 4.6 X 109 459 1.55 X 108 toluene 488 4.0 X 106 513 1.45 X 106 6.6 X lo-' 521 1.1x 106 19.6 540 5.4 x 10'' 562 2.7 X 10' 581 1.5 X 10' 595 1.0 x 10'' 521 5 x 106' xylenes 540 2.2 x loa ' 21.5 562 1.15 X loa 4.4 x 104 583 6 x 1 0 ' ' 606 3.2 X 10' 540 5.7 x 106 ' mesitylene 553 3.2 X 106 561 2.5 X 106 573 1.75 X 106 2x104 23.3 583 1.1 x 106 ' 600 8 x 10'' 603 6 x 1 0 ' ' 523 5.8 x 108 naphthalene 548 1.8 x 108 1.2 x 104 573 6.2 X 106 25.5 603 1.85 X 106 623 1.0 x 106 1,2,3,5-tetramethyl- 547 3 x 1 0 8 benzene 579 8 x 1 0 6 3.8 x 1v 27.0 610 2.3 X 106 636 9.5 x 10' 1,3-dimethylnaph575 4.5 x l 0 8 , C 3.3 x 108 thalene 598 1.4 X 108 625 5.5 x loa 1.04 X 104 28.0 653 3 X 1 P d 658 2.3 X 105 667 1.8 x loa d 684 1.1 x 106 d 690 7.5 x 10' d 701 8 X 1 0 ' d hexamethylbenzene 515 3.4 x 107 d 531 1.2 x 107 d 5.0 X 1o-S 30.0 548 4.5 x l o s d 577 1.5 X led 579 1.5 X 108 triethylbenzene 7Xlo-S 608 4.6 X 106 31.0 633 2.0 x 106 d 658 7.3 x 10' d anthracene 613 3 X l V d 34.0 631 1.2 x l08d 2.5 X 1o-S 658 4.8 x l o a d Values are from gravimetric isotherms of present study unless otherwise indicated. All K values are expressed in dimensionless form: ((mol/cmsof crystal)/(mol/cm~of gas)). b Gravimetric data of Ruthven and Doetsch (1976). Chromatographic measurements. Some of these data are included in the thesis of Gorkey (1985). From isobars.
Assuming conformitywiththe Langmuir model, one would expect that a plot of log(l/q - l / q s )vs 1/Tshould be linear with slope AH/R, and the experimental isobars do in fact show the expected behavior (Figure 6b). Knowing the saturation vapor pressure at ambient temperature, one
Ind. Eng. Chem. Res., Vol. 32, No. 9, 1993 2049 Table 11. Langmuir Parameters (for Isotherms Which Conform to Eq 2) sorbate naphthalene
T (K)
TMB
TEB
1,3-DMN
bs. (g/(&I"I'rr))
473 523 573 603 623 547 579 610 636 579 608 633 658 580
s. (wt % 1 15
18.2 15.4
26.70 2.57 1.14 0.70 5.7 1.5 0.53 0.22 7.7 2.2 0.92 0.32 9.4
12.5 12.5
13.0 11.0
0
0.01
0.02 0.03 0.04 0.05 0.06 P (Torr)
0.3l lo3
1.o
10'
102 P (Torr)
1.o
(b)
Mesitylene
1
--
-
:: 0.6
z
573 K
-
0 \
0.4
0.2
0
I
I
10
20 1/P
1 30
(Torr-!)
Figure 1. (a) Equilibrium isotherms for mesitylene and (b) correspondingLangmuir plots showing deviating from the Langmuir model.
may calculate directly the Henry constants (eq 31, and these values are included in Table I and Figure 7. The uncertainty in the very low vapor pressure of these sorbates translates directly into a corresponding uncertainty in the Henry constants which are therefore less reliable than the values for the lighter sorbates. Henry Constants and Adsorption Energies. The temperature dependence of the Henry constants (dimensionless basis) is shown in Figure 7, and the corresponding energies of adsorption, calculated from the slopes of these plots (eq 4) are given in Table I. To provide a direct measure of the influence of the aromatic nucleus, comparative data for cyclohexane are also included. The increase in sorption energy with carbon number is approximately linear (Figure €9,and a t least up to CU,the differences between different structural isomers are evidently not significant relative to the scatter of the data. When considering the variation of the preexponential factor (KO),it is helpful to recall that this factor is directly
0 0
50 1/P
100 (Torr-!)
150
Figure 2. (a) Equilibrium isotherms for naphthalene and (b) correspondingLangmuir plots showing conformitywith eq 1 at higher temperatures.
related to the entropy change on adsorption (A&): In KO= ASdR (4) The linear paraffins show a well-defined compensation effect in which In KOdecreases linearly with increasing adsorption energy, implying a direct proportionality between A S 0 and AVO(Ruthven and Kaul). Such behavior is not uncommon, and several possible mechanistic explanations have been suggested (see,for example, Cremer, 1955)). The behavior of the aromatic hydrocarbons appears to be more complex (Figure 9). For the lower homologs KOappears to be almost constant followed by a sharp decline for the higher molecular weight species. A decrease in KOreflects an increasingly negative change in entropy corresponding to increased restriction of movement of the adsorbed molecule. From this perspective the observed pattern of behavior seems understandable. Molecules smaller than mesitylene can move quite
2060 Ind. Eng. Chem. Res., Vol. 32, No. 9, 1993
t
l2
Open Symbols 2pm 0 0.1
0
0.05
0.3
0.2
0.1
0.15
P (Torr)
P (Torr)
0 0’ 0
I
5
I
10
1
15
I
I
I
20 25 30 1 / P (Torr.’]
I
I
I
I
35
40
05
50
Figure 3. (a)Equilibriumisothermsand (b)corresponding Langmuir plots for 1,2,3,5-tetramethylbenzeneshowing conformity with the Langmuir model.
freely within the cage or channel intersection of the NaX pore system. The addition of methyl groups therefore has little effect on the rotational freedom of the adsorbed molecule. With the addition of further methyl groups or additional aromatic rings, the rotational freedom becomes increasingly restricted with a corresponding decrease in entropy. Saturation Limit. The temperature dependence of the saturation limit, derived from the Langmuir plots, is shown in Figure 10. The trends are somewhat more consistent when the saturation limit is expressed as molecules per cage, rather than as weight percent. The saturation limit decreases with increasing temperature and for all four sorbates for which the data were sufficiently extensive to allow the extraction of reliable values, the high-temperature limit appears to be somewhat greater than 1 molecule/cage. The data are consistent with the hypothesis that at sufficiently high temperatures the limit of 1molecule/cage will be approached. Such a result would explain conformity with the ideal Langmuir model, since for sufficiently large molecules steric restrictions imposed by the pore system greatly reduce the contact and, therefore, the potential for interaction between neighboring molecules.
Conclusions The adsorption of higher molecular weight aromatic hydrocarbons on NaX zeolite shows surprising simplicity in that the isotherms approach the ideal Langmuir form with an apparent saturation limit close to 1 molecule/
20
40 l/P
60 (Torrl”
100
80
120
Figure4. (a) Equilibriumisothermsand (b) correapondingLangmuir plots for 1,3,5-triethylbenzeneshowing conformitywith the Langmuir model. (Filled circles in (a) represent replicate measurements with 5O-wm crystals.)
2t/4
0
0,004
0.008 0,012
0.016
0.020
200
250
P (Torr)
0-
I
0
50
I
1
100 150 1/P (Torr)
300
Figure6. (a) Equilibrium isotherms at 580 K and (b) corresponding Langmuir plot for 1,3-dimethylnaphthaleneshowingconformitywith Langmuir model.
cage. Such behavior is intuitively understandable from steric considerations which, for large enough molecules, prohibit close sorbate-sorbate interactions. Sorption
Ind. Eng. Chem. Res., Vol. 32, No. 9, 1993 2051
\
81
Hexamethyl Benzene P=0.003 Torr
Carbon Number
Figure 8. Variation of limiting energy of adsorption with carbon number (0,uneubstituted aromatics; 0,substituted aromatics).
0.01
‘
1.4
1
1.5
I
I
1.6 1.7 1 0 4 ~( ~
I
I
1.8
1.9
I 2.0
-9
Figure 6. (a) Isobars at saturation vapor pressure and (b) corresponding plots of log(l/q - l/qJ w 1/T for hexamethylbenzene, anthracene and 1,3-dimethylnaphthalene.
I
108 NaX
I I
m
e C
0
g
10’-
i I
106-
a,
E
0, Y
io5104
-
1.3
1.5
1.7 lO’/T
1.9
2.1
2.3
2.5
[K’)
Figure 7. Temperature dependence of Henry’s law constants for aromatic hydrocarbons in NaX.
energies increase regularly with carbon number, and there appears to be little difference in adsorption equilibrium between different structures with the same carbon number such as naphthalene and tetramethylbenzene. The variation of the preexponential factor (or the entropy of sorption) is more complex. The decreasing volatility of the higher molecular weight sorbates make the accurate measurement of the isotherms difficult. As a result, the accuracy and reliability of the data for the higher homologs are substantially reduced
20
25 Au.
-
30
35
Figure 9. Variation of preexponential factor (KO) with adsorption energy.
and conclusions concerning the trends of AU or the form of the isotherms become less definite. T w o series of experiments are suggested as possible ways to confirm the conclusions from the present study. (i) Direct calorimetric measurements of heats of sorption should be more reliable than isosteric heat measurements, and if such measurements can be made as a function of loading, one has the possibility of obtaining independent confirmation of Langmuirian behavior. (ii) Measurement of binary equilibria for two large
2052 Ind. Eng. Chem. Res., Vol. 32, No. 9,1993
AVO= limiting energy of adsorption v = interstitial gas velocity c = voidage of chromatographic column s 2a
'''-
TMBTEEDMN
5 400
Literature Cited
-
500 T (IO
600
Figure 10. Variation of saturation limit with temperature for naphthalene and mesitylene: (a) wt % and (b) molecules/cage.
aromatics, such as naphthalene and tetramethylbenzene, provide a potentially sensitive test of Langmuirian behavior. We hope to pursue these approaches. If confirmed the Langmuir model should provide a comparatively simple basis from which to interpret the adsorptive and catalytic behavior of the higher aromatics and their methylsubstituted isomers.
Nomenclature b = Langmuir equilibrium constant c = Sorbate concentration in gas phase K = dimensionless adsorption equilibrium constant (q* = Kc) KO= preexponential factor (eq 4) L = column length p = sorbate pressure or partial pressure q = adsorbed-phase concentration qs = saturation limit R = gas constant AS0 = entropy change on sorption t = mean retention time (in chromatographic experiment) T = absolute temperature
Barrer, R. M.; Davies, J. A. Sorption in Decationated Zeolites. h o c . R. SOC.1970,A320,289. Barrer, R. M.; Bulitude, F. W.; Sutherland, J. W. Structures of Faujasite and Properties of Its Inclusion Complexes with Hydrocarbons. Trans. Faraday SOC.1957,53,1111-1121. Barthomeuf, D.; Ha, B. H. Adsorption of Benzene and Cyclohexane on Faujasite type Zeolites. J. Chem. SOC.,Faraday Trans. 1 1974, 70, 2147. Charnell, J. F. Synthesis of Large Crystals of A and X Zeolites. J. Cryst. Growth 1971,8,291-296. Cremer, E. The Compensation Effect in Heterogeneous Catalysis. Adv. Catal. 1955,1 , 75. Goddard, M.; Ruthven, D. M. Adsorption of C8 Aromatics on NaY Zeolite. Proceedings of the Sixth International Zeolite Conference; Olson, D., Bisio, A., Eds.; Butterworths: Guildford, UK, 1984;pp 268-275. Gorkey, Y. Adsorption of CS Aromatics in Zeolite. MScE Thesis, University of New Brunswich, Fredericton, 1985. Ruthven, D. M.Principles of Adsorption and Adsorption Processes; Wiley: New York, 1984. Ruthven, D. M.; Doetsch, I. H. Diffusion of Hydrocarbons in 13X Zeolite. AIChE J . 1976,22,882-886. Ruthven, D. M.; Goddard, M. Correlation and Analysisof Equilibrium Isotherms for Hydrocarbons in Zeolites. In Fundamentals of Adsorption; Myers, A. L., Belfort, G., Eds.; Engineering Foundation: New York, 1984,pp 533-543. Ruthven, D. M.; Goddard, M. Sorptionand Diffusion of Cs Aromatics in Faujasite type Zeolites I. Zeolites 1986,6, 275-282. Ruthven, D. M.; Kaul, B. K. Unpublished experimental data. Stach, H. Experimentelle und Theoretische Untersuchun gen zum Adsorptionsgleichgewicht von unpolaren und polaren MolekCllen an Zeolithen von Faujasittyp. Thesis (Promotion B), Academy of Sciences of the DDR, Berlin, 1977. Stach, H.; Thamm, J.; Jhchen, J.; Fiedler, R.; Schirmer, W. Experimental and Theoretical Investigations of the Adsorption of n-Paraffins, Olefins and Aromatics in Zeolites. Proceedings of the Sirth International Zeolite Conference;Olson, D., Bisio, A., Eds.; Butterworths: Guildford, UK, 1984;pp 225-231.
Received for review December 14,1992 Revised manuscript received May 28, 1993 Accepted June 11, 1993.
* Abstract published in Advance ACS Abstracts, August 15, 1993.