2555
Adsorption of Cyclohexane and Benzene on Silica
2449 (1973). (4)H. A. Berman and T. R. Stengle, J. Phys. Chem., 79, 1001 (1975).
only or microfiche (105 X 148 mm, 24X reduction, negatives) containing all of the supplementary material for the papers in this issue may be obtained from the Business Office, Books and Journals Division, American Chemical Society, 1155 16th St. N.W., Washington, D.C. 20036. Remit check or money order for $4.50 for photocopy or $2.50 for microfiche, referring to code number JPC-75-2551.
(5) J. A. Pople, W. G. Schnelder, and H. J. Bernstein, “High Resolution Nuclear Magnetic Resonance”, McGraw-Hill, New York, N.Y., 1959,Chapter 9. (6)Y. M. Cahen, P. R. Handy, E. T. Roach, and A. I. Popov, J. Phys. Chem., 79, EO (1975). (7)The donor number is the negative of the enthalpy of reaction of the donor solvent and SbC15 in the “inert” medium 1.2dichloroethane. V. Gutmann, “Coordination Chemistry in Nonaqueous Solvents”, SpringerVerlag, Vienna, 1968,p 19. (8)H. A. Berman, Ph.D. Thesis, University of Massachusetts, Amherst, Mass., 1974,and in the supplementary material. (9)K. L. Craighead and R. G. Bryant, J. Phys. Chem., 79, 1602 (1975). Chim. Belg., 49, 59 (1940). (IO)V. de Landsberg, Bull. SOC. (11) T. R. Stengle, Y. E. Pan, and C. H. Langford, J. Am. Chem. SOC., 94,
References and Notes (1)C. W. Davies, ”Ion Association”, Butterworths, Washington, D.C., 1962. (2)J. P. Coetzee and W. R. Sharpe, J. Solution Chem., 1, 77 (1972). (3)M. S. Greenberg, R . L. Bodner, and A. I. Popov, J. Phys. Chem., 77,
9037 (1972).
Adsorption of Cyclohexane and Benzene on Two Modified Silica Supports Donald Barry* and Martha Cook Department of Chemistry, Universityof Houston, Houston, Texas 77004 (Received December 4, 1974; RevisedMnuscrlpt ReceivedJuly 28, 1975) Publicafloncosts assisted by The Robert A. Welch Foundation
The adsorption of cyclohexane and benzene were studied on two modified silica supports. The first support had the form of SSi-CH2CH2CH2NH2 and the second support had the form [ 3 i C H ~ C H ~ C H ~ N H ~ P ~ ([PtCL]. N H ~ ) SThe ] methods of preparation of these supports are presented in the paper. The surface characteristics of these supports were studied using a constant volume system to determine adsorption isotherms. These isotherms were used to derive the isosteric heats of adsorption of both benzene and cyclohexane on the two supports. The isotherms were found to follow a Freundlich relationship and information about the surfaces is inferred from the coefficients of the Freundlich expressions and from the heats of adsorption.
Introduction The bonding of transition metal complexes to organic polymers or silica has provided an easy method for combining some of the most advantageous characteristics of homogeneous and heterogeneous catalysts. The major advantages of this process of “heterogenization” are the easy recovery of the catalyst and the extra selectivity provided by the presence of a surface.l The organic polymers which are used as supports have the major disadvantages that they tend to swell in organic solvents and that a great part of the catalytic reaction tends to occur in polymer channels where diffusion of reactants can become rate limiting. Grubbs and Krol12 successfully bonded Wilkinson’s catalyst to polystyrene beads and they showed that the bound catalyst exhibited a great deal of activity and selectivity in the hydrogenation of olefins. Other workers3v4have developed methods for attaching enzymes or peptides to silica and Schwetlich5 has recently adapted their method to bond transition metal complexes to a silica surface. This method involves the condensation of either a substituted trichloro- or triethoxysilane with a silica surface. The product of the reaction is a siloxane bond between the silica surface and the silane derivative. By varying the substituents attached to the trichloro- or triethoxysilanes he was able to bond a variety of
I
transition metal complexes to the silica surface. The principal disadvantage of this method is the ease of hydrolysis of the siloxane bond. Locke6 and coworkers have developed a method for bonding a variety of alkyl or aryl substituents to siliceous surfaces using Grignard reagents. This method has the principle advantage of leading to a hydrolytically stable silicon-carbon bond. We have extended this method so as to allow us to bind Magnus salt, [Pt(NH&][PtC14], to a modified silica surface. Our method can be extended to allow the binding of any amine containing complex to a silica surface. The crystal structure of Magnus salt has been shown to consist of metal complex units stacked in columns with the metal atoms in close enough proximity to bond to each othern7It has also been found that Magnus salt can display anisotropic semiconductor b e h a ~ i o r in ~ , ~the direction of the metal-metal bonds. Since the structure of Magnus salt exhibits coordinative unsaturation at the surface, we felt that the salt would provide a useful link between platinum metal catalysts and their homogeneous analogs. The behavior of the bound Magnus salt was studied by determining the physical adsorption of benzene and cyclohexane on it. It has been shown that the shape of isatherms can be somewhat sensitive to the surface structure of the adsorbent.1° Additionally, the magnitude of the heats of The Journal of Physical Chemistry, Vol. 79, No. 23, 1975
2556
Donald Barry and Martha Cook
adsorption derived from these isotherms can be indicative of surface structure. Experimental Section Preparation of Chemically Bound Magnus Salt. The Magnus salt was bound to Porasil E (Waters Associates) using the following reaction scheme: \I
,Si-OH
\
/
+ Tic14
Ai-Cl t HzC=CH-CHzMgBr
-
+
>kCl
(1)
Si-CHz-CH=CHZ
The replacement of the surface hydroxyls on the Porasil
E was done by refluxing 25 g of TiC1411 in 100 ml of dry pentane with 10 g of Porasil E. The chlorinated silica was then throughly washed with dry pentane and dried in a vacuum oven overnight. The Grignard reagent was prepared by the reaction of 5 ml of 3-bromopropene, (Matheson Coleman Bell) which was used without further purification with a large excess of Mg (10 g) in 100 ml of anhydrous ether. A crystal of I2 was added to facilitate the reaction. The large excess of magnesium was used to prevent coupling between the Grignard reagent and the 3-bromopropene. The Grignard reagent was added to the chlorinated silica and the reaction was assumed to be complete after the bubbling ceased. Additional samples of the Grignard reagent did not react any further with the silica and it was therefore assumed that the entire silica surface had reacted. The treated silica was then filtered and washed repeatedly with water to hydrolyze the excess Grignard and to remove the salts. The treated silica was then washed with ether, acetone, and chloroform. The nonbonded organics were removed by heating the treated silica in a vacuum oven at 100°C for 48 hr. The sample was then analyzed for carbon content by heating it in a porcelain crucible and determining the weight loss. The carbon moiety bound to the silica was aminated using the procedure of Brown and coworkers.12In this reaction, 30 ml of a 1.89 M solution of diborane in tetrahydrofuran, under a nitrogen stream, was added by syringe to 10 g of the treated Porasil E which was covered by 40 ml of freshly distilled tetrahydrofuran. After 1 hr, 10 ml of water was added to destroy any residual hydride and this was followed by addition of 70 ml of 3 M NaOH. To this mixture was added 250 ml of a freshly prepared solution of 0.4 M chloramine. The solution was allowed to react for 2 hr at room temperature and was then acidified. The sample was washed throughly with ether and it was then placed in 40 ml of tetrahydrofuran which had been made strongly alkaline. After stirring for 20 min, the treated silica was removed and washed with water and ether and then dried. The platinum tetrammine was prepared from platinum tetrachloride (Englehart Industries) by the method of Keller.I3 A solution of 250 mg of [Pt(NH3)4]C12 in 250 ml of The Journal of Physical Chemistry, Vol. 79, No. 23, 1975
water was added to 5 g of the treated silica. The solution of the platinum salt was equilibrated with the treated silica for a period of 2 days. The stirred mixture was heated at 6OoC and the reaction was allowed to continue until no further ammonia evolution could be detected. The treated silica was throughly washed with hot water and dried. A sample of the silica was then removed for analysis of the platinum content. The platinum complex was removed from the silica by treatment with concentrated nitric acid. The solution was then filtered and diluted to a known volume. The platinum content was determined using a Perkin-Elmer 303 atomic adsorption instrument. The treated silica containing the platinum tetrammine moiety was then allowed to react with a solution of 250 mg of KzPtC14 in 70 ml of water over a 2-hr period. The sample was washed throughly with hot water and dried. A sample of this material was also removed for analysis. The same procedure as given above was used to analyze for the platinum content. The mesh size of all the adsorbents was 80-120. Infrared spectra of the porasil, the aminated porasil, and the porasil with bound Magnus salt were taken with a Beckman IR 4250 spectrophotometer. In addition, a reflectance uv-visible spectra of the bound Magnus salt was taken on a Beckman DK-2A spectrophotometer using MgO as the reference. Isotherms Apparatus. The apparatus used is shown schematically in Figure 1. A vacuum line with just a fore pump was used to evacuate the system. The valves used were Kontes Teflon values (5 mm) where the vacuum seal was formed by a Teflon to glass seal. All Viton 0 rings were shielded by Teflon rings from exposure to the adsorbate. The pressure readings were made using a Texas Instruments Model 145 precision pressure gage in conjunction with a Type 2 bourdon capsule. The dotted lines in Figure 1 indicate the part of the system which was kept thermostated to fO.l°C. The temperature control unit was a Yellow Springs Instrument Model 71 used in conjunction with a variac to control a 20ohm heating coil. The vapor pressure of the adsorbate was controlled by either heating or cooling the adsorbate reservoir. The volume between values two and three was calibrated using mercury. Materials. The cyclohexane (Aldrich Chemical Co., Inc.) used was found to be free of any chromatographic impurities and was used without any further purification. The benzene (Mallinckrodt, Nanograde) used was also found to be free of any chromatographic impurities and it too was used without further purification. Procedure. In a typical run, 1.3 g of the bound Magnus salt was placed in the adsorbent reservoir and this' was sealed on to the system. A sample of cyclohexane was placed in the adsorbate reservoir and frozen. All valves were opened and the system was evacuated. The system was then thermostated for the desired temperature. Valves 1 and 3 were closed and the cyclohexane was slowly melted until the desired pressure was attained. At this point, valve 2 was closed and valve 3 opened, allowing the vapor to adsorb on the bound Magnus salt. After the system had been equilibrated and data collected, the cyclohexane was refrozen, and valves 1 and 2 were opened allowing the system to evacuate. In the case of benzene it was necessary to heat the adsorbent to ensure that no benzene remained. The
2557
Adsorption of Cyclohexane and Benzene on Silica
Varuum
I
Proposed structure for bound Magnus salt. NH3 groups are indicated by 0 and CI- groups are indicated by and Pt by 0 .
Figure 2.
I
II
question now arises as to how these bound metal complexes are orientated on the surface. If we assume the surface of L_ _ - _ - - - - J the silica is smooth, then we would expect that the total Figure 1. Schematic diagram of adsorption apparatus. 1, 2, and 3 surface area of the bound Pt(NH&NH2-R-Si$ groups are teflon valves. Dotted line indicates thermostated region. would be equal to or less than the total surface area of the silica. Messmer15 has noted that the area of a [PtC14]2group is 32 and if we assume that the area of the dead volume, the volume after valve 3, was calibrated using Pt(NH3)3NHz-R-Si< is about the same then the total argon. area of the bound platinum complex is (5 X X (32 Results and Discussion or 16 m2/g. This is four times the surface area of the untreated silica. We can conclude that the surface of the silica Bound M a g n u s Salt. The carbon content of the $Siis not smooth but rather is very rough and contorted in CHzCHCH~species was found to be 0.3 f 0.5% by weight. order to accommodate the amount of platinum complex The error limits were based on the two separate determinaknown to be present. We are still left with the initial questions made for the carbon content. Locke6 has found for a tion as to how many of the bound Magnus salt species are similar reaction that his carbon content was 3% by weight exposed enough to interact with an adsorbate. As a first apfor the formation of the same species. Since his silica was proximation, we will assume that the maximum surface Porasil C with a surface area of 50 m2/g and ours was Porarea of the exposed bound Magnus salt (the “active” site) asil E with a reported surface areal4 of 4 m2/g, we should cannot be greater than the total surface area of the silica or expect a yield of 0.25% which is within the error limits of 4 m2/g. Using this analysis, we find that there should be our yield. Assuming that the reaction to form the amine only 1.25 X 1019 active sites on the surface. Our picture of goes to completion, the expected weight percentage of the bound tetrammineplatinum(I1) chloride should be 1.5%. the bound Magnus salt surface would have only a fourth of the bound complex exposed while the rest of the bound salt The analysis of the platinum content by atomic absorption would be in crevices and valleys and would not be exposed indicated that the platinum content was 1.7% by weight. to the adsorbate because these metal species were shielded The fact that our yield of 1.7% of platinum, by weight, is from exposure by other molecules of the bound Magnus very close to the theoretical weight percentage of 1.5%indisalt. Further substantiation of this analysis may be found cates that there has been total conversion of all surface 1Si-OH / species to ~ S ~ - C H ~ C H C H ~ N H Z P ~The ( N H ~ )in~the . section on cyclohexane adsorption. The ir spectrum of the bound Magnus salt was taken color of the silica, a t tf& point, was a very light yellow. The from a powder sample which had been mixed with KBr. platinum content after the addition of the potassium tetraThe ir spectrum of the porasil itself indicated that the silichloroplatinate(I1) to the treated silica was a 3.3% or twice ca absorbed in all of the regions of interest. Another probas high as the platinum content found for the addition of lem with the spectrum is the fact that the concentration of the tetrammineplatinum(I1) chloride. The reaction to form the bound Magnus salt is very low. Thus we were able to the bound Magnus salt occurred very rapidly and led to a discern only one weak shoulder at 465 cm-’ which did not product which was purple in color. The extreme change in appear in the porasil spectra. It has been reported16 that color and the fact that the final platinum content was twice the v(Pt-N) in Magnus salt occurs at 500 cm-l. We feel the original platinum content led us to believe that the that the peak a t 465 cm-I could be due to the platinumbound Magnus salt had been formed. The color difference nitrogen stretch, where stabilizing interactions with the silbetween pure Magnus salt (green) and our bound sample ica surface would tend to shift the Pt-N stretch to lower could be ascribed to the presence of the silica and also to wave numbers. We could not discern any difference bethe difference in long-range structure between the two tween the ir spectra of porasil E and the aminated porasil E compounds. The proposed structure of the bound com(species 1). The uv-visible reflectance spectra of a powpound is shown in Figure 2, where the ligands are twisted dered sample of Magnus salt taken in this laboratory in the same manner as is found in the crystal. showed two absorbance peaks a t 425 and 625 nm. The uvWe have assumed, in our proposed structure of the visible spectrum of the bound Magnus showed a very weak, bound Magnus salt, that only a single molecule rather than broad absorbance a t 700 nm. We feel this peak is the one a polymer of Magnus salt is found per surface hydroxyl found at 625 nm in Magnus salt which has been changed in group. We have based this conclusion in part, on the final energy because of the interaction of the bound Magnus salt platinum percentages and, in part, on the fact that the with the silica surface. The other peak is most likely hidden treated silica surface was throughly washed after equilibration with the specified platinum complex. We would expect in the broad absorbance band: that the platinum-platinum bond found in crystalline Adsorption Magnus salt would still exist in our sample of the bound Magnus salt and that there would be a coordinatively unCyclohexane. The adsorption isotherms for cyclohexane saturated or “active” site for every molecule of bound Magon the bound Magnus salt are shown in Figure 3. These isonus salt. A simple calculation, based on our platinum contherms were derived assuming that cyclohexane followed tent, reveals that there should be 5 x 10’9 molecules of the ideal gas law and calculations using the van der Waals bound Magnus salt on the surface of the treated silica. The equation showed that the deviation was at most 1%a t the TO
awm
aullgw
a2
a’)
The Journal of Physical Chemistry, Vol. 79, No. 23, 1975
Donald Barry and Martha Cook
2558
P (mm)
Figure 4. Adsorption isotherms of cyclohexane on the aminated silica. The temperatures and corresponding symbols are 301.7 K (0), 317.6K (m),332.0K (A)and 348.0 K (0).
P (mm)
15.1
.
11.4
.
Figure 3. Adsorption isotherms of cyclohexane on the bound Magnus salt. The temperatures and corresponding symbols are 300.0 K (o),317.4 K (M), 332.4 K (A),and 348.0 K (0).
higher pressures. The figure plots a in wmoles adsorbed/ gram of adsorbent vs. the equilibrium pressure. The general shape of these isotherms indicates that they are a modified type I1 is0therm.l' The adsorption isotherms for cyclohexane on the aminated silica are given in Figure 4. Again the general shape of the isotherms indicates that they are modified type I1 isotherms. Since the data was only taken up to a region of PIP0 of 0.4, it was not possible to exclude the possibility that the isotherms were type IV. The type IV isotherm implies the presence of porous substructure in the adsorbent and since the total surface area of these adsorbents was only 4 m2/g, it was felt that no such substructure could exist and therefore these isotherms were of the type I1 variety. The determination of the monolayer was done using the form of the BET equation proposed by Keii and coworkers:l*
1 a(1-X)
(F)
=1 +1 am amC
e(Qads-QL)lRT
b la,
(11)
where alb,/bla, refers to the ratio of frequency terms between various phases, && is the heat of adsorption, and QL is the heat of liquification. As other workers have shown19*20 this form of the BET equation is only applicable for a uniform surface. The calculation of a, as Halsey has noted, appears to be quite valid from the BET equation, however, the C values calculated using the BET technique can differ widely from the C values using other techniques. The a, and C values calculated for cyclohexane adsorption The Journal of Physical Chemistry, Vol. 79, No. 23, 1975
I D1
O(1-X)
7.6
I
4.6
I
I
I
1
I
3
6
9
11
I5
Figure 5. BET plot for the cyclohexane adsorption on the bound Magnus salt. The temperatures and corresponding symbols are 300.0K (O),317.4K (H), 332.4K (A),and 348.0K (0).
(1)
where X refers to the ratio of the equilibrium pressure to the saturation vapor pressure of the adsorbate at the same temperature and a, refers to the value of a a t '6 = 1. An illustration of the BET plots for cyclohexane on bound Magnus salt using this relationship is shown in Figure 5. The intercept is equal to l / a , while the reciprocal of the slope times the intercept gives the value of C. In the simple form of the BET equation:
C =I alb e ( Q a d s - Q L ) / R T
-
on the bound Magnus salt are given in Table I. The corresponding values for the adsorption on the aminated silica are given in Table 11. The results indicate that the a, values remain constant over the temperature range for both the aminated silica and the bound Magnus salt. The C values tend to decrease with increasing temperature and this is in agreement with the simplified expression of C given in eq 11. The values of Qads - QL, or the net heat of adsorption, are also given in these tables and it should be noted that this quantity does not remain constant. The average value of a, for the bound Magnus salt is 21.7 pmol/g and this corresponds to a monolayer coverage of 1.31 X 1019 molecules of cyclohexane/g of adsorbent. If we assume that the area of the cyclohexane is approximately V2I3and calculate V from the liquid density, we find that the molecular area equals 31.8 A2 or 4.2 m2/g. Since the surface area of the bound Magnus salt is 4.2 m2/g then from our previous calculation only 1.25 X 1019 molecules of Magnus salt can be exposed to the adsorbate. It also appears that one cyclohexane molecule is adsorbed per exposed metal site. The surface area found for the aminated silica is 4.04 m2/g.
Adsorption of Cyclohexane and Senzene on Silica
2559
TABLE I: BET Values for the Bound Magnus Salt
In (a KIO') 0.275
am7
T ,K
wol/g
22.1 300.0 317.4 22.1 332.4 21.2 348.0 21.6 a Both Qadaorl,tlnn and Q ues .
~
C
i
~
1.0 7 5
8x18- QL, kcal/mol'
5.22 4.80 4.08 2.93 ~ are ~ taken f l
0.99 0.99 0.95 0.74 to ~ be ~ positive t ~ ~ val~
TABLE 11: BET Values for the Aminated Silica
2.675
1.875
I
I
I
I
I
1
I
I
t
9.1 PO'
I
0
I
I
5.1
I
10.4
a (rrmolelg)
301.7 317.6 332 .O 348.0
20 .o 22.2 21.1 21.2
13.4 9.5 7.5 6.6
1.55 1.42 1.41 1.31
Figure 6. Heat of adsorption as a function of coverage for the bound Magnus salt (solid line) (0). Heat of adsorption as a function of In a is given as the dotted line (0).
Again, there appears to be an average of one cyclohexane adsorbed for every active site. The isosteric heats of adsorption were derived from the isotherms using the Clausius-Clapeyron relationship a t a constant a value, viz
The relationship between Qads and a for adsorption on the bound Magnus salt is shown in Figure 6 and for adsorption on the aminated silica it is shown in Figure 7 . &ads drops rapidly off for the adsorption of cyclohexane and up to an a value of 10.5 it follows a relationship of Qads (constant) In a. At a = 10.5, which corresponds to 6 = $, Q begins to fall quite rapidly and at this point multilayer formation is just beginning and a rapid drop in Q should be expected as the Q is now a function of both &ads and QL.The major information about the adsorbent is contained in the region of a = 0 to a = 5. It is in this region that no multilayer formation occurs and that no cooperative phenomena between the cyclohexane molecules can occur. For the cyclohexane on the bound Magnus salt this is the region of the steepest fall for Qads and this indicates that the surface is heterogeneous. The availability of different energy sites on the bound Magnus salt may be due to two types of sites available, namely, the silica itself and the platinum moiety or it may be due to geometrical distortions on the surface which we have shown must exist. The heat of adsorption of cyclohexane on the aminated silica shows a small initial rise in Qads followed by a rapid decrease. The rise at small a values may be an experimental artifact or it could possibly be due to repulsions between the polar aminated surface and the cyclohexane. In either case there is a large difference in surface characteristics between the bound Magnus salt and the aminated silica. The large difference in Qads between both systems leads to the conclusion that the bound Magnus salt has a stronger interaction with the cyclohexane than does the silica. The calculated values of the net adsorption from the C values of the BET equation do not appear to agree with these conclusions. However, since the C value in BET calculations assumes a constant Qads, and this is obviously not the case, then we will probably get a BET Qads value which is averaged over the entire 6 range.
0.4
1 0
I
I
I 5.2
lo 10.4
awmole I g 1
Figure 7. Heat of adsorption as a function of coverage for cyclohexane on the aminated silica (solid line) (0).Heat of adsorption as a function of In a is given as the dotted line (0). By coincidence this will give a higher net absorption value for the silica then the bound Magnus salt. It has been proposed1° that the limits of physical adsorption are given by the following inequality
5.5RT 1 Qads - QL (IV) where T is the highest temperature possible for physical adsorption. Applying this to our system and assuming that T = 353 K, we find that Qads can have an uppermost value of 11.75 kcal/mol where QL is 7.9 kcal/mol.21 Since the maximum value of &ads for either of the adsorbents occurs with the bound Magnus salt a t a = 0 and is extrapolated to be 10.25 kcal/mol, we feel sure that we are studying reversible physical adsorption. The adsorption isobars, derived from the isotherms, do not show the type of behavior expected for systems having both chemical and physical adsorption and this information further supports the conclusion that we are just studying physical adsorption. The Journal of Physical Chemistry, Vol. 79,No. 23, 1975
Donald Barry and Martha Cook
2560
TABLE 111: Values of Freundlich Coefficients for the Bound Maanus Salt
TABLE V: BET Values for the Bound Magnus Salt 1o6a,,
Qads
- QL,
T,K
106d
l/n
Q,, cal/mol
T ,K
wol/g
C
cal/mol
300.0 317.4 332.4 348.0
1.28 0.67 0.32 0.13
0.825 0.840 0.890 0.924
723 751 742 748
298.5 317.5 332.4 348.1
25.6 27.1 31.7 32.2
17.40 12.30 13.70 11.40
1694 1583 1728 1683
TABLE IV: Values of Freundlich Coefficients for the Aminated Silica T,K
106d
l/n
301.7 317.6 332 .O 348 .O
1.97 0.84 0.46 0.29
0.827 0.857 0.879 0.910
Q,,
TABLE VI: BET Values for the Aminated Silica
cal/mol 725 736 750 760
~
302.1 317.2 332.6 348.3
32.2 32.2 30.3 29.4
~~
30.0 46.6 34.6 27.8
~
2042 2421 2342 2301
Magnus salt in Table I11 and for the aminated silica in Table IV. In the theoretical derivation of the Freundlich isotherm the quanities d , and lln are given below:1°
RT
1/n = 7
where Qm is a constant and
where f A is the partition function of the adsorbed species, Fg is the partition function of the gas phase per unit volFlgure 8. Adsorption isotherms of benzene on the bound Magnus ume, and A is a conversion factor. For the bound Magnus salt. The temperatures and corresponding symbols are 298.5 K (O), salt Qm = 741 f 5 and for the aminated silica Qm = 145 f 317.5 K (A),332.4 K (H)and 348.t K (0). 5 or they are identical. We may ascribe the constant Qm to being a function of the adsorbate and it appears to be relatively insensitive to the nature of the adsorbent. The ratio l5 r of the d coefficients of the equation leads to the following relationship: P imml
12
-dsilica
-
d M . salt - 9
P
(ao)silicaRT'Qm (a0)Magnus saltRT/Qm
Since Qm is the same for both species the ratio reduces to the followingat the same temperature:
1
(
O 6
Pimml
Flgure 9. Adsorption isotherms of benzene on the aminated silica.
T h e temperatures and corresponding symbols are 302.1 K (0), 317.2 K (A),332.6 K (H),and 348.3 K (0).
Since both the aminated silica and the bound Magnus salt appear to follow the relationship of Qads = (constant) X In a , it was determined that they would follow a Freundlich isotherm. The derivation of the isotherm which has the form a = dP1/",where both d and l / n are temperature dependent variables, is based on the concept of surface heterogeneity. The values of d and l / n are given for the bound The Journal of Physical Chemistry, Vol. 79, No. 23, 1975
Csilica fA(si1ica) >RT/Qm CMCsalt fA(MagnUS Salt) Since RTIQm > 0 for our system and Csilica 1 . 5 c M . C . salt (at T = 300 K) then fABilica > fAMagnus Green This result can be interpreted in terms of the greater mobility of cyclohexane on the silica. Benzene. The adsorption isotherms for benzene on the bound Magnus salt are shown in Figure 8. The general shape of the isotherms indicates that they are modified type I1 isotherms. The adsorption isotherms for the modified aminated silica are shown in Figure 9 and they also indicate that the isotherms are modified type I1 isotherms. In these cases the ratio of PIP0 had as upper limits a value of 0.10. Since these upper limits are on the borderline of BET linearity the a, and C values derived from these isotherms are very suspect. However, the extrapolated values of a , and C from these isotherms using the previously mentioned eq I are given in Table V. The bound Magnus salt
2561
Adsorption of Cyclohexane and Benzene on Silica In i a x IO? 1.275
0.275 11.5
10.7
TABLE VII: Values of Freundlich Coefficients for the Bound Magnus Salt
2.275
‘i;
T,K
1 06d
l/n
Q,, cal/mol
298.5 317.5 332.4 348.1
4.08 1.58 0.79 0.50
0.709 0.789 0.867 0.881
837 800 762 785
TABLE VIII: Values of Freundlich Coefficients for the Aminated Silica
I
8.3
1 5.2
10.4
15.6
aiumolelgi
Flgure 10. Heat of adsorption as a function of coverage for benzene on the bound Magnus sa%(solid line) (0).Heat of adsorption as a function of In a is given as the dotted line (A).
10.3
r9
I\
aiumolelgi
Figure 11- Heat of adsorption as a on the aminated silica.
function of coverage for benzene
has an average value for a, = 29.2. The values for a, and C derived for the aminated silica are given in Table VI. The average value of a, is 31.0. The area of benzene derived from its density is 27.9 A. The surface area derived from the equation (a,)(area) X 6 X is equal to 6.4 m2/g. Since this calculation is only approximate this is in fairly good agreement with the known surface area. The values of C in both Tables V and VI do tend to decrease with increasing temperature, however, the uncertainty in their values is very large. Since the interaction of the surface is greater with benzene than with cyclohexane, it is reasonable to expect that the deviation from the BET plot would be much larger. The values of the net adsorption at each of the temperatures is given in Table V and Table VI. The values for the bound Magnus salt predict a net heat of adsorption of 1675 cal/mol. The values for the aminated silica predict a value of 2276 cal/mal. This could be interpreted to mean that the heat of adsorption is greater on silica, however, the degree of error in the values of both a, and C is relatively large and thus an interpretation of these values will not be attempted.
302.1 317.2 332.6 348.3
7.1 3.9 2.9 1.4
0.706 0.732 0.750 0.815
850 861 881 849
The heats of adsorption derived from the isotherms of the benzene adsorption on the bound Magnus salt are given in Figure 10. The plot shows a rapid fall in Qads a t low 6 values. The relationship of Qads = (constant) ln a is linear and thus the isotherms follow a Freundlich relationship. The extrapolated value of Qads at a = 0 (6 = 0 ) shows that QadS = 12.0 kcal. Since the reported value of QL for benzene over this temperature range is 8.09 kcal, the net heat of adsorption at a = 0 is 3.91 kcal/mol. It appears that Qads approaches QL as 0 approaches 1. A simplified calculation of the average value of the net heat of adsorption would be given by [(Qads),+O - Qvap]/2 = 1.95 kcal/mol and this is just slightly higher than the value we calculate from the C values of the BET plot. The relationship between the heat of adsorption and a for benzene on the aminated silica is given in Figure 11. Again there is a rapid drop in Qads with a but in this case it is more linear than logarithmic and it could be interpreted on the basis of a Tempkin isotherm, however, for the convenience of comparison we will also assume that the adsorption on the aminated silica follows a Freundlich isotherm. The extrapolated value at a = 0 shows a Qads = 10.5 kcal for the aminated silica. The maximum net heat of adsorption is then 2.41 kcal/mol. This indicates that in the area of the absorbate-adsorbent interaction (6 = 01, the bound Magnus salt shows an interaction with benzene which is 1.5 kcal/mol stronger than the aminated silica. Unger22 had reported that the net isosteric heat of adsorption for benzene on a fully methylated silica is 1.6 kcal/mol and on a fully phenylated silica it is 1.3 kcal/ mol. Both of these values are lower than either of the values we find for the bound Magnus salt or the aminated silica. Using the previous approximation we find that the C value for the aminated silica should be of the order of 1.2 kcal/mol; the reported C value is far too high and this may indicate that the silica does not approach QL as 0 approaches 1 but rather some much higher value. The uncertainty in the C values for the treated silica are of a much higher order than the Magnus salt. Using eq IV and assuming that T = 353, the maximum net heat of adsorption permissible for physical adsorption is 3.86 kcal. The adsorption of benzene at 6 = 0 is above this figure and thus is on the borderline between physical and chemical adsorption. The adsorption on the aminated silica has a maximum net heat of adsorption of 2.41 which is clearly within the physical adsorption range. The isobars The Journal of Physical Chemistry, Vol. 79, No.23, 1975
David L. Beveridge and Gary W. Schnuelle
2562
of benzene adsorption on both the aminated silica and the bound Magnus salt both indicate that only one process, physical adsorption is taking place. The adsorptions in both cases are therefore reversible, physical adsorption. The drop of the value of Qads below the value of QL for the benzene adsorption on the aminated silica may be caused by repulsions between the adsorbed species. The rapid return of the curve to the QL value may be caused by the onset of cooperative phenomena between the adsorbed phase and the forming multilayers. The Freundlich coefficients for the bound Magnus salt are given in Table VI1 and for the aminated silica in Table VIII. The average Qm value is 796 f 10 for the bound Magnus salt and for the aminated silica it is 860 f 5. There does appear to be a surface effect upon the value of Qm in this case. The ratios of CsiliJCbenzene range from 1.75 to 3.10. In all cases then fA(Silica) > f A~~~~~~~ and this ratio is greater than the one seen with the cyclohexane adsorption. We can conclude then that the benzene is more mobile on the aminated silica than on the bound Magnus salt. Conclusions The heats of adsorption derived from the isotherms of both cyclohexane and benzene on an aminated silica and a bound Magnus salt indicate that the interaction of the adsorbate with Magnus salt is greater than the interaction with an aminated silica. The adsorption interaction of both the cyclohexane and the benzene appears to be a one siteone adsorbate species interaction. Overall the adsorption of the benzene on the bound Magnus salt approaches chemisorption. However, all of the adsorptions were reversible physical adsorption. The system of bound Magnus salt is thus an ideal candidate for a model catalytic system and we are now studying the catalytic possibilities of this system.
Acknowledgments. We wish to thank the referees for their helpful comments and suggestions. We are indebted to the Robert A. Welch Foundation and the Research Corporation for supporting this research.
References and Notes (1) N. Kohler and F. Dawans, Rev. lnst. Fr. Pet. Ann. Combust. Liq.. 27, 105 (1972). (2) R. H. Grubbs and L. C. Kroll, J. Am. Chem. Soc., 93, 3062 (1971). (3) H. H. Weetall, Science, 166, 615 (1969). (4) W. Parr and K. Grohmann, Tetrahedron Len., 2633 (1971). ( 5 ) K. Schwetlick, J. Pelz, and K. Unverferth. Proc. lnt. Conf. Coord. Chem., 16th, 44 (1974). (6)D. C. Locke, J. T. Schmermund, and B. Banner, Anal. Chem., 44, 90 (1972). (7) M. Atoji, J. W. Richardson, and R. E. Rundle, J. Am. Chem. SOC.,79, 3017 (1957). ( 8 ) L. V. lnterrante and F. P. Bundy, lnorg. Chem., 10, 1169 (1971). (9) B. G. Anex, S.I. Foster, and A. F. Fucaloro, Chem. Phys. Lett., 18, 126 (1973). (10) D. 0. Hayward and B. M. W. Trapnell, “Chemisorption”, Butterworths, London, 1964. (11) It has been noted by the referees that the use of TIC14 to chlorinate the surface could lead to surface bound Ti species. They have suggested that SOClp would be a far better chlorinating agent. (12) H. C. Brown, W. R. Heydkamp, E. Breuer, and W. S. Murphy, J. Am. Chem. SOC.,86,3565 (1964). (13) R. N. Keller. “Inorganic Synthesis”, W. C. Fernelius, Ed. Vol. 2, Wiley, New York, N.Y., 1946, p 250. (14) This is the value of the surface area reported by the manufacturer (Waters Assoc.). (15) R. P. Messmer, U. Wahlgren, and K. H. Johnson, Chem. Phys. Lett., 18, 7 (1973). (16) R. J. H. Clark and C. S. Williams, J. Chem. SOC.A, 1425 (1966). (17) S. Brunaver, “The Adsorption of Gases and Vapors”, Vol I, Princeton University Press, Princeton, N.J., 1945. (18) T. Keii, T. Takagi, and S. Kanetaka, Anal. Chem., 33, 1967 (1961). (19) G. Halsey. J. Chem. Phys., 16, 931 (1948). (20) W. A. Steele. J. Chem. Phys., 25, 819 (1956). (21) American Petroleum Institute, Report on Research Project 44, Supp. VOl. A-62, 1972. (22) K.Unger, Agnew. Chem., lnt. Ed. Engl., 11, 267 (1972).
Free Energy of a Charge Distribution in Concentric Dielectric Continua David L. Beverldge, and Gary W. Schnuelle Department of Chemistry, Hunter College of The City university of New York, New York, New York 10021 (Received April 16, 1975) Publicatton costs asslsted by CUNY Faculty Research A ward Program
A theoretical treatment for dealing with the energetics of an arbitrary charge distribution imbedded in a central spherical cavity surrounded by two concentric dielectric continua is described. The results provide a general means for treating environmental effects using the continuum model. The form of the solution is particularly suited to identifying the contributions of the various dielectric regions.
I. Introduction The simplest model for the theoretical treatmentof ronmental effects on structure, properties, and chemical remodel.l actions in liquids and so~utionsis the the continuum model, the dissolved system is represented as discrete charges q k , k = 1, M or a charge distribution p(r) in a cavity imbedded in a structureless polarizable dielectric. The charges induce a reaction potential in the diThe Journal of Physical Chemistry, Vol. 79, No. 23, 1975
electric. The reaction potential acts back on the dissolved charges. The energy of interaction of the distribution with the environment is just the reversible work involved in charging the distribution in the presence of dielectric. This energy is thus a Helmholz free energy of polarization 1
A = - E qk@pR(rk)= 2 k
j’
p(r)+R(r)
dr
(1)
