J. Phys. Chem. B 2000, 104, 7311-7319
7311
An IR, FR, and TPD Study on the Acidity of H-ZSM-5, Sulfated Zirconia, and Sulfated Zirconia-Titania Using Ammonia as the Probe Molecule R. Barthos, F. Lo´ nyi, Gy. Onyestya´ k, and J. Valyon* Institute of Chemistry, Chemical Research Center, Hungarian Academy of Sciences, P.O. Box 17, H-1525 Budapest, Hungary ReceiVed: March 13, 2000; In Final Form: May 16, 2000
Infrared (IR) and frequency-response (FR) spectra of zeolite H-ZSM-5, ZrO2/SO42-, and ZrO2-TiO2/SO42were recorded under 133 Pa of NH3 pressure in the temperature range 293-673 K. It was shown that ionpair complexes comprising the conjugated base sites of the solid Bro¨nsted acid and H-bonded [NH4‚nNH3]+ associations were formed. Smaller associations or NH4+ ions of higher acid strengths were obtained as NH3 coverage decreased at higher temperatures. Desorption of NH3 was accompanied by proton back-transfer from the cations to the sulfated oxides, indicating that the association contributed significantly to the energy stabilizing the ion pair. In the NH3/H-ZSM-5 system, virtually all the protons remained localized in associations or NH4+ ions up to 673 K. Thus, the NH4+ ion-zeolite framework interaction stabilizes the ion pair more effectively than the interaction of the NH4+ ion with the sulfated zirconia. The deprotonation energy of the acid sites and also the stabilization (media) effect determine the efficiency of the acid in protonating a base, i.e., the acid strength. Results suggest that deprotonation energy alone, or any spectroscopic parameter reflecting the strength of the O-H bond, is not sufficient for comparing the acidities of solids that are chemically and structurally as different as sulfated zirconia and zeolite. In acid-base interactions, H-ZSM-5 exhibits stronger acidity than ZrO2/SO42- due to the better stabilization of the adsorbed base or the ion pair in the zeolite channels than on the zirconia surface. Results of IR, FR, and temperature-programmed desorption (TPD) examinations suggest that sulfated ZrO2 contains two kinds of Lewis acid sites of distinctly different acid strengths and Bro¨nsted sites with a broad acid strength distribution.
Introduction In the chemical and refinery industries, solid acids are of great importance as catalysts or as supports for other active components. Zirconia modified with sulfate ions (ZrO2/SO42-) was found to be a highly acidic solid that can catalyze a large number of reactions of industrial importance.1,2 Numerous studies concern its acidity; however, there is controversy regarding the nature and the strength of the acid sites. Sulfated zirconia was found to be active in the lowtemperature isomerization of n-butane.3,4 The reaction was believed to proceed at very strong Bro¨nsted acid sites; thus suggesting that ZrO2/SO42- was behaving as a solid superacid. The Hammett acidity function (H0), determined by the indicator method, was low enough (H0 , -12) to support this notion.1,2,5,6 However, by use of a spectrophotometric modification of the indicator method, ZrO2/SO42- was found to have an acid strength close to that of 100% H2SO4 (H0 ≈ -12).7 These data suggested that ZrO2/SO42- was a weaker acid than the mild superacid high-silica zeolites. In general, acid strength is measured by studying the interaction of an acid with a base. For instance, adsorption calorimetry and temperature-programmed desorption (TPD) permit acidity to be characterized by adsorption heats and desorption activation energies, respectively. An acid-base reaction is also followed when protonation efficiency is determined by the indicator method. The results reflect all the * To whom correspondence should be addressed. Fax: (361) 325-7750. E-mail:
[email protected].
possible interactions of the base, including possible solvation effects and all the interactions with the solid media. These effects were considered to introduce some uncertainty into the evaluation of the data. Therefore, the methods often preferred are those not involving any acid-base interaction, such as theoretical calculations and some IR8,9 or 1H MAS NMR10,11 spectroscopic examinations. The latter methods provide us with the deprotonation energy (DE) of a Bro¨nsted acid site or with parameters that can be directly related to the DE. The sulfate modification of zirconia induced a very significant downfield shift of the 1H MAS NMR signal (δH). Since the absolute value of the chemical shift was much higher than that for zeolite H-ZSM-5, it was concluded that ZrO2/SO42- is the stronger Bro¨nsted acid, most probably a superacid.10b The theoretical calculations of Babou et al.12 substantiated that the superacidity is of the Bro¨nsted type. In contrast, IR, TPD, and calorimetric studies of base adsorption13,14 showed that the strongest acid sites are of the Lewis type. The Bro¨nsted sites adsorbed NH3 with a heat of adsorption of about 125-145 kJ mol-1.14 This adsorption heat is somewhat lower than that obtained for H-ZSM-5.15 Most often, sorption of a base weak enough not to induce full proton transfer from the acidic OH groups is examined to characterize acidity. It is generally assumed that solvation and medium effects can be disregarded and that the 1H NMR shift (∆δH),11 the red shift of the νOH band (∆νOH),9,11,16 or the shift of the UV-visible spectrum of an indicator base7,17 induced by the acid-base interaction can be correlated with the strength
10.1021/jp000937m CCC: $19.00 © 2000 American Chemical Society Published on Web 07/18/2000
7312 J. Phys. Chem. B, Vol. 104, No. 31, 2000 of the OH bond. The results obtained suggest unequivocally that H-ZSM-5 is a stronger acid than ZrO2/SO42-.9,11 Over ZrO2/SO42-, several groups of acid sites could be distinguished by their acid strengths. The preparations studied and the pretreatment conditions applied in the various studies were usually different; however, this does not seem sufficient to explain the contradictory results obtained. With the use of weak-base probe molecules, one kind of Bro¨nsted acid site could be detected.18,19 However, Spielbauer et al.20 interpreted the IR spectra obtained from ammonia adsorption as being indicative of the existence of two different Bro¨nsted acid sorption sites. Similarly, different numbers were reported for the possible types of Lewis acidity ranging from 1 to 4.9,14,18-21 In the present study, ammonia adsorption over ZrO2/SO42and H-ZSM-5 was examined by conventional methods, i.e., TPD and IR spectroscopy, and by the frequency response (FR) method. By the FR method, parallel sorption processes over distinctly different acid sites can be distinguished from their dynamic parameters.22 Results suggest that H-zeolites are stronger Bro¨nsted acids than sulfated zirconia primarily due to the more favorable stabilization of the protonated base within the zeolitic framework than over the zirconia surface. Methods disregarding such medium effects are of minor practical value in characterizing solid acids. Experimental Section Materials. A detailed characterization of the samples was given in earlier papers.13,22 The H-form of ZSM-5 was prepared by in situ thermal decomposition of the NH4+-form. The unit cell composition of the sample was (NH4)4.6Na0.03Al5.8Si90.2O192; i.e., about 80% of the Al atoms were in the zeolite framework. The sulfated oxides were prepared by the hydrolysis of Zr(npropoxide)4 and an equimolar Zr(n-propoxide)4/Ti(isopropoxide)4 mixture, drying the precipitate at 383 K, contacting the dry gel by 0.25 M H2SO4 solution, drying again, and calcining in air at 823 K. The specific surface areas of the ZrO2/ SO42- and ZrO2-TiO2/SO42- samples were 73 and 258 m2 g-1, respectively. The sulfate content of each sample was about 10 wt %. The surface areas of corresponding sulfate-free ZrO2 and ZrO2-TiO2 samples, prepared from the gel by a 3-h thermal treatment at 823 K, were 16 and 77 m2 g-1, respectively. The sulfated preparations showed structural and catalytic properties corresponding to those reported for the strong solid acid ZrO2/ SO42-.13 The sorbate ammonia gas was of 99.96% purity, and it was further purified by the freeze-pump-thaw method. Temperature-Programmed Desorption of Ammonia (NH3TPD). The NH3-TPD measurements were carried out using a conventional flow-through reactor. The reactor was attached to a vacuum system, which permitted us to evacuate the reactor and to contact the sample with ammonia gas. About 600 mg of sample was activated with flowing O2 at 773 K for 1 h, the reactor was then evacuated for 30 min, and the resulting sample was cooled to room temperature and equilibrated with NH3 at 13.3 kPa of pressure. Weakly bound NH3 was removed by a 30-min evacuation at 423 K. A helium flow of 20 cm3 min-1 was passed through the reactor, a dry ice trap, and a thermal conductivity detector. The reactor temperature was ramped up to 973 K at a rate of 10 K min-1. The detector signal was recorded, and the data were processed using a computer. IR Measurements. A self-supporting wafer of about 10 mg cm-2 thickness was prepared and placed in a heatable sample holder between the CaF2 windows of a metal IR cell. The wafer was treated in situ with flowing oxygen at 773 K for 1 h, after which the cell was evacuated for 30 min. The wafer was then
Barthos et al. cooled to room temperature and contacted with ammonia at 133 Pa of pressure. The cell containing the NH3 gas was closed, and single-beam spectra were recorded at 298, 373, 473, 573, and 673 K using a Nicolet 5PC FTIR spectrometer; 256 scans were averaged at a nominal resolution of 4 cm-1. To obtain the spectrum of the wafer with the adsorbed species, the ratio of the spectral intensities of each single-beam spectrum and the corresponding background spectrum, recorded without the sample in place, was computed. The overlapping peaks were resolved using a curve-fitting computer program. FR Measurements. The principles and technical details of FR measurements were reviewed previously.23 Fifty or hundred milligrams of finely powdered sample distributed in a ball of glass wool was placed in the FR chamber. A pressure wave was generated by applying a periodic square-wave perturbation to the volume of the FR chamber containing ammonia gas and the solid sample in sorption equilibrium at a pressure of 133 Pa. The maximum perturbation was (1% of the mean volume. The perturbation frequency was alternated between 0.01 and 10 Hz. The higher odd harmonics of the Fourier transformation of the response signals were used to obtain data up to about 90 Hz. Measurements were carried out with and without a sorbent under the same conditions. The phase difference and the amplitude ratio of the pressure waves were determined, and a response function was derived. The in-phase (real) and out-ofphase (imaginary) components of the response function were plotted against the perturbation frequency to generate the FR spectrum. The intensity of the FR response is a function of the pressure change effected by the volume perturbation and is proportional to the slope of the adsorption isotherm at the equilibrium pressure. The magnitudes of the out-of-phase peaks reflect the fractional sorption capacities associated with the corresponding mass transports. Resonance occurs when the characteristic time of a rate-controlling transport process is comparable with that of the periodic perturbation. Spectra show steps in the in-phase and peaks in the out-of-phase functions at the frequency of resonance. The theoretical FR function derived by Yasuda22c for rate-determining parallel sorption processes was used to fit the experimental FR spectra. The effect of temperature was studied in the 373-673 K range. Results NH3-TPD. ZrO2 and ZrO2-TiO2 are weak Lewis acids.1,2,6,13,21 A broad peak on the temperature-programmed NH3 desorption (NH3-TPD) curve with a maximum in the 500-700 K range characterizes the Lewis acid sites of these oxides (Figure 1, peak L′). On the TPD curves of sulfated oxides and H-ZSM-5, more distinct peaks appeared (Figure 1). The amount of ammonia corresponding to the low-temperature peak of H-ZSM-5 was roughly equivalent to the number of framework Al atoms (0.8 mmol g-1). The ammonia released at high temperature was equivalent to the total Al content (1.0 mmol g-1). Data suggest that two kinds of species were bound to the sample at the beginning of the temperature ramp-up: a protonated ammonia dimer, such as [NH4‚NH3]+, balancing the negative charge on the framework, and NH3 bound to extraframework Al (EFAl) species, often referred to as “true” Lewis acid sites. The dimeric species released ammonia in two steps. Its conceivable that the peak appearing at about 630 K was mainly due to desorption of ammonia from Lewis acid NH4+ ions ([NH4‚NH3]+ f NH4+ + NH3v; see Figure 1, peak Lb). The NH4+ ions probably decompose to ammonia and a protonic acid site at significantly higher temperatures (NH4+ f NH3 + H+). This step is paralleled by desorption of ammonia
Acidity Study of H-ZSM-5 and Sulfated Zirconia
Figure 1. NH3-TPD curves for H-ZSM-5, ZrO2-TiO2/SO42-, and ZrO2/SO42- (thick solid lines) and for the corresponding nonsulfated oxides (thin solid lines). The dashed line gives the rate of N2 evolution during the temperature-programmed desorption of ammonia from ZrO2-TiO2/SO42-. About 600 mg of sample was pretreated at 773 K with an O2 flow for 1 h. The reactor was evacuated for 30 min at 773 K, the sample was equilibrated with NH3 (13.3 kPa, 298 K), and the reactor was then evacuated at 373 K for 30 min. The TPD curve was recorded while the sample was heated to 923 K at a rate of 10 K min-1 in a 20 cm3 min-1 He flow. The effluent gas was passed through a dry ice trap before it entered the TC detector. To detect N2 formation, the released NH3 was also retained in a trap cooled by liquid N2. h ) 0.5, except for H-ZSM-5, where h ) 1. The peak of Bro¨nsted-bound ammonia is indicated by B. L and L′ designate the peaks of ammonia bound to different Lewis acid sites of the solid. Lb and Ll indicate peaks assigned to desorption of ammonia H-bound to the ammonia adsorbed on the strong-acid Bro¨nsted (B) and Lewis (L) sites of the solid.
coordinated to the strong Lewis acid sites (EFAl species). The last two processes give a single high-temperature peak at about 860 K (see Figure 1, peak B-L). ZrO2/SO42- and ZrO2-TiO2/SO42- have different specific surface areas, but their acidic and catalytic properties are quite similar.13,24 The sulfate groups generate strong Lewis and Bro¨nsted acidity in both solids. The strongest acid sites release NH3 at temperatures as high as about 900 K. These sites were shown to be of Lewis acid nature (see peak L in Figure 1).13 At about 900 K, N2 also appeared in the effluent, indicating that some of the ammonia was oxidized by the sulfate groups (Figure 1, dotted line). The broad peak at temperatures below 900 K was treated as an envelope consisting of overlapping component peaks. On the TPD curve of the high surface area ZrO2-TiO2/SO42-, the components can be clearly distinguished (Figure 1). The IR and FR results (vide infra) suggest that ammonia was released from weak Lewis acid sites, such as NH4+ ions and coordinately unsaturated zirconium and titanium ions (Figure 1, peaks Lb and L′). Peak Ll is tentatively assigned to desorption of ammonia coordinated to the NH3 strongly bound to the strong Lewis acid sites. In about the same temperature range, ammonia was also obtained from the decomposition of the NH4+ ions (Figure 1, peak B). In the latter process, the NH3
J. Phys. Chem. B, Vol. 104, No. 31, 2000 7313 release attained its maximum rate at somewhat higher temperature than desorption from sites Lb and Ll. IR Spectroscopy. The IR spectra of the samples are discussed here primarily with regard to the spectral region shown in Figure 2. For sulfated oxides the νSdO band of the sulfate groups appeared at about 1380 cm-1. Zeolite H-ZSM-5 had no bands in the shown frequency range (Figure 2A, dashed line). The adsorption of ammonia generated a broad δNH4+ envelope in the 1350-1550-cm-1 range, while the νSdO band shifted to lower wavenumbers out of the range displayed. The red shift of the νSdO band on adsorption of different basic molecules has already been reported.13,19 According to the explanation proposed, adsorption of a base increases the electron density on the surface and, thereby, decreases the bond order of the SdO bond. As a result, the νSdO band shifts to lower wavenumbers. When the temperature was increased, the equilibrium ammonia coverage decreased and the change was gradually reversed, as indicated by the reappearance of the νSdO band (Figure 2B). The sulfate band of ZrO2-TiO2/SO42shifted on adsorption of NH3 as a whole, while the sulfate band of ZrO2/SO42- split into two: to a shifted and to a nonshifted band (Figure 2A). This finding suggests that some of the sulfate groups, probably embedded in the bulk of the ZrO2/SO42sample, remained unperturbed by the surface-bound ammonia. Component bands could be resolved from the δNH4+ envelope by computer curve fitting. At 673 K, i.e., at the highest temperature of the measurement and at the corresponding lowest equilibrium NH3 coverage, the resolved bands appeared at about 1400 and 1450 cm-1 for H-ZSM-5 and at about 1430 cm-1 for the sulfated oxides. At lower temperatures, the NH3 coverages of the samples and also the frequencies of these bands slightly increased and an additional band appeared for each sample in the 1480-1505-cm-1 range (Figure 2). A simplified but descriptive band assignment applying the group frequency concept is provided. It is assumed here that frequencies of the resolved bands are characteristic of N-H bending vibrations affected by H-bonds of different strengths. Each of the bands mentioned is assigned to δNH vibrations of surface-bound NH4+ ions, which are unperturbed or perturbed by adsorbed NH3 within protonated associations, such as [NH4‚nNH3]+, where n ) 1, 2, 3, etc. Under the isobaric conditions of the measurements, the band at about 1480-1500 cm-1 was found to vanish rapidly, as coverage became lower at higher temperatures (Figures 2 and 3). This band was assigned to the vibration of an N-H group H-bonding to another nitrogen atom. It is known that association of molecules decreases the vibrational frequency of the N-H stretching mode and increases that of the bending mode. The 1480-1500-cm-1 band was the highest frequency NH4+ bending peak resolved from the δNH4+ envelope. The νNH region of the spectrum obtained from adsorption of ammonia over ZrO2-TiO2/SO42- is shown in Figure 4. The band at 3053 cm-1 is the lowest frequency νNH band. The intensity of this band and that at 1480 cm-1 decreased similarly and rapidly when the temperature was increased, suggesting that the bands belong to the same species, most probably to N-H groups of NH4+ ions affected by H-bound NH3. At 673 K, in the absence of coordinated ammonia, the spectrum shows bands of NH4+ bending vibrations perturbed by H-bonding interactions with surface oxide ions. Thus, it can be suggested that the band at about 1400 cm-1 stems from N-H groups coordinated to weak-base oxide ions of the ZSM-5 zeolite (tN-H‚‚‚O-Zeol). The weakest base oxide ions are the conjugated base sites of the strongest Bro¨nsted acid sites. The rest of the framework and nonframework oxide ions are more
7314 J. Phys. Chem. B, Vol. 104, No. 31, 2000
Barthos et al.
Figure 2. IR spectra obtained from adsorption of NH3 on (A) H-ZSM-5, ZrO2-TiO2/SO42-, ZrO2/SO42-, and ZrO2 at 298 K and (B) ZrO2-TiO2/ SO42- at 373, 473, 573, and 673 K (thick solid lines). Spectra were recorded with the samples in adsorption equilibrium with NH3 under about 133 Pa of pressure. Under each spectrum are shown the best-fit component bands (thin solid lines). The dashed lines show the spectra of the wafers before NH3 adsorption. Self-supporting wafers were treated in situ with an O2 flow for 1 h, the reactor was evacuated for 30 min at 773 K, and the wafers were then contacted with NH3 gas at 298 K. Absorbance is normalized to 10 mg cm-2 wafer thickness.
basic; therefore, their H-bonding to N-H groups must be stronger. The band at about 1450 cm-1 could stem from vibrations of N-H groups coordinated to these oxygen atoms. Recently, a similar interpretation was given by Fripiat et al.25 for the spectra of zeolite-bound NH4+ ions. The appearance of a single band at 1430 cm-1 suggests that sulfated oxides comprise oxide ions of base strength that is about intermediate between that of the two kinds of oxide ions in the H-ZSM-5 sample (see Figure 2A,B). Bands attributed to bending modes of surface-bound NH3 and NH4+ were also detected in the 1600-1630-cm-1 region and at about 1680 cm-1, respectively. These bands were weak; therefore, they were found to be less useful for characterizing the surface. Spectra below about 1300 cm-1 were not informative due to the strong absorption of the solid. Pure zirconia was an exception; together with the band of the asymmetric bending vibration of adsorbed NH3 at about 1600 cm-1, bands of symmetric bending were also observed at 1125 and 1214 cm-1 (not shown). The TPD, IR, and FR results (vide infra) substantiate that, under 133 Pa of ammonia up to at least 573 K, almost all Bro¨nsted protons of the ZSM-5 sample remain captured by NH3, forming NH4+ ions. If the NH3 coverage of the NH4+ sites decreases, the concentration of the N-H groups not affected by coordinated NH3 can be expected to increase. However, the integrated absorbances of the NH4+ bands at about 1450 and 1400 cm-1 were found to be virtually unaffected by temperature in the 298-673 K range (Figure 3). It is known that, with a decreasing extent of association, the absorption coefficient of the N-H groups becomes smaller. It is conceivable that a
species with a smaller absorption coefficient was formed in higher concentration as the NH3 coverage decreased at higher temperatures. This may explain why, as a net result, the integrated areas of the bands near 1400 and 1450 cm-1 remained virtually unchanged in the examined temperature range. For the sulfated oxides, the integrated δNH absorbance vs temperature plots show pronounced differences from those obtained for H-ZSM-5 (Figure 3). The intensity of the 1480cm-1 band monotonically decreases with decreasing NH3 coverage. In contrast, the intensity of the peak at about 1430 cm-1 increases up to about 373 K and then, at higher temperatures, gradually decreases. The concentration of the NH3associated N-H species decreases, and at low temperatures, the concentration of the nonassociated or oxide-ion-associated N-H groups increases. The concentration increase is not fully balanced by the decrease of the absorption coefficient, as indicated by the growing intensity of the 1430-cm-1 band up to about 400 K. The intensity loss at higher temperatures is attributed to conversion of NH4+ to NH3. The 1430-cm-1 band does not disappear up to 673 K, showing that some protonated ammonia was retained even at this temperature. FR Spectroscopy. The “rate spectra” of NH3 sorption are shown in Figure 5 for the studied NH3/sorbent systems. The symbols and lines give the experimental data and the best-fit characteristic FR curves, respectively. Dotted lines represent the characteristic FR functions resolved for parallel sorption processes. It should be understood that sorption sites can be examined by the FR method if perturbation of the system modulates the coverage of the sites, i.e., if the slope of the adsorption isotherm is not zero at the pressure of the measure-
Acidity Study of H-ZSM-5 and Sulfated Zirconia
J. Phys. Chem. B, Vol. 104, No. 31, 2000 7315
Figure 4. The νNH region of the spectra shown in Figure 2B.
Figure 3. IR band intensity of adsorbed ammonia vs temperature plots at equilibrium NH3 pressure of about 133 Pa. The peak areas were obtained by resolving and integrating δNH bands in the 1300-1600cm-1 wavenumber region shown in Figure 2.
ment. A comparison of panels D and d in Figure 5 suggests that adsorption sites, which are fully covered at 573 K, become activated for FR examination at 673 K. The time constant of the sorption process and the desorption rate constant are related quantities.22b Thus, at 673 K, at least two sorption processes of different desorption activation energies can be distinguished, suggesting that H-ZSM-5 contains at least two kinds of sorption sites for ammonia. The high-frequency resonance can be related to a sorption process characterized by a weak sorption interaction. It is suggested that this resonance is predominantly due to ammonia sorption over Lewis acid NH4+ ions (Figure 5D, peak Lb). The sorption equilibrium of this species only is perturbed
at 573 K (Figure 5d). At 673 K, the ammonia coverage of the strong Bro¨nsted and Lewis acid sites is also affected, resulting in the appearance of an additional low-frequency peak at about 0.2 Hz (Figure 5D, peak B-L). The resonance peaks resolved from the rate spectrum of ammonia sorption over zirconia are attributed to sorption over weak Lewis acid sites, conceivably over coordinately unsaturated surface zirconium atoms (Figure 5A, peak L′). On the rate spectrum of ZrO2/SO42-, an additional resonance is observed at 0.15 Hz, indicating a slower sorption process (Figure 5B, peak B). It is known that surface-bound sulfate ions generate Bro¨nsted and Lewis acidities. The generated strong Lewis acid sites, according to the TPD results shown in Figure 1 (peak L), start to release NH3 above about 800 K, while some NH3 is converted first to N2. At 673 K under 133 Pa of NH3 pressure, all these sites must be fully covered by ammonia. Therefore, detection of sorption processes over these sites by the FR examination could not be expected. Most probably, the resonance at 0.15 Hz is due to sorption over sulfate-induced Bro¨nsted acid sites. Again, peak Ll is tentatively assigned to the adsorption-desorption process occurring in the coordination sphere of the NH3 species strongly bound to Lewis acid surface sites (cf. Figures 1 and 5). In general, the FR spectrum of ZrO2-TiO2/SO42- corresponds to that obtained for ZrO2/SO42-. Comparison of the spectral intensities shows that larger ammonia transport was effected by the same pressure modulation above the mixed oxide of higher surface area and sorption capacity (Figure 5). At a given equilibrium pressure and volume (pressure) modulation, the FR intensity of a sorption system depends on the temperature. For the applied range of modulation frequency, the total FR intensity (I) is shown as a function of the temperature (T) in Figure 6. The I vs T plot of the NH3/HZSM-5 system exhibits the characteristic maximum and minimum at about 450 and 570 K, respectively. In contrast, the FR intensity of the NH3 sorption over the zirconia-containing samples remains almost unchanged in the examined temperature range. For species obeying Langmuir type sorption kinetics, it was recently shown that, at constant pressure, the total FR intensity passes through a maximum as a function of temperature.22b The maximum has to appear at a temperature where 50% of the sorption sites are covered by the species exhibiting the FR signal. The two maxima on the FR intensity vs
7316 J. Phys. Chem. B, Vol. 104, No. 31, 2000
Barthos et al.
Figure 5. FR spectra of ammonia sorption over (A) ZrO2, (B) ZrO2/SO42-, (C) ZrO2-TiO2/SO42-, and (D, d) H-ZSM-5 at (A-D) 673 K and (d) 573 K. 100 and 50 mg samples of sulfated oxide and NH4-ZSM-5, respectively, were pretreated in situ by 1-h evacuation at 773 K. The samples were then contacted with NH3 at 133 Pa of pressure at the temperature of the FR measurements. The same designations were applied for the out-of-phase component peaks (B, L′, Lb, Ll) as were used in Figure 1.
Figure 6. FR intensity vs temperature plots for sorption of NH3 over (A) ZrO2, (B) ZrO2/SO42-, (C) ZrO2-TiO2/SO42-, and (D) H-ZSM-5. For measurement conditions, see the caption to Figure 5.
temperature plot of the H-form zeolite suggest that energetically distinctly different sorption sites are active at lower and at higher temperatures. At low temperatures, most probably, the NH3 coverage of the NH4+ ions (0.5 < ΘNH4+ < 1) was effected by the applied modulation. At higher temperatures, ΘNH4+ decreased below 0.5 and the Bro¨nsted-bound ammonia became activated for adsorption-desorption (ΘH+ < 1) (Figure 6). The almost temperature-independent FR intensity obtained for the sulfated oxides suggests that the surface of the oxides is energetically heterogeneous; i.e., about equal fractions of the surface sites are activated for adsorption-desorption at each temperature of the measurement. Discussion Upon adsorption of ammonia on sulfated zirconia or the zeolite H-ZSM-5, proton transfer was observed. The νOH band was found to disappear from the vibrational spectra of the solids, whereas new bands stemming from NH4+ ions appeared (Figures 2 and 4). The spectrum of ammonia coordinated to Lewis acid sites was also observed. An important conclusion of this work is that a significant fraction of the adsorbed ammonia molecules coordinate NH4+ ions even at relatively high temperatures. As was mentioned previously, different studies distinguished different kinds of Lewis and Bro¨nsted acid centers according to their acid strengths on the surfaces of sulfated zirconia. Our
results (Figures 1 and 5) suggest that two kinds of Lewis sites (L and L′) coexist on the surfaces of our preparations. The strongly acidic sorption sites (L) were generated by the sulfate. Some of the weaker sites, similar to those on the nonsulfated sample, may have been created as a result of adsorption, which can remove some covalently bound sulfate ions. Ionic surface species, such as NH4+ and SO42-, were obtained, as well as uncovered Zr4+ ions. These Zr4+ ions are weak Lewis acid sites (L′), as suggested by Babou et al.21 No distinctly different Bro¨nsted acid sites could be resolved; rather we observed a group of sites characterized by a broad acid strength distribution (Figures 1 and 5, peak B). The broad distribution is probably due to stabilization of the NH4+ ions over a large number of energetically different surface sites. Most of the NH4+ ions are stabilized in ammonia associations (vide infra). TPD and FR peaks were assigned to ammonia Lewis-bound to the NH4+ ions and to the strongly polarized ammonia on the L sites (Figures 1 and 5, Lb and Ll). Thus, all-together four different kinds of Lewis sites can be observed, but only two seem to stem from direct ammonia-surface interactions. Only the strongest Bro¨nsted acid sites were of acidity comparable to the acidity of the sites on H-ZSM-5. Over H-ZSM-5, ammonia associations were also formed; however, the ion-pair complexes were found to remain stable when all of the excess ammonia was already removed. The IR spectroscopic results also provided evidence for the formation of protonated ammonia associations over the studied solids. The δNH bands were assigned above to vibrations of N-H groups in H-bonds of different strengths, namely, in H-bonds to oxide ions or to NH3 molecules. However, an analysis regarding the possible symmetries and normal modes of the bound NH4+ species can give a better understanding of the ionpair interactions and the surface properties.26,27 The “free” NH4+ ion (not involved in H-bonds) has Td symmetry. It has a Raman-active stretching and a doubly degenerated bending mode, as well as an IR-active stretching and a bending mode, both 3-fold degenerated. The latter modes give rise to IR bands at 3310 and 1418 cm-1, respectively.28 H-bonding decreases the symmetry of the ion. As a result, all modes become IR active. The bands, depending on the sym-
Acidity Study of H-ZSM-5 and Sulfated Zirconia
J. Phys. Chem. B, Vol. 104, No. 31, 2000 7317
SCHEME 1: Energy Diagram of the Ion Pair Formation from Ammonia and a Solid Bro1 nsted Acid, Showing (A) Hypothetical Reaction Steps Permitting Theoretical Calculations and (B) Energy Changes along a Possible Reaction Coordinate
metry, split to a maximum of four νNH and five δNH4+ bands. In addition, the combinations of the frustrated rotations with the deformation modes give rise to two weak bands in the 22501700-cm-1 region.27 The stretching bands appear in a broad envelope. The assignment of the component bands is difficult and uncertain, since this part of the spectrum is greatly altered by Fermi-type resonance interactions occurring between overtones and combinations of the bending modes and some N-H stretching vibrations. In relation to the present work, the 1550-1350-cm-1 spectral region is especially significant. Depending on the symmetry of the NH4+ ion, two (C3V) or three (C2V and Cs) bending modes can give rise to two or three bands in this wavenumber region. Thus, the appearance of two or more IR bands does not necessarily indicate the presence of different kinds of Bro¨nsted acid sites. Using the harmonic approximation, ab initio calculations predicted bands at 1518, 1421, and 1324 cm-1 for NH4+ coordinated by two H-bonds to the zeolite framework (C2V symmetry).28 With H-faujasite, δNH4+ bands were detected at 1480, 1430, and 1370 cm-1.26 At low coverage, a similar triplet could be discerned in the spectra of ammonia adsorbed on H-ZSM-5, H-β, and H-SAPO-34, but only a broad band was obtained for the NH3/H-mordenite system.27 Because of Hbonding to similar or different basic oxide ions and, at higher coverages, to ammonia, the NH4+ ions can assume mono-, di-, tri-, and tetradentate configurations. It is conceivable that a broad peak occurs in the 1550-1350-cm-1 region if more NH4+ forms exist together. The peak is an envelope of overlapping doublets and triplets of bands. Even for the very same configurations, the corresponding bands of the species can have slightly different wavenumbers depending on the strengths of the interactions. In such a case, the bands of the different vibrational modes cannot be resolved. Most probably, this is the situation that was observed with H-mordenite27 and was experienced with the
systems studied in this work. The highest δNH4+ frequencies are most probably obtained when the NH4+ ions associate with strong-base NH3 molecules. Thus, it is suggested that the band resolved near 1500 cm-1 at high NH3 coverage indicates that some of the NH4+ ions are H-bonding not only to oxide ions but also to ammonia molecules (cf. Figures 2 and 3). The acid strength of a Bro¨nsted site is best characterized by thermodynamic quantities, such as the deprotonation energy (DE) of the site or heat of adsorption (∆Hads) of a base. It should be understood that DE is independent of any acid-base interaction while ∆Hads characterizes the acid site in an interaction with a base involving all the possible medium effects. Our results should be interpreted with regard to the energetics of proton transfer and the stabilization of an ion-pair complex. The adsorption-protonation of ammonia on a Bro¨nsted acid site is a facile, exothermic reaction:
AOH + NH3 f AO-NH4+
(1)
where AOH and AO- represent a surface acid group and its conjugated base, respectively. The heat of adsorption (∆Hads) can be obtained experimentally, or it can be calculated as
∆Hads ) ∆Hst + ∆Hpt
(2)
∆Hst is the stabilization energy of the ion-pair complex, and ∆Hpt is the proton-transfer energy. ∆Hpt is the sum of the deprotonation energy (DE) of the Bro¨nsted acid site and the protonation energy of ammonia in the gas phase. The latter quantities are often referred to as the negative proton affinity of the AO- ion (DE ) -PAAO-) and the proton affinity of ammonia (PANH3), respectively (Scheme 1A). The proton affinity of a zeolite (PAZO-) can be determined experimentally or predicted by ab initio quantum chemical cluster calculations.
7318 J. Phys. Chem. B, Vol. 104, No. 31, 2000 From the frequencies of fundamental transitions and overtones of hydroxyl groups, Kazansky8 calculated the energies of homolytic OH dissociation for various oxides and zeolites and corrected these values with the energies of electron transfer from the hydrogen atom to the solids to obtain the energies of heterolytic dissociation, i.e., the deprotonation energies. The DE values obtained for H-faujasite and H-mordenite were about 1200 kJ mol-1. Using IR spectroscopy of probe molecules, Datka et al.16 determined similar values for H-ZSM-5. It was shown that H-ZSM-5 contains five different OH groups, having deprotonation energies in the 1200-1320 kJ mol-1 range. There results correspond to those obtained from the 1H NMR shifts (∆δH) induced by adsorption of weak bases.11b The theoretically calculated DE values for small H-ZSM-5 clusters were between 1240 and 1650 kJ mol-1 depending on the models used.28-32 The proton affinity of ammonia in the gas phase is about 820900 kJ mol-1.15,31,33,34 As a consequence, ∆Hst must be larger than ∆Hpt by more than about 300 kJ mol-1 so that the proton transfer is energetically favored. For H-ZSM-5, reaction 1 is exothermic with about 150 kJ mol-1, implying that the zeolite lattice must stabilize the NH4+ ion with an energy of more than ∼450 kJ mol-1. Separation of the adsorption process into (i) proton transfer at an infinite distance from the sorbent and (ii) binding of the obtained “free” NH4+ ion to the zeolite permits us to calculate the enthalpy change of each step. The calculations suggest that proton transfer between an ammonia molecule and an acid site in the absence of a medium providing extra energy (∆Hst) is energetically not feasible. Proton transfer occurs along a very complicated reaction coordinate. An [AOH‚‚‚NH3] sorption complex is formed first, representing a local minimum on the energy diagram of the protonation process. The system then passes through an energy barrier, if any, before it is stabilized as an [AO-NH4+] ion-pair complex. During the process of proton transfer, the deprotonation energy of the acid site and the proton affinity of NH3 change simultaneously until PA(NH3)a e -DE(AOH)a ≡ PA(AO-)a, where (AOH)a and (NH3)a represent species of the transition state complex (Scheme 1B). Cluster calculations of Kyrlidis et al.32 suggest that ionic adsorbed species are favored over neutral adsorbed species by only about 10 kJ mol-1 for the zeolite H-ZSM-5. The proton transfer can be described as proceeding by a mechanism that permits maximum energy compensation of the bonds to be broken by those to be formed. Similar ideas were advanced by Kazansky8 to explain the process of alkoxide formation over zeolites. The formation of an alkoxide requires proton transfer between a zeolite acid and an alkene, which is a much weaker base than ammonia. The process was referred to as a “concerted mechanism” emphasizing that proton transfer is due to concerted effects of the Bro¨nsted acidic moieties and the neighboring basic oxide ions. It is well-known that the medium (solvent) affects proton affinities. For instance, NH3‚HCl complexes are H-bonded in the gas phase35 but partial proton transfer can be detected in an N2 or Ar matrix at low temperatures.36,37 The proton affinity of an ammonia cluster is also higher than that of a single ammonia molecule. With -850 kJ mol-1 as a reference value for NH3, the proton affinity of (NH3)2 was estimated to be -950 kJ mol-1 and that of a cluster of five NH3 molecules to be about -1056 kJ mol-1.34 Over H-zeolites, ammonia is converted to NH4+ at a coverage of ΘH+ ,1. Thus, it is conceivable that the proton affinity of a single ammonia molecule H-bonded to a zeolite framework is
Barthos et al. higher (larger negative value) than that of the free molecule [PA(NH3)a < PANH3]. Cluster calculations provide evidence that hydrogen bonding to ammonia can also decrease the proton affinity of a zeolite framework (DE(ZOH)a ≡ -PA(ZO-)a < DEZOH ≡ -PAZO-). It was shown that the deprotonation energy of a zeolitic T-O(H)-T (T ) Si, Al) site tends to be larger for stronger T-O bonds and smaller T-O-T angles.28-30,38 Calculations suggest that the angle of the ionic [T-O-T] fragment is larger, substantiating a lower -PAZO- (lower DE) compared to that of the fragment carrying a proton [T-O(H)-T].31,32 The hydrogen bonding of NH3 to a T-O(H)-T site slightly shifts parameters such as bond angle and bond length toward the anionic values, inferring that NH3 sorption decreases the deprotonation energy of the T-O(H)-T site. The association provides the [NH4‚nNH3]+ ion with an extra energy of stabilization exceeding that of the NH4+ ion. It follows from thermodynamic considerations34 that the extra stabilization energy is the difference in the enthalpies of the reactions giving ionic {NH4+ + (NH3)n f [NH4‚nNH3]+} and nonionic, Hbonded associations {(n + 1)NH3 f (NH3)n+1}, or in other terms, the energy gain of the system is PA(NH3)n+1 - PANH3. When ammonia is released from the ionic association, part of this stabilization energy is lost. If the coordination of the NH4+ ion to the solid is not sufficient to stabilize the system, the separation of charges becomes energetically less favored and the proton is transferred from the NH4+ cation (from the acid) to an oxide ion of the solid (to the conjugated base of the Bro¨nsted acid). This was observed with the sulfated zirconia or zirconia-titania. The decrease of the ammonia coverage was found to be paralleled by the decomposition of the NH4+ ions (cf. Figures 1 and 3). In contrast, all the weakly bound ammonia could be removed without affecting the decomposition of the NH4+ ions in the zeolite H-ZSM-5. This implies that the NH4+ ions are more effectively stabilized within the micropores of the H-ZSM-5 framework than over the surface of ZrO2/SO42or ZrO2-TiO2/SO42-. Neither calculated nor measured values are available for the deprotonation energy of ZrO2/SO42-. It was presumed, however, that the correlation between the 1H NMR chemical shift (δH) and the DE would allow at least some qualitative conclusions.11,39,40 For ZrO2/SO42-, the value of δH was found to be significantly higher than that for ZrO2 or H-ZSM-5, suggesting that some sites of sulfated zirconia have lower DE values than the sites of H-ZSM-5.10,11 It was concluded, therefore, that ZrO2/ SO42- is a stronger acid than H-ZSM-5, probably a superacid.10 In contrast, ZrO2/SO42- was always found to be a weaker acid than zeolite H-ZSM-5 when sites were probed using a base molecule.7,9,11 It was also found to be less active in catalytic reactions requiring strong Bro¨nsted acid active sites, such as alkane isomerization or cracking.13 Whatever the actual value of the DE, it is certainly much higher than -PANH3. The present work has shown that, for the NH4+ ions, ∆Hst of ZrO2/SO42is smaller than that of H-ZSM-5. Thus, acidity, which always appears in the interaction with a base, is decisively determined by the stabilization (medium) effect of the solid. The medium can also modify the proton affinity of the adsorbed weak bases, such as hydrocarbons and CO, and, thereby, the strength of H-bonding to the Bro¨nsted sites even if full proton transfer is not induced. H-bonding to a base weakens the O-H bond. The weakening of the bond can be followed by various spectroscopic techniques. The measured spectral parameters ∆δH+ and ∆νOH correlate with the DE of the surface hydroxyl group only if the stabilizations of the base interacting with different OH groups
Acidity Study of H-ZSM-5 and Sulfated Zirconia of a solid or with hydroxyl groups of different solids are similar. This condition is not satisfied when acid sites are probed with different bases or when very different solid acids, such as ZrO2/ SO42- and H-ZSM-5, are probed with a base. The inequality of the medium effects provides an explanation for the controversial results obtained with and without a basic probe. The adsorption of ammonia increases the electron density on the solid acids. This is partly due to the electron donation of the Lewis-bond base. The negative charge on the solid is also increased when covalent O-H bonds are broken and stable ion pairs are formed.7,13 The increasing electron density on the solid is reflected by the red shift of the νSdO band. This shift indicates decreased bond order, i.e., weaker SdO bonds. Depending on the structure of the solid acid, the electrons are delocalized to some extent over the conjugated base of the acid. The delocalization of the negative charge increases the stability of the ion pair complex. This also means that the strength of the unconverted O-H bonds and, therefore, the DE of the Bro¨nsted acid sites increase as ammonia coverage is increased. It follows from Scheme 1A that the increasing coverage results in decreasing ∆Hads if ∆Hst is constant. In this connection, it should be noted that the ∆Hads values measured for H-ZSM-5 were often found to be almost invariant with coverage up to ΘH+ ≈1. This finding suggests that the increase in DE can be paralleled with increasing ∆Hst, i.e., with a more favorable stabilization of the NH4+ ions interacting with more basic oxide ions. Conclusion Ion-pair interactions with a ZSM-5 framework were found to provide sufficient energy to stabilize NH4+ ions. In contrast, the associated ammonia contributes significantly to stabilizing the charge separation over sulfated zirconia. Data suggest that, in the absence of medium effects, proton transfer is not energetically favored from these solid acids, even to a strong base like ammonia. The deprotonation energy of sulfated zirconia is probably lower than or comparable to that of H-ZSM5. However, H-ZSM-5 is a stronger acid due to the preferable stabilization of the adsorbed or protonated bases within the zeolitic framework. Acknowledgment. We are grateful to Professor L. V. C. Rees, Department of Chemistry, University of Edinburgh, Edinburgh, U.K., for the opportunity to collect FR data in his laboratory. Thanks are also expressed to the Hungarian Scientific Research Fund (OTKA, Contract No. T 029717) and to the Royal Society (CEE Project Grant) for financial support. References and Notes (1) Yadav, G. D.; Nair, J. J. Microporous Mesoporous Mater. 1999, 33, 1. (2) Song, X.; Sayari, A. Catal. ReV.sSci. Eng. 1996, 38, 329. (3) Fogash, K. B.; Yaluris, G.; Gonzalez, M. R.; Ouraipryvan, P.; Ward, D. A.; Ko, E. I.; Dumesic, J. A. Catal. Lett. 1995, 32, 241. (4) Cheung, T. J. Catal. 1995, 153, 344. (5) Arata, K. AdV. Catal. 1990, 37, 165. (6) Sohn, J. R.; Kim, H. W. J. Mol. Catal. 1989, 52, 361. (7) Umansky, B.; Engelhardt, J.; Hall, W. K. J. Catal. 1991, 127, 128. (8) (a) Kazansky, V. B. In Acidity and Basicity of Solids: Theory, Assessment and Utility; Fraissard, J., Petrakis, L., Eds.; Kluwer Academic
J. Phys. Chem. B, Vol. 104, No. 31, 2000 7319 Publishers: London, 1994; p 335. (b) Kazansky, V. B. Acc. Chem. Res. 1991, 24, 379. (9) (a) Kustov, L. M.; Kazansky, V. B.; Figueras, F.; Tichit; D. J. Catal. 1994, 150, 143. (b) Armendariz, H.; Sanchez Sierra, C.; Figueras, F.; Coq, B.; Mirodatos, C.; Lefebvre, F.; Tichit, D. J. Catal. 1997, 171, 85. (10) (a) Pfeifer, H. NMR Basic Principles and Progress; Springer: Berlin, 1994; Vol. 31, p 31. (b) Riemer, T.; Spielbauer, D.; Hunger, M.; Mekhemer, G. A. H.; Kno¨zinger, H. J. Chem. Soc., Chem. Commun. 1994, 1181. (11) (a) Brunner, E.; Ka¨rger, J.; Koch, M.; Pfeifer, H.; Sachsenro¨der, H.; Staudte, B. In Progress in Zeolite and Microporous Materials; Chon, H., Ihm, S.-K., Uh, Y. S., Eds.; Studies in Surface Science and Catalysis, Vol. 105; Elsevier: Amsterdam, 1997; p 463. (b) Adeeva, V.; de Haan, J. W.; Ja¨nchen, J.; Lei, G. D.; Schu¨nemann, V.; van de Ven, L. J. M.; Sachtler, W. M. H.; van Santen, R. A. J. Catal. 1995, 151, 364. (12) Babou, F.; Bigot, B.; Sautet, P. J. Phys. Chem. 1993, 97, 11501. (13) Lo´nyi, F.; Valyon, J.; Engelhardt, J.; Mizukami, F. J. Catal. 1996, 160, 279. (14) Yaluris, G.; Larson, R. B.; Kobe, J. M.; Gonza´lez, M. R.; Fogash, K. B.; Dumesic, J. A. J. Catal. 1996, 158, 336. (15) (a) Parillo, D. J.; Gorte, R. J.; Farneth, W. E. J. Am. Chem. Soc. 1993, 115, 12441. (b) Parillo, D. J.; Biaglo, A.; Gorte, R. J.; White, D. In Zeolites and Related Materials: State of Art; Weitkamp, J., Karge, H. G., Pfeifer, H., Ho¨lderich, W., Eds.; Elsevier: Amsterdam, 1994; p 701. (c) Shannon, R. D.; Staley, R. H.; Vega, A. J.; Fischer, R. X.; Baur, W. H.; Auroux, A. J. Phys. Chem. 1989, 93, 2019. (d) Shi, Z. C.; Auroux, A.; Taarit, Y. B. Can. J. Chem. 1988, 66, 1013. (16) Datka, J.; Boczar, M.; Rymarowicz, P. J. Catal. 1988, 114, 368. (17) Umansky, B.; Hall, W. K. J. Catal. 1990, 124, 97. (18) Lunsford, J. H.; Sang, H.; Campbell, S. M.; Liang, C.-H.; Anthony, R. G. Catal. Lett. 1994, 27, 305. (19) Platero, E. E.; Mentruit, M. P.; Area´n, C. O.; Zecchina, A. J. Catal. 1996, 162, 268. (20) Spielbauer, D.; Mekhemer, G. A. H.; Zali, M. I.; Kno¨zinger, H. Catal. Lett. 1996, 40, 71. (21) Babou, F.; Coudurier, G.; Vedrine, J. C. J. Catal. 1995, 152, 341. (22) (a) Valyon, J.; Onyestya´k, Gy.; Rees, L. V. C. J. Phys. Chem. B 1998, 102, 8994. (b) Valyon, J.; Onyestya´k, Gy.; Rees, L. V. C. Langmuir 2000, 16, 1331. (c) Yasuda, Y. Heterog. Chem. ReV. 1994, 1, 103. (23) Rees, L. V. C.; Shen, D. Gas Sep. Purif. 1993, 7, 83. (24) Barthos, R.; Lo´nyi, F.; Engelhardt, J.; Valyon, J. Top. Catal. 2000, 10, 79. (25) Yin, F.; Blumenfeld, A. L.; Gruver, V.; Fripiat, J. J. J. Phys. Chem. B 1997, 101, 1824. (26) Earl, W. L.; Fritz, P. O.; Gibson, A. A. V.; Lunsford, J. H. J. Phys. Chem. 1987, 91, 2091. (27) Zecchina, A.; Marchese, L.; Bordiga, S.; Paze`, C.; Gianotti, E. J. Phys. Chem. B 1997, 101, 10128. (28) Sauer, J.; Ugliengo, P.; Garrone, E.; Saunders, V. R. Chem. ReV. 1994, 94, 2095. (29) Brand, H. V.; Curtiss, L. A.; Iton, L. E. J. Phys. Chem. 1992, 96, 7725. (30) Teunissen, E. H.; van Duijneveldt, F. B.; van Santen, R. A. J. Phys. Chem. 1992, 96, 366. (31) Teunissen, E. H.; van Santen, R. A.; Jansen, A. P. J.; van Duijneveldt, F. B. J. Phys. Chem. 1993, 97, 203. (32) Kyrlidis, A.; Cook, S. J.; Chakraborty, A. K.; Bell, A. T.; Theodoru, D. N. J. Phys. Chem. 1995, 99, 1505. (33) Bartmess, J. E.; McIver, R. T. In Gas Phase Ion Chemistry; Bowers, M. T., Ed.; Academic: New York, 1979; Vol. 2, Chapter 11. (34) Chesnovsky, O.; Leutwyler, S. J. Chem. Phys. 1988, 88, 4127. (35) Brciz, A.; Karpfen, A.; Lischka, H.; Schuster, P. Chem. Phys. 1984, 89, 337. (36) (a) Ault, B. S.; Pimentel, G. C. J. Chem. Phys. 1973, 77, 1649. (b) Ault, B. S.; Steinback, E.; Pimentel, G. C. J. Phys. Chem. 1975, 79, 615. (37) (a) Barnes, A. J.; Beech, T. R.; Mielke, Z. J. Chem. Soc., Faraday Trans. 2 1984, 80, 455. (b) Barnes, A. J.; Kuzniorski, J. N. S.; Mielke, Z. J. Chem. Soc., Faraday Trans. 2 1984, 80, 465. (38) Rabo, J. A.; Gajda, G. J. Catal. ReV.sSci. Eng. 1989-90, 31, 385. (39) Freude, D.; Hunger, M.; Pfeifer, H. Z. Phys. Chem. (Munich) 1987, 152, 171. (40) Haw, J. F.; Hall, M. B.; Alvarado-Swaisgood, A. E.; Munson, E. J.; Lin, Z.; Beck, L. W.; Howard, T. J. Am. Chem. Soc. 1994, 116, 7308.