Temperature Dependence of the Raman Spectrum of Barium Peroxide

The Raman spectra of barium peroxide (BaO2) in the O-O stretching region were studied at various temperatures between 22 and 850 °C in both O2 and He...
1 downloads 0 Views 325KB Size
J. Phys. Chem. 1996, 100, 551-555

551

Temperature Dependence of the Raman Spectrum of Barium Peroxide Kristjan Haller,† Jack H. Lunsford, and Jaan Laane* Department of Chemistry, Texas A&M UniVersity, College Station, Texas 77843-3255 ReceiVed: August 21, 1995X

The Raman spectra of barium peroxide (BaO2) in the O-O stretching region were studied at various temperatures between 22 and 850 °C in both O2 and He atmospheres. Although only one peroxide stretching band is expected for the crystal symmetry, three Raman bands, which were assigned to peroxide ions in different crystal environments due to the relative locations of oxide impurities, were observed at all temperatures. The temperature dependencies of the Raman frequencies, bandwidths, and the absolute and partial intensities were investigated in detail, and these were explained on the basis of the sample composition and structural changes.

Introduction The physical and chemical properties of barium peroxide have been studied for more than a century, and the preparation of pure oxygen from its decomposition has been of special interest. During the past decade there has been a growing interest in peroxides including BaO2 for their possible role as catalysts in oxidation processes allowing the conversion of abundant hydrocarbon resources to more useful chemicals and fuels.1,2 Recently, we investigated the Raman spectra of the O-O stretching vibration of BaO2 in order to study the effect of temperature, the influence of CO2, H2O, or He atmosphere, and the effect of BaO2 concentration on Ba/MgO catalysts.3 At 100 °C we found the barium peroxide Raman spectrum to be characterized by a major peak at 842 cm-1 with weaker shoulders at 829 and 821 cm-1. Raman measurements of isotopically substituted (18O) samples verified that all three components were related to the O-O vibration, but the origin of this unexpected multiplet structure was unclear. The vibrational spectrum of BaO2 has been investigated before,4 but its multiplet structure was not reported. In the present work we have undertaken detailed Raman studies of BaO2 in the O-O stretching region at different temperatures from 22 to 850 °C. Samples were studied under both O2 and He atmospheres, and different preparation schemes involving various temperatures were used in order to understand the effects of these factors. Experimental Section Two sources of BaO2 were used in this work: a commercial sample from Aldrich and a sample prepared using the sample cell described earlier.3 In the latter case, pure powdered BaO was pressed into the sample holder where it was kept under an atmosphere of high-purity oxygen, free from both H2O and CO2. On the final step of the purification, the sample was maintained at 800 °C for 4-8 h in order to eliminate the carbonate present, before it was cooled to the temperature at which the spectra were recorded. The same sample cell was also used for collecting the Raman spectra (90° scattering) at various temperatures. A Jobin Yvon U-1000 double monochromator equipped with holographic gratings (1800 grooves/mm) and a Coherent Innova 20 argon ion laser excitation source operating at 514.5 nm were used. A RCA C31034-02 photomultiplier tube and photon counting were utilized. For some spectra a Coherent DPSS 532-200 diode† Permanent address: Institute of Physics, Estonian Acad. Sci., Riia 142, EE2400 Tartu, Estonia. X Abstract published in AdVance ACS Abstracts, December 15, 1995.

0022-3654/96/20100-0551$12.00/0

pumped, solid-state laser operating at 532 nm and/or a Spectraview 2D liquid-nitrogen-cooled charge coupled device (CCD) from Instruments SA were used. The excitation power at the sample was maintained at a constant value ((5%) in the 150400 mW range. The Raman spectra were measured with a spectral resolution better than 0.9 cm-1 and an absolute accuracy of (0.5 cm-1. The reproducibility of both Raman frequencies and bandwidths was better than 0.2 cm-1. LabCalc software was used for curve fitting of the experimental line shapes and for calculating spectral parameters. Normal Vibrations of BaO2 The elementary cell of BaO2 has tetragonal symmetry, I4/ 5 mmm ≡ D17 4h, with two structural units. The primitive cell for the BaO2 lattice is its Wigner-Seitz cell of D4h symmetry with one molecular unit. A group theoretical analysis of the normal modes of crystalline BaO2 shows that they have the following irreducible representations:

Γ ) A1g + Eg + 2A2u + 2Eu The A1g and Eg internal normal modes (Figure 1) are Ramanactive. The A1g O-O stretching mode occurs near 830 cm-1 and will be the focus of this investigation. The Eg modes occur at much lower frequency. The remaining modes are the three acoustic (A2u + Eu) and three external translational (A2u + Eu) vibrations along the three principal axes of the elementary cell. The Ba2+ ion does not move in either the A1g or Eg vibrations in the k ) 0 limit. (Only the vibrations in the center of the Brillouin zone can be detected in the first-order Raman scattering spectra.) Results and Discussion The Raman spectra of the O-O stretching vibration of BaO2 powder under O2 atmosphere at 22 and 650 °C are presented in Figure 2. According to group theoretical analysis, only one totally symmetric O-O stretching vibration should exist in BaO2, but the Raman spectra show a main peak at 838 cm-1 (a) with at least two sidebands at 826 (b) and 814 cm-1 (c) at room temperature. This multiplet structure characterizes the Raman spectrum of both commercial and freshly made BaO2 at all temperatures investigated (22-850 °C) under both O2 and helium atmospheres. The room temperature spectrum was readily curve-fit by utilizing three spectral components (Figure 2A). No significant improvement in curve fitting was achieved when additional components were used. At higher temperatures, especially near 500 °C, a slightly poorer curve fitting was achieved with three components, but this was still considered to be an acceptable approximation even at very high temperatures (Figure 2B). © 1996 American Chemical Society

552 J. Phys. Chem., Vol. 100, No. 2, 1996

Haller et al.

Figure 1. Raman-active normal vibrations of BaO2.

Figure 3. Temperature dependences of the Raman shifts (A), bandwidth - fwhm (B), integrated intensities (C), and partial intensities (D) of the three curve-fit spectral components (a, b, and c) during heating (continuous line) and cooling (dotted line).

Figure 2. Raman spectrum of BaO2 under an O2 atmosphere in the O-O stretching region at 22 (A) and 650 °C (B). The curve fit with three spectral components a, b, and c is shown.

In order to understand the origin of the multicomponent structure of the O-O stretching vibration in BaO2, a detailed Raman investigation at temperatures between 22 and 850 °C and using different heating/cooling cycles was carried out. The spectra were found to vary with temperature as well as with the nature of the heating/cooling sequence. Figure 3A-C shows how each frequency, bandwidth (full width at half-maximum ) fwhm), and integrated intensity of bands a, b, and c of the sample varies with temperature and with the heating and cooling cycles. Figure 3D shows the percentage contribution from each spectral band to the total O-O stretching intensity. The Raman spectra were measured at least three times at each temperature: first, just after the desired temperature was reached, and then 30 and 60 min later. BaO2 starts to decompose fairly rapidly at about 500 °C. Oxygen, its different ions, and other possible impurities (BaCO3, BaO, Ba(NO3)2, Ba(OH)2, MgO, MgCO3) either have no firstorder Raman spectra (e.g. oxides) or have no Raman bands in the barium peroxide stretching region. Thus, these species do not contribute to the spectral features under investigation. It

was also confirmed that there was no contribution from twophonon scattering in this spectral region. The Raman spectrum of the commercial BaO2 under a helium atmosphere shows the same three-component shape with the same frequencies. The Raman spectra were independent of the excitation wavelength (515.4 or 532.1 nm). In each of the curves shown in Figure 3 the solid line represents behavior of the commercial sample to 800 °C. Upon cooling under the O2 atmosphere back to room temperature the freshly made sample is produced. What is particularly evident in Figure 3C is that the Raman spectrum of the commercial sample shows the presence of much less peroxide ion, especially of the type giving rise to band a, than the freshly prepared sample produced upon cooling. The commercial sample, described as having 95% purity by Aldrich, thus clearly contains less peroxide than the freshly prepared sample. Since the carbonate concentration, as described below, changes only little during the heating and cooling cycles, the commercial sample must contain much more oxide than the fresh sample, particularly at the surface of the finely ground particles which the Raman technique probes most effectively. (Deeper within the individual particles, the peroxide concentration may be higher.) Integration of the total peak area for the three O-O stretching bands has increased by a factor of 3 after heating and then cooling to produce the fresh sample. Significantly, the integrated intensity of band a, which later we ascribe to the purest form of barium peroxide, increases by a factor of about 6 during this cycle. There are also frequency and bandwidth differences between the commercial and fresh samples (Figure 3A,B), and the values on the cooling curve should be considered to be more representative of “pure” peroxide than the heated commercial sample. It is important to note that, after the fresh sample has been produced after heating above 800 °C and then cooled, the heating curves for Figure 3 closely match the cooling curves.

Raman Spectrum of Barium Peroxide

Figure 4. Raman spectrum of BaO2 in the O-O stretching region at 22 °C after 5 min (A) and after 150 min (B) of laser irradiation.

In other words, the solid lines for the commercial sample are characteristic only for the initial samples which clearly have a large amount of oxide impurity, especially on the sample surfaces. We have also examined (Figure 4) whether the Raman spectra of BaO2 demonstrate any photochemical changes produced by laser irradiation. The 838 cm-1 peak a in a commercial sample shifts to 839 cm-1 and continues to increase in intensity during 150 min of irradiation. In addition, the intensity of this peak increases relative to the two weaker bands (b and c). At longer irradiation times the frequency remains fairly constant. This is not an “aging” effect, since no changes were found between two Raman spectra taken just before the laser was turned off and 15 h later. Analogous effects were observed at all other temperatures. This seems to indicate that laser irradiation is helping to convert oxide ions to peroxides through a photochemical effect, perhaps through intermediate species (such as O3-) which can absorb light in this wavelength region. When the effect of heating was examined, all three Raman components showed shifts toward lower frequencies (in the opposite direction to that resulting from laser irradiation at constant temperature) at the average rate of about 0.015 ( 0.005 cm-1/deg up to the decomposition temperature (Figure 3A). The frequency shift was not exactly linear, especially for the main peak, as the largest frequency decrease took place at temperatures close to the BaO2 decomposition temperature. An accelerated decrease of frequency and an increase of fwhm could be characteristic of a soft mode.6 However, soft mode behavior is not common for stretching vibrations. Moreover, for BaO2 neither the frequency nor fwhm vs temperature experimental data could be approximated according to the soft mode behavior of a reasonable phase transition temperature. Upon changing the temperature up and down (within a temperature region below the decomposition temperature of the peroxide ion), a slight increase in the frequency was observed. Within experimental accuracy the measured Raman shifts were the same in both the oxygen and the helium atmospheres. When BaO2 was heated beyond its decomposition temperature of 800 °C and then cooled under the O2 atmosphere to a fixed temperature, a frequency shift ∆νT up to 9 cm-1 (compared with the Raman frequency at the same temperature before heating beyond 800 °C) was observed (Figure 3A). ∆νT is larger at lower temperatures, and it exists for all three Raman components. After heating the sample beyond its decomposition temperature and cooling, the shift in frequency vs temperature increased so that dν/dT ) -(0.022 ( 0.005) cm-1/deg. This value is applicable to both the main peak a and the stronger sideband b. Due to its broadness and weakness, it was not possible to accurately measure the frequency shift with tem-

J. Phys. Chem., Vol. 100, No. 2, 1996 553

Figure 5. Raman spectrum of BaO2 in oxygen atmosphere in the O-O stretching region at 22 °C before (A) and after (B) heating to 800 °C.

Figure 6. Raman spectrum of BaO2 under an oxygen atmosphere in the O-O stretching region at 600 °C in the cooling (A), then heating (B), then recooling (C), and then reheating (D) cycles for temperatures varied between 400 and 800 °C.

perature for the third Raman component c after heating at 800 °C, but its behavior looks very similar to that of the two first components. Repeating the heating-cooling cycle at temperatures below 500 °C several times always resulted in the same frequencies ((1 cm-1) following the dashed curves in Figure 3. The frequencies for the “fresh” sample, which resulted from heating beyond 800 °C and then recooling, persisted for at least 14 days when the sample was stored either under the O2 atmosphere or in air (Figure 5). As shown in Figure 6, the frequencies of the Raman bands recorded at 600 °C were found to shift depending on whether the sample was heated or cooled. The Raman frequencies of all the three components were 3.5 ( 0.5 cm-1 higher during the cooling cycles than during the heating cycles. This 3.5 cm-1 frequency difference was reproducible when the temperature cycle was repeated several times. The addition of CO2 to the O2 atmosphere had no noticeable effect on the frequency behavior described, supporting the view that the presence of CO32- is not of paramount importance. The temperature dependence of the bandwidth (fwhm) of all three Raman components during the heating and cooling cycles is presented in Figure 3B. The heating of a commercial or fresh sample results in a broadening of all Raman components. At room temperature the Raman components of the commercial sample have the following fwhm: Γ(a) ≈ 6.0 ( 0.5 cm-1, Γ(b) ≈ 8.0 ( 0.5 cm-1, and Γ(c) ) 6.0 ( 0.5 cm-1. The two components a and b have similar temperature dependencies, while c starts broadening much faster at a lower temperature than the other two, i.e. at about 300 °C instead of 500 °C. One can see an analogous behavior of the bandwidths in a helium

554 J. Phys. Chem., Vol. 100, No. 2, 1996

Haller et al.

atmosphere. At room temperature in helium Γ(a) = Γ(b) = 5.5 ( 0.5 cm-1 and Γ(c) = 6.5 ( 0.5 cm-1, while at 500 °C Γ(a) = Γ(b) = 11 cm-1 is very different from Γ(c) = 22 cm-1. There are significant differences in the bandwidth vs temperature dependencies after the sample is heated above 800 °C. The strongest component a broadens most, up to two times, while b shows even a smaller fwhm’s at T < 600 °C in the cooling cycle. The third component is also broadened after heating, but it is difficult to measure this effect with high accuracy. As was the case for the Raman frequencies, after a sample was heated above 800 °C, there were no large changes observed in the bandwidth curves during subsequent heating-cooling cycles. The Raman intensity measurements were made for samples under O2 atmospheres where the peroxide formation-decomposition ratio was only dependent on temperature. Under a He atmosphere the Raman intensity is time-dependent and varies from experiment to experiment. The absolute integrated intensities of all the three components (Figure 3C) increase with temperature, reaching maximum intensities at different temperatures: ∼400 °C for a, ∼500 °C for b, and ∼600 °C for c. As a result, the total absolute intensity increased almost linearly up to 500-600 °C, after which a rapid intensity decrease takes place. On cooling the sample (after it was heated at 800 °C) a very rapid rise of the total absolute intensity takes place. The maximum total intensity at 500-600 °C in the cooling cycle is twice as high as the maximum intensity at the same temperature in the heating cycle. There is a principal difference in the behavior of a compared to that of b and c. Namely, in the cooling cycle the temperature dependencies of I(b) and I(c) are similar to the corresponding dependencies in the heating cycle, while I(a) increases throughout the cooling cycle and practically reaches saturation at lower temperatures. On cooling to about 550 °C, I(a) reaches the same intensity as on heating for the same temperature, but this component is at least 6 times more intense when room temperature is reached again. The temperature dependencies of the partial intensities (I(i)/ ∑I(i)) are also different for I(a) and I(b), I(c) (Figure 3D). The partial intensity for the most intense component a decreases with temperature, while b and c show the opposite behavior. The partial intensities of the components a and c increase while that of I(b) decreases in the cooling cycle after the treatment of the sample at 800 °C. Within experimental error ((10%) there were no differences in the temperature dependencies of the partial intensities when the sample was heated several times to 800 °C and then cooled. In addition to the Raman line intensity measurements, the background levels were recorded at 650, 750, and 870 cm-1 at different temperatures. The background approximately doubles upon heating the sample to 500 °C. The greatest increase in the background (approximately by a factor of 1.6) occurs between 500 and 600 °C. At higher temperatures the background exhibits slower increases with temperature. An increase of the background level in nonresonant Raman spectra is related to the increase of irregularity and disorder in the crystal structure (see, for example, refs 7 and 8). The crystal lattice of BaO2 has the greatest defects when decomposition takes place, i.e., at temperatures higher than 500 °C. In addition to different possible oxygen species, a variety of oxygen-barium compounds9 may exist as impurities in the BaO2 lattice, each contributing to the background in the Raman spectrum. It is well-known that even at lower temperatures it is difficult to obtain BaO2 of purity higher than 90%.9 The chemical equilibrium

BaO + 1/2O2 a BaO2 or O2- + 1/2O2 a O22-

(1)

depends on the porosity of BaO and is very sensitive to the

TABLE 1: Relative Integrated Raman Intensities for the Carbonate Band Compared to the Multiplet Peroxide Band of BaO2 at Different Temperatures in the Heating and Cooling Cycles [I(CO32-)/I(O22-)] heating cycle (up to 800 °C)

cooling cycle (down from 800 °C)

temp, °C

immediately after T is reached

30 min later

immediately after T is reached

30 min later

25 300 500

0.068 0.056 0.052

0.067 0.059 0.060

0.053 0.053 0.037

0.061 0.051 0.058

content of impurities, especially to metal oxides which catalytically decrease the decomposition temperature of BaO2.10 Except at very low pressures, oxygen enters into the BaO2/BaO lattice in two ways:11 formation of more peroxide ions and/or insertion of molecular or atomic oxygen into the lattice structure. The latter leads to the rise of inhomogeneity of the BaO2 lattice and an increase in the background of the vibrational spectrum. The equilibrium in the BaO-BaO2 system is dependent on both temperature and pressure of oxygen. It has been proposed9,12 that at different oxygen pressures (i.e., at different BaO-BaO2 concentration ratios) three different regions can be distinguished in the phase diagram: BaO2 in BaO (when the BaO2 concentration is considerably less than 10%), BaO in BaO2 (for BaO2 concentrations 60-80% or greater, depending on temperature), and an intermediate phase. For our present experiments the systems can primarily be described as BaO in BaO2 where the BaO component is at least 5% at room temperature. BaO2 is readily contaminated by CO2 from the air to produce BaCO3, and we were concerned that some of the effects described above could be caused by the presence of carbonate ions in the BaO2. Thus, along with the O-O stretching region near 830 cm-1, we also recorded the Raman spectrum for the carbonate ion stretching region near 1055 cm-1 at all temperatures. The relative integrated Raman intensities of the carbonate stretching vibration compared to the peroxide multiplet structure in the O-O stretching region of BaO2 are presented in Table 1. The relative intensities calculated do not directly show the CO32-/O22- concentration ratio since the bands for the two molecules have different Raman scattering cross sections. However, our calibration studies of the integrated band intensities demonstrated that, in fact, the cross sections for the carbonate and peroxide bands are similar. Because the peroxide samples also contain some oxides, the actual relative amounts of carbonate in the samples will be somewhat higher than the values in Table 1. Thus, the amount of carbonate in the sample studied appears to be in the range of 5% for temperatures between 25 and 500 °C. There is no large difference in the carbonate content before and after the thermal treatment. (The somewhat lower concentration of CO32- at 500 °C in the rapid cooling cycle immediately after this temperature has been reached is caused by nonequilibrium conditions.) Despite the fact that there are essentially no carbonate ions present after heating to 800 °C, the re-formation of carbonate is evident as its Raman line appears again on cooling the sample even in O2 atmosphere. Its intensity rises nearly proportionally to the intensity of the O-O stretching band. As one can see from the Table 1, there is less than a 25% decrease in carbonate content compared with the peroxide content after the heating cycle to 800 °C has been completed. Evidently the carbonate content is smaller inside the sample and larger on the surface. This indicates that the carbonate ion content alone is most likely not responsible for all the large changes in the peroxide Raman spectrum. It is evident that the observed multiplet structure cannot be readily explained with a pure (perfect) BaO2 model. In addition to the carbonate ions, a crystal lattice of BaO2 is distorted by the presence of oxide ions and possibly by interstitial oxygen.

Raman Spectrum of Barium Peroxide The Raman spectra of both commercial and homemade (from BaO) samples show this. There are primarly two different ways in which BaO can be accommodated into the BaO2 crystal lattice: as oxide impurities replacing peroxides or as unistructural clusters or microcrystallites. In the BaO2 lattice the peroxide ion O22- has a tetragonal coordination with two different oxygen-barium distances, 2.68 Å (along the z axis in D4h symmetry) and 2.79 Å (in xy surface).13 This corresponds to two Ba2+ ions at 3.39 Å and four at 2.65 Å from the center of the peroxide ion (which may be replaced by an oxide ion upon doping with BaO). While the substitution of peroxide ions by oxide ions may alter these distances somewhat, the interactions between the peroxide ions and the two different types of Ba2+ neighbors should differ enough to affect the peroxide stretching frequency. Thus, for low (∼5%) concentrations of oxide impurities the three observed peroxide bands can be assigned as (1) “pure” BaO2, where none of the Ba2+ ions neighboring the peroxide ion have oxide ions as nearest neighbors (the highest frequency band a), (2) BaO2 with some oxide impurity, where one of the four nearer Ba2+ ions has an oxide ion as a closest neighbor (the middle frequency band b), and (3) BaO2 with some oxide impurity, where one of the two more distant Ba2+ ions has an oxide ion as a closest neighbor (the lowest frequency band c). If it is assumed that 5% of the peroxide positions have been replaced by oxide ions, 73% of peroxides are “pure”, of type a, while 16% and 8% are of types b and c, respectively. These calculated values are similar in magnitude to the partial intensities at room temperature for the fresh sample shown in Figure 3D and thus lend support to the model described. It should be noted that the assignments of the bands b and c could be reversed. However, the most intense and highest frequency band a should correspond to the “pure” BaO2. With this interpretation the data in Figure 3 can be understood in terms of models where more or less of the peroxide sites are occupied by oxide ions as the equilibrium of eq 1 is affected. Higher temperatures give rise to more oxide and less pure peroxide. As shown in Figure 3D, at 650 °C band c increases relative to band a. For the above explanation of the multiplet band structure for the peroxide, the frequency differences between the three bands are also reasonable. The shifts of -13 and -21 cm-1 in the peroxide frequencies are caused by interaction with the slightly altered electronic structures and positions of Ba2+ ions which have been somewhat perturbed by the replacement of a single peroxide unit by an oxide ion. The real structure of the BaO2 sample is naturally more complicated due to the presence of carbonate ions. The CO32ion as an impurity disturbs the BaO2 lattice much more than an oxide impurity, due to its larger size. However, we believe that its effect on the O-O stretching frequency is less specific (more random) than that from the oxide impurity. In other words, while the presence of oxide ions appears to give rise to three distinct Raman bands, the carbonate ions tend to only broaden the spectrum. A second model which could explain the multiplet structure of the O-O stretching band assumes an inhomogeneous distribution of oxide ions such that microcrystallites may be formed. An inhomogeneous distribution of BaO2 has been found in BaO2/MgO catalyst mixtures,14 where XPS measurements showed that the pure BaO2 layers covered pure MgO microcrystallites. The structure of such a BaO2-BaO-O2 system would probably be very complicated and would strongly depend on physical conditions.9 The chemical and phase contents in the BaO2-BaO-O2 system are also dependent on oxygen pressure, and a complicated phase diagram of three phases has been proposed.9,12 For such a case, the Raman

J. Phys. Chem., Vol. 100, No. 2, 1996 555 spectrum could show sevreal different O-O stretching frequencies. However, the origin of three specific bands would be difficult to identify. Conclusion Both of these structural models for the BaO2 with BaO and O2 can be used to explain the temperature dependencies of the spectral characteristics given in Figure 3. However, the more homogeneous model with three specific types of peroxide ions appears to be more nicely consistent with our observation. In summary, we have found the following: (1) Initially (and surprisingly) there is quite a large oxide concentration in the commercial barium oxide. Within individual particulates the oxide concentration may be lower and the peroxide concentration may be higher since the Raman technique produces the strongest spectra nearer the particulate surfaces. (2) Heating the commercial sample under an oxygen atmosphere first produces more peroxide (up to about 500 or 600 °C) but then results in nearly total decomposition of peroxide to oxide after 800 °C is reached. (3) Cooling the sample from 800 °C under the O2 atmosphere produces the purest BaO2 sample which has Raman bands with 3 times the integrated intensity of the commercial sample when the temperature approaches room temperature. (4) The three Raman peaks arise from peroxide ions with slightly different chemical environments. Peak a represents the “pure” sample, and this is seen to grow in intensity when oxide impurity concentrations are decreasing. Peaks b and c are produced through the interaction of oxide ions with Ba2+ ions, thus slightly altering the peroxide/barium ion interactions. After heating above 800 °C and then cooling, the partial intensity of b decreases substantially, reflecting the decrease in oxide impurity concentration. (5) The carbonate ion concentrations appears to change little with temperature. After heating to 800 °C when the carbonate decomposes to produce CO2, the slow purge of O2 gas should remove the CO2 gas. However, the re-formation of carbonate is observed, possibly indicating that CO2 molecules may be trapped within the crystallites. The presence of carbonate does not seem to add any spectral features to the peroxide stretching region. Acknowledgment. The authors thank the National Science Foundation and Robert A. Welch Foundation for financial support. References and Notes (1) Lunsford, J. H. Catal. Today 1990, 6, 235. (2) Wolf, E. E., Ed. Methane ConVersion by OxidatiVe Processes; Van Nostrand/Reinhold: New York, 1992. (3) Lunsford, J. H.; Yang, X.; Haller, K.; Laane, J.; Mestl, G.; Kno¨zinger, H. J. Phys. Chem. 1993, 97, 13810. (4) Eysel, H. H.; Thym, S. Z. Anorg. Chem. 1975, 411, 97. (5) Foeppel, H. Z. Anorg. Allg. Chem. 1957, 291, 46. (6) Scott, J. F. ReV. Mod. Phys. 1974, 46, 83. (7) Brodsky, M. H. Raman Scattering in Amorphous Semiconductors. In Light Scattering in Solids; Cardona, M., Ed.; Springer-Verlag: Berlin, 1975; p 205. (8) Taylor, P. C. Laser Spectroscopy in Amorphous Solids. In Laser Spectroscopy of Solids II; Xen, W. M., Ed.; Springer-Verlag: Berlin, 1988; p 257. (9) Vol’nov, I. I. Perekisnye Sojedinenija Stchjolotchno-zemel’nykh MetalloV; Nauka: Moscow, 1983. (10) Kendall, J. K.; Fuchs, F. J. J. Am. Chem. Soc. 1921, 43, 2023. (11) Kozlenko, T. A. SoV. J. Phys. Chem. 1967, 6, 1369. (12) Kedrovskij, O. V.; Koovtunenko, P. V.; Bundel’, A. A. SoV. J. Phys. Chem. 1967, 51, 414. (13) Abrahams, S. C.; Kalnais, J. Acta Crystallogr. 1954, 7, 841. (14) Dissanayake, D.; Lunsford, J. H.; Rosynek, M. P. J. Catal. 1993, 143, 286.

JP952426J