J. Chem. SOC., Faraday Trans. 1, 1985, 81, 19-36 Study of Methanol and Water Chemisorbed on Molybdenum Oxide BY JONG s. CHUNG, RAUL MIRANDAT AND CARROLL 0. BENNETT* Department of Chemical Engineering, University of Connecticut, Storrs, Connecticut 06268, U.S.A. Received 16th January, 1984 Methanol and water chemisorbed on finely divided molybdenum oxide have been studied by infrared spectroscopy and by dynamic adsorption and desorption techniques. Isolated hydroxyl and two kinds of water on the surface desorb at ca. 150 "C, while hydrogen-bonded hydroxyls desorb at 350 "C. Most of the methanol chemisorbs dissociatively, even at room temperaure, and at least three forms of chemisorbed methanol exist : methanol dissociated into methoxy and hydrogen across Mo=O (form C), methoxy on the vacancy of terminal-bonded oxygen Mo=O (form A) and methoxy on the vacancy of bridge-bonded oxygen Mo-0-Mo (form B).Most of the form C desorbs reversibly as methanol below 110 "C and a part of the form A is desorbed as methanol with a peak at 110 "C during temperature-programmed desorption. The rest of the methoxy groups are decomposed into formaldehyde and CO at ca. 260 "C. The mechanism of the selective oxidation of methanol on Mo-Fe-0 mixed oxides has been extensively studied since the early There is a growing concern about the importance of the methoxy group as a precursor of various products.276-8 However, the presence of the species has not been confirmed by infrared spectroscopy, since the surface area of conventionally made MOO, or molybdate is very low.Also it is known that water in the gas phase is a strong inhibitore and that the selectivity for the products depends on the hydroxyl content on the catalyst surface.l0 This paper presents a basic study of the chemisorption of water and methanol on MOO, in order to contribute to our understanding of the nature of the reaction intermediates on active sites. Infrared spectroscopy and dynamic adsorption/ desorption techniques have been used to study water and methanol adsorbed on unsupported MOO,. Finely divided MoO,ll prepared in a flame reactor is suitable for the infrared study because of its high surface area and good transmission qualities. EXPERIMENTAL CATALYST PREPARATION The catalyst consisted of finely divided unsupported MOO, particles made in a flame reactor and obtained from Teichner's laboratory in Lyon.ll The B.E.T.surface area after extensive treatment in oxygen at 380 "C was 27 m2 g-l and the average particle size measured by X-ray line broadening was 450 A. The reduced MOO, for the adsorption experiments was obtained by passing 3.6% methanol in helium over the catalyst at 150 "C, then passing helium for < 5 min to clean the surface. The reduced catalyst was cooled to room temperature quickly in -f Present address : Department of Chemical Engineering, University of Louisville, Louisville, Kentucky 40292, U.S.A. 19 2-220 CHEMISORPTION OF H20 AND CH,OH ON MOO, order to minimize oxygen diffusion from the bulk. The desired degree of reduction could be controlled by the length of the reduction with methanol. A higher degree of reduction was possible by repeating the reduction and cleaning cycle.The degree of reduction on the surface of the catalyst was measured by the normalized absorbance at 1320 cm-l, as described previously.12a Oxodized MOO, refers to the catalyst obtained by heating at 380 "C overnight in oxygen and cooling to room temperature in oxygen. DYNAMIC ADSORPTION AND DESORPTION EXPERIMENTS EXPERIMENTAL SYSTEM The experimental system consisted of the feed system, which was the same as that for the infrared system, the reactor and the mass spectrometer with a minicomputer. The differential reactor and tubing were made of SS316. The grain size of the pressed catalyst particles was between 60 and 100 mesh. The analysis of gases was made with an on-line magnetic-sector-type (12 in* radius analyser) nuclide mass spectrometer which was connected to a MINC[(R)DEC] minicomputer to follow up to 3 peaks per second.Since various components had contributions to several of the monitored peaks, their individual cracking patterns were determined using pure components in order to convert the collected data into concentrations against time. EXPERIMENTAL PROCEDURE Just before an adsorption experiment MOO, was subjected to another temperature increase up to 400 "C in 3 % 0, +He to remove hydrated water. The catalyst was then cooled, purged with helium for 10 min and an adsorbate admitted. The pulse duration was 4 min, a time sufficient to attain a steady state. Subsequently, helium was introduced into the reactor for 5 min.Still in flowing helium, the temperature of the reactor was increased at an initial rate of 60 "C min-l and an average rate of 23 "C min-l until desorption activity ceased. ANALYSIS The analysis of the reactor inlet and outlet was accomplished by monitoring peaks at mass/charge ratios of 1 8(H,O), 28(CO), 30(CH,O), 3 1 (CH,OH) and 40(Ar), representing all the species present below 100 "C. Sensitivities stayed in the 1 ppm range with white noise of k2 ppm. In order to eliminate some residual noise (60 Hz), 100 samples were taken in 35 ms and averaged to get each peak. In order to calculate the amount of methanol adsorbed (or desorbed) during the step input of gaseous methanol (or helium), a tracer component, Ar, in the gas mixture was used as a standard curve of step change in the reactor system.After the concentration of Ar was set equal to that of 3.6% methanol, the difference in the response curves between adsorbate and Ar was used to calculate the amount of methanol adsorbed (or desorbed). INFRARED METHODS The whole system was the same as that described When the chemisorbed intermediates of methanol and water were observed, methanol and water were passed through both the sample and reference cell and the infrared beam was balanced carefully between the sample and reference cell by adjusting the distance between the surface of the catalyst (or the stainless-steel mirror in the reference cell) and the infrared window. In order to reduce the severe transmission loss in the high-frequency region, above 3 100 cm-', CaF, of 1 mm thickness was used for the window instead of ZnS.A low concentration of oxygen in the methanol mixture also results in the loss of transmission in the high-frequency region, especially at high temperatures, because of the reduction of MOO,. To avoid this, methanol in oxygen was used during the chemisorption of methanol on MOO,. The optical system was purged with dry air to minimize the contamination of infrared bands by moisture. Real transmission values of the samples are < 5% in the 4000 cm-l region and ca. 10-2074 in the region from 2000 to 1000 cm-l. * 1 in = 2.54 x m.J. S. CHUNG, R. MIRANDA AND C. 0. BENNETT 21 RESULTS AND DISCUSSION INFRARED SPECTRA OF HYDROXYL GROUPS Fig. 1 shows the OH bands of water adsorbed and irreversibly held in helium at various temperatures.Bianchi et al.13 have observed OH bands of adsorbed water on the same catalyst and concluded that water did not chemisorb dissociatively. This conclusion was based on the bands observed at 3500 and 1610 cm-l. Besides these bands we detected more bands at 3685-3705,3375,3255 and 1625 cm-l. The impurity bands above 3705 cm-l result from moisture in the optical chamber. Even though dry air was purged into the optical chamber, the contamination could not be removed completely because the balanced beam was greatly magnified in order to observe the weak OH bands of water. This contamination became more severe when OH bands of chemisorbed methanol were observed. The band at 3685-3705cm-l is also confirmed by observing OD bands with deuterated methanol (vide infra) and is assigned to isolated hydroxyl.The band at 3500cm-l is assigned to physically adsorbed water. The bands at 1690 and 1640 cm-l have already been assigned to an impurity and to a structural band of MOO,, re~pective1y.l~ A close look at fig. 1 reveals that the bands at 3480-3500 and 1610 cm-l disappear at ca. 100 "C and the bands at 3585 and 1625 cm-1 remain up to 185 "C. Therefore these two sets of bands correspond to water chemisorbed undissociatively, since they have water deformation bands at 16 10 and 1625 cm-l. There remain no corresponding water deformation bands for the bands at 3375 and 3255 cm-l. Dihydrated MOO, has coordinated and interlayer water bands at 3520, 3410,3200, 3140 and 1615 cm-l, and monohydrated molybdenum oxide has coordinated water bands at 3430 and 1605 crn-l.l4 However, it is known that upon heating, the dihydrate and mono- hydrate bands disappear at 80 and 130 OC, respectively.Therefore the two bands at 3375 and 3255 cm-l cannot be water bands of hydrated MOO, because they remain up to 350 "C. The assignments of these two bands will be discussed later. In order to find proper chemisorption sites for the two chemisorbed waters which have bands at ca. 3480 and 1610 cm-l and ca. 3580 and 1625 cm-l, we have studied the effect of traces of water in the carrier gas on the chemisorption of water on MOO,; the results are shown in fig. 2. Carrier-grade helium and zero-grade oxygen (Aero All Gas Co., Hartford, CT) were used without further purification, so that they contained traces of water.The band at 3400 cm-l is developed by the water in the oxygen and the band at 3500 cm-l by the water in the helium. If we assume that there is less moisture impurity in helium than in oxygen the results indicate that oxygen has a blocking effect for the chemisorption of water on reduced sites. This leads us to the conclusions that the water band at ca. 3500-3580 cm-l in fig. 1 and 2 corresponds to water chemisorbed on a reduced site of MOO, and that the band at ca. 3460- 3480 cm-l corresponds to water on an oxygen of MOO,. The results also imply that there are vacancies on the oxidized MOO,. Fig. 3(A) shows the OH bands which are developed after passing methanol+ oxygen over MOO, at 27°C. All of the OH bands which are developed by the adsorption of water on MOO, are also found here at the same frequencies.The absence of any new hydroxyl bands indicates that methanol dissociates to form methoxy and hydrogen on the surface of MOO,. The main difference between the OH bands produced by water and by methanol is that water adsorption shows a maximum peak at 3500 cm-l, whereas methanol adsorption shows a maximum at 3375 cm-l. Almost no increase in the water deformation band at 1625 cm-l is observed at room temperature, suggesting that the bands at 3375 and 3255 cm-1 do not arise from water22 CHEMISORPTION OF H20 AND CH,OH ON MOO, Fig. 1. Infrared bands of hydroxyl group and water in OH stretching (left) and water in deformation region (right) on MOO,, produced by water adsorption and desorption at (a) 320, (b) 250, (c) 185, (d) 145, (e) 90, cf) 50 and (g) 27 "C.(-) 4.6 Torr water in helium and ( * * - -) after a purge with helium for 20 min. The ordinates of the spectra are displaced to avoid overlapping of traces. molecules. Hydroxyl bands near 3375 cm-l are also observed with Ti0215 and a-Fe,0,.16 Hydroxyl bands developed by the chemisorption of methanol at temperatures > 27 "C are shown in fig. 3(B). When methanol in oxygen is present in the gas phase at 60 "C, the intensities of the water peaks are greater than those of the bands at 3375 and 3325 cm-l, indicating that water desorption is the rate-determining step at this temperature. As the temperature increases, however, the amount of chemisorbed water decreases sharply and almost disappears at 200 "C in the presence of 3.6% methanol in oxygen.Unfortunately it is impossible to observe the OH bands in the presence of lower oxygen concentrations than 3.6% CH,OH + 96.4% oxygen, especially at high temperatures, because of the severe loss of the background transmission at high wavenumber. The shift in the hydroxyl band at 3375 to 3395 cm-l above 100 "C is probably related to the reduction of MOO,. In order to see the effect of the reduction of MOO,, OD bands were studied because there was still enough transmission in the OD vibration region with reduced MOO,.J. S. CHUNG, R. MIRANDA AND C. 0. BENIWTT 23 wavenumber/cm-' Fig. 2. Inhibition effect of oxygen on the chemisorption of water on MOO, observed in the OH stretching region at 27 "C. (- - - -) Background of MOO,. (a) He, 2 h ; (b) He, 1 day; (c) 0,, 1 day; ( d ) 0,, 2 days.Fig. 4(A) shows OD bands on the oxidized (upper part) and 10% reduced MOO, (lower part) after D,O, CH,OD and CD,OD have been passed over the catalysts for 20 min at 27 "C, and fig. 4(B) shows the OD bands remaining on both oxides after a purge with helium. On the oxidized MOO,, all the OD bands which correspond to the five OH bands in fig. 1 and 3 are observed as shown on the lower part in fig. 4(B). On the reduced MOO,, four of them are shifted to higher wavenumbers compared with those on the oxidized MOO,, but the band 2435 cm-l is not shifted. Table 1 shows the band positions of all OH and OD bands on both oxidized and reduced MOO,. We have already assigned the two water bands based on the results in fig. 2.With the results obtained so far and based on the structure of MOO, we will now assign the three hydroxyl bands. MOO, is known to have a layered structure17 with a protruding terminal double- bonded oxygen (Mo=O), two single-bonded oxygens shared by two Mo and another three shared by three Mo (Mo-0-Mo). From now on we will use OT to denote the double-bonded oxygen in Mo=O, 0, for the bridge-bonded oxygen in Mo-0-Mo, and V, and V, for their corresponding oxygen vacancies. For water and methanol to produce hydroxyl on the surface they must be dissociated as follows: H,O+V+OL + HOV+HOL (1) (2) where the subscript L represents lattice oxygen to distinguish it from the oxygen of adsorbates, V represents an oxygen vacancy of either VT or V, and * is used to represent a site on the surface on which methoxy chemisorbs.According to reactions (1) and (2), hydroxyl can be produced by an oxygen vacancy upon decomposition of chemisorbed water, while this is not true for methanol unless chemisorbed methanol H,COH + * + 0, -+H,CO* + HOL24 CHEMISORPTION OF H,O AND CH,OH ON Moo, 3900 3700 3500 3300 3100 2900 wavenum ber/cm-' Fig. 3. For legend see facing page. can break the CO bond as easily as water does the OH bond. Breaking the OH bond in water is much easier than breaking the CO bond in methanol on the reduced sites of MOO,. Evidence for this is provided by the result that the irreversibly chemisorbed hydroxyls produced by D,O chemisorption have higher intensities than those formed by methanol chemisorption, especially when MOO, is reduced, as shown on the upper part of fig.4(b). On the oxidized catalyst, the hydroxyls produced by CD,OD show the highest intensity among the three adsorbates shown in fig. 4(b), indicating that breaking the CD bond, in other words the'production of surface formaldehyde, occurs even at room temperature. Once the surface is covered with hydroxyls formed by either reaction (1) or (2), they may be linked with each other by multiple hydrogen bondings: H H I I I I I I I I 0. .. . . . H . . . . . O . . . .. . H . . . . . .O Mo-0-Mo-0- Mo VT VB The hydrogen-bonded hydroxyl on V, (or HOB) is affected more strongly than the free hydroxyl on V, (or HOT), since the former interacts directly with neighbouring terminal oxygens. As the catalyst is reduced, oxygen vacancies are formed on the OT and OB sites so that the chance of hydrogen bonding between hydroxyls is reduced.J.S. CHUNG, R. MIRANDA AND C. 0. BENNETT /-I_ 3395 I 3700 3500 3300 3100 2900 wavenumber / cm- l 25 Fig. 3. (A) Hydroxyl bands produced by adsorption and desorption of methanol on MOO, at 27 "C. (-) 3.6% methanol in oxygen and ( * * .) after purging with helium for (a) 35, (b) 25 and ( c ) 6 min. The ordinates of spectra are displaced to avoid overlapping of traces. (B) As (A) but taken at (a) 200, (b) 160, (c) 100 and (d) 60 "C. (. - - .) After a purge with helium for 20 min. Thus the hydroxyl band observed at 3685-3740 cm-l is assigned to free hydroxyl on V, and the band at 3375-3430 cm-l to multiple hydrogen-bonded HOB.The band at 3235-3255 cm-l is at an unusually low wavenumber for a hydroxyl band. Chromia18 showed the same kind of broad band at ca. 3280cm-l, but no assignment was given. On MOO, it probably results from hydroxyl perturbed by hydrocarbon impurities. Strong interactions between hydroxyl and chemisorbed hydrocarbons develop wide bands at 3300cm-l on ~i1ica.l~ Oxidized MOO, has impurity bands at 1950, 1930 and 1880cm-l, which are similar to the bands of a molybdenum carbonyl complex.20 However, the mechanism of interactions between hydroxyl and impurity bands is not clear. The band could be assigned to a hydroxyl in the interlayer of MOO,, because only this band is not shifted upon the reduction of MOO, and no corresponding water deformation band is observed.Molybdenum oxide is known to form hydrogen bronzes H,MoO, (0 < x < 2). Recent studies,' show that H is held between the MOO, layers forming OH, and for higher concentration of H forming OH, groups.26 CHEMISORPTION OF H20 AND CH,OH ON Moo, 2592 2800 2700 2600 2500 2400 2300 wavenumber/cm- 1 Fig. 4. For legend see facing page. INFRARED SPECTRA OF METHOXY GROUPS In the case of methanol chemisorbed on other oXides22-2s the chemisorbed form of methanol is usually regarded as methoxy after being compared with the bands of metal methoxides. It has been assumed that the methoxy on metal oxides is formed either by a condensation reaction or by the opening of the metal-oxygen bond:22-26 H CH3 I 1 CH, H I I 0 0 I I M+CH,OH+M + M. /"\ M (3') Generally three strong bands are observed at ca.3000,292&2960 and 28 15-2860 cm-l and assigned to overtones or combinations of CH, deformation (6) and asymmetric (v,) and symmetric (v,) stretching, respectively. In the case of silica,24 4 bands areJ. S. CHUNG, R. MIRANDA AND C. 0. BENNETT 27 (B) 2690 I I ox; d ; sed M O 03 2735 26kO . 2435 2580 I 2500 wavenumber/cm-' Fig. 4. (A) Infrared bands of OD on reduced and oxidized MOO, at 27 "C after adsorptions of (a) 3.6% CH,OD+He, (b) 3.6% CD,OD and (c) 5 Torr D,O. (B) As (A) for irreversibly bound OD remaining after a purge with helium for 20 min. Table 1. Wavenumbers and assignments for the irreversible hydroxyl bands at 27 "C wavenumber/cm-l oxidized MOO, reduced MOO, OH OD OH OD (3685) 2735 (3740)" 2770 3580 2650 (363 1) 2690 3480 2580 (3534) 2618 3375 2500 (3529) 2540 3255 2435 (3287) 2435 assignment free hydroxyl of HOT water in oxygen vacancies water on oxygen of MOO, hydrogen-bonded HOB hydroxyl perturbed by hydrogen impurities, or in interlayer of MOO, " The number in parenthesis is a value calculated with the ratio of v(OH)/v(OD) = 1.35.28 CHEMISORPTION OF H20 AND CH,OH ON Moo, t 2955 3100 3000 2900 2800 2700 wavenumber/crn-l Fig.5. Infrared spectra of methanol in the gas phase (a) and chemisorbed on MOO, (b) at 27 "C. (-) 3.6% methanol + oxygen, ( * * a ) after purging with helium for 20 min and (- - -) MOO,. observed above 2900 cm-l, and the band at 2928 cm-l is assigned to an overtone or combination of CH, deformations. Fig. 5 shows the CH stretching bands of gas-phase methanol and methanol adsorbed on MOO, at 27 "C.According to the assignment of others the bands at 3006, 2955-2965 and 2845-2854 cm-l in fig. 5 are assigned to 6, v, and v,, respectively. For the two weak bands at 2930 and 2830 cm-l, we prefer the assignment as v, and v, of a different type of methoxy, since these bands do not seem to correspond to any reasonable combination of those of the strong methoxy. Also, the intensity ratio and the distance between these two weak bands are similar to those between the bands at 2955 and 2845 cm-l. The second weak v, at 2830 cm-l has not been observed on other oxides. The infrared spectra of [2H,]methanol chemisorbed on MOO, also confirm the existence of the two weak bands, as shown in fig. 6. All the corresponding bands developed by chemisorbed methanol are detected by the chemisorption of [2H,]methanol, although the band intensity at 2177 cm-l is weaker than the intensity of the corresponding band 2955 cm-l in fig.5. Wavenumbers and assignments of the observed bands are given in table 2. Fig. 7 shows the bands in the CH, deformation region developed by the chemi- sorption of methanol on molybdenum oxides. Other bands which are not specified in fig. 7 result from moisture in the optical chamber. The band at 1430 cm-l is alsoJ. S. CHUNG, R. MIRANDA AND C. 0. BENNETT 29 , I I' 2 2 4 0 I I I I 2000 2300 2200 2100 wavenum ber/cm-l Fig. 6. Infrared spectra of chemisorbed CD,OD in the CD stretching region at 27 "C. (-) 3.6% methanol+ helium, (. * * .) after purging with helium for 20 min and (- - -) MOO,.Table 2. Wavenumbers and assignments of observed bands of chemisorbed methanol and [2H,]methanol at 27 "C CH,OH CD,OD a b a b assignment 3006 - 2955 2965 2925 2928 2845 2854 - 2825 1467 1467 1445 1445 13781 1345 2248 2284 overtone or combination of CH, deformation, form A 2225 overtone or combination of CH, deformation, form B 2177 2185 v,,(CH,), form A 2136 2139 vas(CH3), form B 2075 2085 v,(CH,), form A 2068 v,(CH,), form B - - - - CH, deformation, form A - - CH, deformation, form B CH deformation, formaldehyde - - a In the presence of 3.6% methanol in the gas phase. After purging with helium.30 CHEMISORPTION OF H,O AND CH,OH ON Moo, 1445 I I I I 1550 1450 1350 wavenumber/cm- ' Fig. 7. Infrared spectra of chemisorbed CH,OH in the CH, deformation region at 27 "C with 3.6% CH,OH+He.(a) After catalyst is oxidized at 350 "C for 2 h, (b) reduced with 3.6% methanol+ helium at 145 "C for 12 min and (c) reduced under the same conditions as (b) for 45 min. observed on haematite22 and is assigned to a vibrational or electronic transition produced by depletion of structural oxygen atoms from the surface. Both chemisorbed formaldehyde and formic acid can develop the bands at 1378 and 1345 cm-l. In the case of MOO,, these are assigned to formaldehyde since the band at 2785 cm-l, which belongs to chemisorbed formaldehyde, is observed in fig. 6 (2033 cm-l as CD). The two bands at 1467 and 1445 em-l are CH, deformation bands, but it is not clear whether they represent two different species of methoxy or not.Although the methanol molecule shows three CH, modes, usually only one band is observed at 1465-1475 cm-l in the case of other oxides.26-28 There is an exception: haematite22 shows two 6(CH,) bands at 1460 and 1440 cm-l. However, they are regarded as arising from one type of methoxy, because only one set of v(CH) at 2920 and 2815 cm-1 is observed. Fig. 8 is further evidence of the existence of two different types of methoxy. As the degree of reduction of MOO, increases, the intensity ratio of the band at 1445 cm-l to that at 1467 cm-l increases. This ratio increases even more in the case of methanol held irreversibly after purging with helium, as shown on the right-hand side of fig. 8. Of course the total amount of methoxy on the surface increases as the reduction of MOO, increases.Note that the intensity of the formaldehyde band at 1378 em-l decreases as the reduction increases. This results from a deficiency of surface oxygen capable of taking hydrogen from chemisorbed methoxy. that oxygen of Mo=O is more labile than that in Mo-0-Mo on the surface and that the intensities of the weak CH stretching bands at 2930 and We haveJ. S. CHUNG, R. MIRANDA AND C. 0. BENNETT 31 I I I 1550 1450 1350 wavenumber/cm-' i ; 1378 . . I . ( C ) : I , , .' I , , I 1467 j W'.. . : , . L, 1448 wavenumber/cm-' I I I 1550 1450 1350 Fig. 8. Effect of reduction on relative strength of two different species of methoxy observed at 27 "C in the presence of 3.6% CH,OH + He (-) and after a purge with helium for 20 min ( - . . .).(a) Oxidized at 350 "C for 2 h, (b) 3% reduced on the surface and (c) 10% reduced on the surface. 2830 cm-l increase with increased reduction of MOO,. Therefore the 6(CH3) band at 1445 cm-l is connected to the v(CH) bands at 2930 and 2830 cm-l, and the 6(CH3) band at 1467 cm-l to the v(CH) bands at 2965 and 2854 cm-l. Since the Mo=O is more labile than Mo-0-Mo, it is natural to assume that the abundant methoxy with bands at 2965,2854 and 1467 cm-l chemisorbs on the oxygen vacancy of Mo=O (V,) and the methoxy with weak bands at 2930, 2830 and 1445 cm-l chemisorbs on the oxygen vacancy of Mo-0-Mo (V,). When MOO, is reduced at 150 "C, the oxygen vacancy of Mo-0-Mo (V,) is also generated and this is related to the increase of the bands at 2930, 2830 and 1445 cm-l.The results in fig. 8 show that methoxy chemisorbs on V, more strongly than on V,. Takezawa and Kobayashi30 have correlated the CH stretching bands of methoxy chemisorbed on metal oxides with the acidity (electronegativity) of the oxide; on more acidic oxides the CH stretching bands of methoxy are located at higher frequencies. Therefore the V, site is more electronegative (acidic) than the V, site because methoxy on V, has CH stretching bands at higher frequencies than that on V,. This can also be explained by the geometric effect. The methoxy on V, has more tendency to hydrogen bonding with neighbouring 0, than that on V,, which may cause CH stretching bands of methoxy on V, to shift to lower frequencies. The lower frequencies and higher probability of hydrogen bonding in the methoxy chemisorbed on V, indicate that this species is probably responsible for the pro-32 CHJZMISORPTION OF H,O AND CH30H ON MOO, duction of the observed byproducts such as dimethyl ether, dimethoxymethane and methyl formate when combined with methoxy on the V,, as follows: CH3 I I 0 ----- CH3 I 0 MO- - CHSOCHB (DME) CH3 I I CHOOCH3(MF) .CH - 0 --___ 0 MO - I The CH stretching bands of methoxy observed above 27 "C are shown in fig. 9. The results show that, especially in the presence of methanol+oxygen in the gas phase, the frequencies of the bands shift to higher values as the temperature increases. In the case of methoxy groups chemisorbed on other oxides, no band shift is observed as a function of temperat~re.~*9~~ Therefore the band shifts observed at low temperatures and in the presence of methanol in the gas phase probably indicate the existence of another type of methoxy or methanol. The existence of methanol adsorbed undissociatively is negligible if we recall that there is no new hydroxyl band observed on the chemisorption of methanol in contrast to those formed on the chemisorption of water, Methanol could decompose into methoxy and hydrogen across Mo=O as follows: HqC 0 0 Mo-00-MO + CH,OHT-- Mo-O-Mo II I1 \ / O II (4) form c Some loss (ca.10%) of Mo=O upon adsorption of methanol has been observed even at temperatures < 100 "C.lab After adsorption, however, a switch to helium returns the loss to < 10% of its original value, indicating that reaction (4) is reversible. If reaction (4) is reversible, form C will disappear after flushing the surface with helium at high temperatures.Therefore a negligible shift of the bands at 110 "C after a switch from methanol to helium indicates that the bands observed at 110 "C in the presence of helium in fig. 9 represent the pure methoxy groups chemisorbed in the V, and V,: CH3 I 0 0 I II Mo-0-MO V,, 2964 and 2858 cm-l form A H3 I 0 c o II I II Mo-0-MO VB, 2938 and 2834 cm-l. form BJ. S. CHUNG, R. MIRANDA AND C. 0. BENNETT 33 2958 ~~ 3100 3000 2900 2800 2700 wavenumber/crn-' Fig. 9. Infrared spectra of methanol chemisorbed on MOO, in the CH stretching region at (a) I 10, (b) 90 and (c) 50 "C. (-) 3.6% methanol + oxygen and ( . . - .) after purging with helium for 20 min. The ordinates of spectra are displaced to avoid overlapping of traces.DYNAMIC ADSORPTION AND DESORPTION MEASURED BY MASS SPECTROMETRY Because of appreciable reduction of molybdenum oxide in the presence of methanol above 100 "C,lZa this experiment was limited to temperatures < 100 "C. Fig. 10(a) and (b) show the results obtained during the adsorption of CH,OH at 30 "C and temperature-programmed desorption of the chemisorbed species. Even at 30 "C, water [I11 in fig. lO(b)] is produced when methanol+ 5% oxygen is passed over MOO,. This indicates that methanol is dissociated into methoxy and hydrogen and that some of the lattice oxygens are so labile they produce water: HOL+HOL + H,O+OL. ( 5 ) Upon a switch to helium, methanol chemisorbed reversibly [I in fig. 10 (a)] and water [IV in fig.lO(b)] are desorbed. The amount of formaldehyde and CO desorbed are negligible. The catalyst temperature is then raised with helium (region A'). First methanol [I1 in fig. lO(a)] is observed with a peak maximum at 110 "C, followed by a water peak at 155 "C. Formaldehyde and CO are detected at higher temperatures with peak maxima at 235 and 245 "C, respectively. Another water peak appears as a shoulder at 265 "C. The results, together with the results obtained by infrared spectroscopy, show that there are at least three kinds of chemisorbed methoxy: reversible, irreversible and desorbed into methanol, and irreversible and decomposed34 5 4 n s c 3 .- Y (d 0 & - E o 2 1 ' 0 CHEMISORPTION OF H,O AND CH,OH ON MOO, I A 1 A' ....- ...... I-* 0 3.5 7:0 10.5 14 time/min 0 .8 ' . O 0 . 2 I 0 3: 5 710 10.5 14.0 time/min Fig. 10. (a) Dynamic adsorption and desorption of methanol on 118 mg MOO, with flow rate of 0.0013 mol min-l. Region A, pure helium; region B, 3.6% methanol+5% oxygen+3% Ar + balance helium. Region A' represents temperature-programmed desorption with helium. (-.--) Water, (. - * -) methanol and (-) Ar tracer. (b) Same as (a): (--.--.) water, (-) formaldehyde and ( - - - a ) CO. into formaldehyde and CO upon desorption. The desorption of water with a peak at 155 "C during the temperature-programmed desorption is related to the disappear- ances of the two water bands chemisorbed on oxidized and reduced sites (3480 and 3580 cm-l) and the isolated hydroxyl band (3685-3705 cm-l) in fig. 1. The hydroxyl bands at 3375 and 3255 cm-l in fig.1 correspond to the desorption of water with a peak at 265 "C in fig. lO(b). Fig. 11 shows the amounts of the three different kinds of chemisorbed methanol as a function of adsorption temperature. The total amount of methanol measured by the infrared optical densities is made to agree with that found by mass spectrometry at 30 "C. The agreement at higher temperatures is good. The total irreversible methanol (t.i.m.) obtained by mass spectrometry is the sum of methanol desorbed at ca. 110 "C and the methanol desorbed as formaldehyde and CO (DM) in fig. 1O(a) and (b). The t.i.m. obtained by i.r. spectroscopy is based on the optical density of the CH stretching bands after a switch to helium at the adsorption temperature. However, using the same calibration for the optical densities, the methanol remaining adsorbed after a helium purge (t.i.m.) measured by the infrared method appears to be much less than that measured by mass spectrometry.The smaller t.i.m. measured by i.r. spectroscopy than that measured by the volumetric method probably indicates that, after purging with helium, a part of the methoxy held irreversibly transforms into other kinds of hydro- carbons which were not detected by infrared spectroscopy. The distance between the curve of the total amount of methanol and that of the t.i.m. is the amount of methanol absorbed reversibly at the adsorption temperature. This decreases rapidly with temperature and becomes zero at temperatures > 130 "C. This agrees with the results in fig.9, where the shifts in the band positions of CH stretching bands after a switch from methanol to helium become smaller as theJ. S. CHUNG, R. MIRANDA AND C. 0. BENNETT 35 0 30 50 70 90 T / T Fig. 11. Amounts of three different types of methanol on MOO, as a function of temperaure. Circles, volumetric; squares, i.r.; 0 and m, total amount; 0 and 0, tirn., total amount of irreversible chemisorption left on the surface after purging with helium; @, DM, the amount desorbed as formaldehyde and CO during temperature-programmed desorption. temperature increases and are negligible at 110 "C. Therefore the reversible methanol desorbed below 110 "C in fig. 10(a) corresponds to the methanol dissociated into methoxy and hydrogen across Mo=O (form C) and reaction (4) is reversible.This type of chemisorption must have a weak bond strength between methanol and MOO, and compiicated hydrogen bonding is expected. The extent of hydrogen bonding decreases with increasing temperature, as shown by the increase in the frequency of the CH stretching bands in fig. 9 up to 110 "C. The desorption of methanol with a peak at 110 "C is probably related to the methoxy chemisorbed on acidic V,, and the methoxy which is decomposed into formaldehyde or CO is related to the methoxy on the less acidic V, and/or V, sites. The chemisorption behaviour of methanol above 150 "C is quite different from that below 110 "C because of the reduction of MOO,. This will be discussed in another paper. CONCLUSIONS This work has led to the following conclusions.(1) Chemisorption of water or methanol develops three hydroxyl bands and two water bands. All except the band at 3255 cm-l are shifted to higher frequencies upon reduction of MOO,. (2) The isolated hydroxyl band and the two water bands disappear at ca. 150 "C, while the hydrogen-bonded hydroxyl band remains up to 350 "C. ( 3 ) Most of the methanol chemisorbs dissociatively and at least three fonns of chemisorbed methanol exist : methanol dissociated into methoxy and hydrogen cross Mo=O (form C), methoxy on the oxygen vacancy of Mo=O (form A) and methoxy on the oxygen vacancy of Mo-0-Mo (form B). (4) The bond strength between methoxy and MOO, decreases in the order: form B > form A > form C. ( 5 ) Form C reversibly desorbs as methanol36 CHEMISORPTION OF H,O AND CH,OH ON MOO, below 110 "C and a part of form A is desorbed as methanol above 110 "C.The rest of form A and form B desorb as formaldehyde and CO at ca. 240 "C. (6) There exist some oxygen vacancies capable of forming methoxide even after MOO, is oxidized at 380 "C overnight. We thank the National Science Foundation for support under grant no. CPE-7091018. G. K. Boreskov, Proc. 3rd Znt. Congr. Catal. (Elsevier, Amsterdam, 1964), p. 213. N. Pernicone, F. Lazerrin, G. Liberti and G. Lanzavecchia, J. Catal., 1969, 14, 293. J. Novakova, P. Jim and V. Zavadil, J. Catal., 1970, 17, 93. N. Evmenko ind Ya. Gorokhovatskii, Kinet. Katal., 1969, 10, 1299. R. Mann and K. Hahn, Znd. Eng. Chem., Process Des. Dev., 1970, 9, 43. J. Edwards, J. Nikcolaidis, M. B. Cutlip and C. 0. Bennett, J. Catal., 1977, 50, 24. M. Niwa, M. Mimtani, M. Takahashi and Y. Murakami, J. Catal., 1981, 70, 14. C. J. Machiels and A. W. Sleight, J. Catal., 1982, 76, 238. N. Pernicone. F. Lazerrin and G. Lanzavecchia, J. Catal., 1968, 10, 83. P. Jiru, M. Krivanek, J. Novakova and B. Wichterlova, Proc. 4th Znt. Congr. Catal. (Nauka, Moscow, 1969), p. 1919. l1 M. Formenti, F. Juillet, P. Meriaudeau, S. J. Teichner and P. Vergon, J. Colloid Interface Sci., 1972, 39, 79. l2 (a) C. 0. Bennett, ACS Symp. Ser., 1981, 178, 1; (6) J. S. Chung, Ph.D. Dissertation (University of Connecticut, 1984). l3 D. Bianchi, J. L. Bernard, M. Camelot, R. Bernali-Chaoui and S. J. Teichner, Bull. Soc. Chim. Fr., l4 J . R. Gunter, J. Solid State Chem., 1972, 5, 354. l5 M. Primet, P. Pichat and M. Mathieu, J. Phys. Chem., 1971, 75, 1216. 1980, 7-8, 1-275. C. H. Rochester and S. A. Topham, J. Chem. Soc., Faraday Trans. I , 1979,75, 1073. G. Anderson and A. Magnelli, Acta Chem. Scand., 1950, 4, 793. A. Zecchia, S. Coluccia, E. Guglieminotti and G. J. Ghiotti, J. Phys. Chem., 1971, 75, 2774. le L. H. Little, Infrared Spectra of Adsorbed Species (Academic Press, New York, 1966), p. 285. 2o R. T. Jernigan and G. R. Dobson, Inorg. Chem., 1972, 11, 81. 21 P. G. Dickens, J. J. Birtill and C. J. Wright, J. Solid State Chem., 1979, 28, 185; 1979, 29, 367. 22 G. Busca and V. Lorenzelli, J. Catal., 1980, 66, 155. 23 R. G. Greenler, J. Chem. Phys., 1962, 37, 2094. 24 B. A. Morrow, J. Chem. Soc., Faraday Trans. I , 1974, 70, 1527. 25 J. C. McManus, K. Matsushita and M. J. D. Low, Can. J. Chem., 1969,47, 1077. 26 R. 0. Kagel and R. G. Greenler, J. Chem. Phys., 1968,49, 1638. 27 E. W. Thornton and P. G. Harrison, J. Chem. Soc., Faraday Trans. 1, 1975,71,2468. 28 A. A. Davydov, V. M. Shchekochikhin, P. M. Zaitsev, Ya. M. Shchekochikhin and N. P. Keier, 29 K. Miranda, J. S. Chung and C. 0. Bennett, Proc. 8th Znt. Congr. Catal. (Springer, Berlin, 1984) 30 N. Takezawa and H. Kobayashi, J. Catal., 1973, 28, 335. Kinet. Katal., 1971, 12, 694. p. 111-347. (PAPER 4/088)