Adsorption and thermal decomposition of methanol on magnesium

Adsorption of n Methanol Molecules on MgO(100) with n = 1 to 4: A Theoretical Study. Angel Horacio Rodríguez, María Marta Branda, and Norberto Jorge...
0 downloads 0 Views 794KB Size
Langmuir 1985,1, 593-599

593

Adsorption and Thermal Decomposition of Methanol on Magnesium Oxide. 13C NMR Studies Septimus H. C. Liang and Ian D. Gay* Department of Chemistry, Simon Fraser University, Burnaby, British Columbia, Canada V5A 1S6 Received July 25, 1984. In Final Form: March 13, 1985

NMR and adsorption studies have been carried out for methanol on three different preparations of MgO. High-resolution 13Cspectra have been obtained, together with 'H relaxation data, at several coverages. These data indicate the only significant chemisorbed species to be methoxide, at room temperature. Relatively minor differences are observed between MgO from different sources. The adsorbed layer is stable upon heating up to 300 "C. At higher temperatures, decomposition sets in, yielding a complex mixture of gas-phase products. A single chemisorbed species formed by high-temperaturedecomposition is found to be bicarbonate or a hydrogen-bonded carbonate. Introduction The application of conventional 13C NMR spectroscopy to surface chemistry and catalysis has become nearly routine. Investigation of surface acidity and acid sites1v2 and monitoring of surface reactions3 have been achieved with moderate success. However, these results are based on the premise that the adsorbed molecules undergo isotropic rotation at a rate of at least 106 Hz. This is because the C-H dipolar couplings are of the order of lo4 Hz and correspondingly rapid motion is required to average this to a small value, permitting high-resolution spectroscopy. It is obvious that this conventional 13C NMR technique will not be applicable in the case of strongly chemisorbed species which exhibit a more rigid structure and limited freedom of motion. Recent developments in high-resolution solid-state NMR technique4 have shown that this dipolar coupling can be suppressed by high-power irradiation of the protons and high-resolution 13C spectra obtained. By combination of this with cross-pol"tion,H the signal-to-noise ratio can also be increased. The application of this high-power decoupling plus cross-polarization (CP) technique in surface chemistry and polymers is still relatively The main difficulty in these experiments is that the 13C chemical shift anisotropies can be large; thus overlapping powder patterns result for species containing more than one carbon. This will also degrade the signal-to-noiseratio, although the use of selectively enriched 13C compoundsgCcould solve this problem. This report is a continuation of previous work from this laboratory on chemisorption of methanol on MgO" utilizing the solid-state 13CNMR technique. In addition to data for static samples, we present some initial results (1) Gay, 1. D.; Liang, S. J. Catal. 1976,44,306. Gay, I. D. Ibid. 1977, 43,430. Liang, S.; Gay, 1. D. Ibid. 1980, 66, 294. (2) Bernstein, T.; Kitaev, L.;Michel, D.; Pfeifer, H. J. Chem. SOC., Faraday Trans 1 1982, 78, 761. (31 Krii.J. F.: Gay. I. D. J.Phvs. Chem. 1976.80.2951. Ali.. 1. T.:. Gay. . . _. I. D.'J.Catal. 1980,-62, 341. (4) Pines, A.; Gibby, M. G.; Waugh, J. S. J. Chem. Phys. 1973,59,569. (5) Hartmann, S. R.; Hahn, E. L. Phys. Reo. 1962, 128, 2042. (6) Lurk, F. M.; Slichter, C. P. Phys. Reu. A 1964, 33, 1108. (7) Kaplan, S.; Resing, H. A.; Waugh, J. S. J. Chem. Phys. 1973,59, 5681. (8) Chang, J. J.; Pines, A.; Fripiat, J. J.; Resing, H. A. Surf. Sci. 1975, 47, 661. (9) (a) Stejskal, E. 0.; Schaefer, J.; Henis, 3. M. S.; Tripodi, M. K. J. Chem. Phys. 1974,61,2351. (b) Sefcik, M. D.; Schaefer, J.; Stejskal, S. 0.A.C.S. Symp. Ser. 1976,34,109. (c) Sefcik, M. D. J . Am. Chem. SOC. 1979,101, 2164.

(10)Duncan,T. M.; Yates, J. T., Jr.; Vaughan, R. W. J.Chem. Phys. 1979, 71, 3129. (11) Gay, I. D.

J. Phys. Chem. 1980,84, 3230. 0743-7463/85/2401-0593$01.50/0

using a recently developed magic-angle spinneP which permits study of sealed samples. We discuss here two aspects of heterogeneous catalysts: difference in catalytic activity due to the different methods of preparation and the extent of catalytic conversion of reactant at different temperatures. The system under investigation is methanol chemisorbed on magnesium oxide. The interaction of alcohols with magnesium oxide has been widely explored.lS2l The general consensus is that MgO is a very selective dehydrogenation catalyst. However, dehydration activity is observed when the catalyst is contaminated with C02,which may result from an incomplete decomposition of a carbonate precursor.22 The conventional methods of preparing MgO are from the thermal decomposition of the hydroxide, carbonate, or oxalate; thus variations in contamination may occur. Pretreatment of MgO with H2 or O2 also affects the catalytic selectivity as shown by Davis.18 Comparisons of different methods of MgO preparation have been detailed.1572"26 It is agreed that MgO prepared by different routes has different surface morphologies and thus the catalytic activity and selectivity vary. Other factors, such as the temperature, the length of treatment time, and the atmosphere in which the substance is heated also have great effects on the activity of the resulting p r o d u ~ t . ~ ' ~ ~ ~ We investigate these effects by studying adsorption isotherms, proton relaxation parameters, and the 13C CP NMR spectra of adsorbed species. (12) Gay, I. D. J.Magn. Reson. 1984, 58, 413. (13) Kagel, R. 0.; Greenler, R. G. J. Chem. Phys. 1968, 49, 1638. (14) Tench, A. J.; Giles, D.; Kibblewhite, J. F. J. Trans. Faraday SOC. 1971, 67, 854. (15) Hattori, H.; Shimazu, K.; Yoshi, N.; Tanabe, K. Bull. Chem. SOC. Jpn. 1976, 49, 969. (16) Foyt, D. C.; White, J. M. J. Catal. 1977, 47, 260. (17) Parrott, S. L.; Rogers, J. W., Jr.; White, J. M. Appl. Surf. Sci. 1978, 1, 443. (18) Davis, B. H. J. Chem. SOC.,Faraday Trans. 1 1980, 76, 92. (19) Takezawa, N.; Hanamaki, C.; Kobayashi, H. J. Res. Inst. Catal., Hokkaido Uniu. 1980,28, 347. (20) Takezawa, N.; Kobayashi, H. J. Catal. 1982, 73, 120. (21) Takezawa, N.; Hanamaki, C.; Kobayashi, H. J. Catal. 1975, 38, 101. (22) Krylov, 0. V.; Fokiva, E. A. Kinet. Katal. 1961, 1, 421. (23) Coluccia, S.; Tench, A. J.; Segall, R. L. J. Chem. SOC.,Faraday Trans. 1 1979, 75, 1769. (24) Iizuka, T.; Saito, M.; Tanabe, K. J. Res. Inst. Catal., Hokkaido Uniu. 1980, 28, 189. (25) Matauda, T.; Tanabe, J.; Hayashi, N.; Sasaki, Y.; Miura, H.; Sugiyama, K. Bull. Chem. SOC. Jpn. 1982,55,990. (26) Mivahara. K.: Murata.. Y.:. Tovoshima,. 1.:. Tanaka.. Y.:. Yokovama, . T. C a d . 1981; 68,186. (27) Anderson, P. J.; Morgan, P. L. Trans. Faraday SOC. 1964,60,930. (28) Razouk,R. I.; Mikhail, R. S.; Ragai,J. J. Appl. Chem. Biotechnol. 1973, 23, 51.

a.

~

0 1985 American Chemical Society

594 Langmuir, Vol. 1, No. 5, 1985

Liang and Gay

Studies of the catalytic conversion products of methanol on MgO have included infrared spectr~scopy,'~J~ mass spectrometry, and flow systems.15 This report is the first by 13C NMR spectroscopy. GC/MS studies only reveal the gaseous and the easily desorbed products, while giving no information about the strongly bound species on the surface. Infrared studies suffer from assignment problems and the problem of assuming an invariance of extinction coefficients upon adsorption. Despite its shortcomings, infrared spectroscopy has revealed several adsorbed species formed from methanol on MgO. These include physisorbed methanol plus several chemisorbed methoxide ~ p e c i e s ' ~atJ ~room temperature. These methoxide species persist on the surface upon heating until 300 OC, at which temperature formate is observed on the surface with the evolution of CO. At even higher temperatures, surface carbonate is observed. This possibly results from the reaction of CO, (from decomposing surface formate species) with the surface. We intend to use 13C NMR spectroscopy to supplement existing information in drawing a clear picture of the conversion of methanol on MgO.

V

V

L 0

Materials Three kinds of MgO were prepared. MgO(1) was made from MgC03.3H20 prepared by the coprecipitation of solutions of 3.6 M Mg(N03)z-6H20and 0.8 M NaHCO3 The carbonate was then washed, dried, and decomposed in vacuum at 500 OC in a thin bed. The MgO(1) so prepared has a surface area of 215 m2/g and a bulk density of 0.3 g/cm3. MgO(2) was prepared from the decomposition of M g (OH), (reagent grade from Matheson, Coleman and Bell) in vacuum at 300 OC for 20 h in a thin bed. MgO(2) has a surface area of 220 m2/g and a bulk density of 0.6 g/cm3. MgO(3) was prepared from MgC20,.2H20, obtained from the coprecipitation of solutions of 0.5 M (NHJ2Cz04-H20and 3.5 M Mg(N0J2.6H20. The oxalate was dried and then decomposed a t 300 "C in vacuum, after which it was further heated at 400 OC in air. This further calcination was deemed necessary when one batch of MgO(3) was found contaminated with some undecomposed impurities. MgO(3) has a surface area of 200 m2/g and a bulk density of 0.6 g/cm3. All surface area determinations were done by the BET method using N2 a t 77 K. The methanol used in these experiments was of spectroscopic grade from Fisher Scientific. It was dried over pretreated molecular sieve type 4A and degassed before use. The COz used was from Canadian Liquid Air Ltd. I3C-enriched methanol and COP were obtained from MSD Isotopes and were vacuum distilled before use. Experimental Section Samples of the different kinds of MgO for static studies were loaded into 12-mm-0.d. NMR tubes to a height of about 2 cm. These were then slowly heated to 500 "C in vacuum for 22 h for outgassing and removal of surface water. Measured amounts of the desired adsorbate were then allowed to adsorb onto the oxide from the gas phase at room temperature. The samples were then sealed off and allowed to stand for several days at room temperature before NMR measurements. Some samples were rerun after a period of 2-3 months to check for reproducibility,which was always found to be excellent. Samples for magic-angle spinning were of about 1.5-cm depth in a 5-mm-0.d. tube and were treated in the same manner. To investigate the conversion products of methanol on MgO, another series of experiments with the same samples was undertaken. A t least three samples having different coverages of methanol on MgO from each preparation were raised to succes-

A

A '

l b i 0 & 4 b & 6 0

>

Pressure (mmHg)

Figure 1. Adsorption isotherms of methanol on all preparations of MgO at room temperature. Samples degassed at 500 "C. (0) MgO(1); (A)MgO(2); (VI MgO(3).

sively higher temperatures. This was done by placing the sealed NMR sample tubes inside a tube furnace at the selected temperatures (170-180,300-320, and 485-500"C) for 3 h. They were then cooled and left for several days before NMR experiments. All the 13CCP NMR spectra were measured at 15.08 MHz on a TT-14 spectrometer which was modified for solid-state NMR studies. Single-contact Hartmann-Hahn cross-polarizationwas used with radiofrequency field strengths (yHl/2r) ranging from 40 to 70 kHz. "Spin-temperaturereversal"" was used on alternate polarizations. Proton spin-lattice relaxation times ( T l ~were ) measured for all samples by the usual 180"-r-90' sequence. These data were used to determine the optimum recycle delay (=1.25TlH)between cross-polarizations. Cross-polarization times (TcH)were measured for selected samples of methanol-MgO to determine the optimum contact time and as a qualitative estimate of the C-H dipolar strength in the system. This was done by varying the contact time and observing the resulting 13C FID signals. Proton spin-lattice relaxation time in the rotating frame (Tip) was also measured for the selected samples on which TIHwas measured. Only those samples in which 60% 13C-enriched methanol was used were chosen, so that strong 13C signals could be observed with the minimum number of repetitions. The proton TlPwas measured by varying the lengths of time between proton spm locking and the '3c contact pulse and observing the amplitude of the resulting 13C signal. This yields the TlPvalue for those protons involved in cross-polarization with 13Cand excludes any isolated surface protons not strongly coupled to those in the adsorbed methanol.

Results and Discussion 1. Methanol and CO, Absorption Isotherms. In order to observe the differences in the surface characteristics of the three kinds of MgO, adsorption isotherms were done for methanol and CO, at room temperature, as shown in Figures 1-3. For Figures 1 and 2, the MgO's were degassed at 500 O C but at 300 "C for Figure 3. Essentially, all isotherms show a type I1 adsorption characteristic. The methanol adsorption isotherms are (29) Stejjskal, E. 0.;Schaefer, J. J. Magn. Reson. 1976, 18, 560. (30)Schaefer, J.; Stejskal, E. 0.; Buchdahl, R. Macromolecules 1977,

10,384.

Langmuir, Vol. 1, No. 5, 1985 595

Adsorption and Thermal Decomposition of Methanol

kl

C

E

0

3 0

V

V

.4

A

40

20

0



6

0

Pressure(mmHg)

lco

80



.C

Figure 2. Adsorption isotherms of COz on all preparations of MgO at room temperature. Samples degassed at 500 “C. (0) MgO(1); (A)MgO(2); (v)MgO(3).

I 0

a 0

O

A

A

A A

I

I

I

Ppm

100

J

0

Figure 4. 13C CP NMR spectra of 90% 13C-enrichedCOz on all MgO preparations. RF field = 40 kHz;contact time = 2 ms; delay time is 1 s; loo00 contacts. Spectra are not normalized. (a) 2.70 pmol/m2 of COz on MgO(1); (b) 2.09 pmol/m2of COz on MgO(2); (c) 2.98 pmol/m2 of COz on MgO(3); 95000 contacts.

24

0

I 200

0

0

I

n

Figure 3. Adsorption isotherms of COz on all preparations of MgO at room temperature. Samples degassed at 300 “C. (0)

spectrum is observed. A single peak is found at the same shift, within our experimental error, f0.2 ppm. As might be expeded, the 13CTl is rather long, about 20 s. No shifts or new peaks appear for interpulse delays up to 60 s. Thus, although an undetected species of arbitrarily long T1 can never be excluded, it seems likely that only a single surface species is present. For bicarbonate formation on the surface, we propose the following mechanism: co2

OII >

0I

MgO(1); (A)MgO(2); (VI MgO(3).

basically the same for all preparations of MgO. There is a slight deviation for MgO(3) whose isotherm lies slightly higher than the others. The C02 isotherms (Figures 2 and 3) show rather more variation with preparation and degassing conditions. For both methanol and COP,a substantial chemisorbed population is achieved at lower pressures than we were able to use and a more loosely bound population is observed which increases with pressure. Qualitatively, we can conclude that the chemisorption level on MgO is 6-7 Fmol/m2 for methanol and 1-2 pmol/m2 for COP. Since COPis slightly acidic, presumably it will absorb on the basic sites on the MgO surface. Thus, if methanol adsorbs on the same sites as C02, then only onequarter of these sites are basic. However, this estimate of basic sites is 10 times bigger than that reported by indicator t i t r a t i ~ n . To ~ ~ investigate further into this aspect, we measure the 13C NMR spectra of 90% 13C-enriched COP adsorbed on all MgO’s. The static 13C CP spectra are shown in Figure 4. Magic-angle spinning (Figure 13) for C02 on MgO(2) reveals a single peak at an isotropic shift of 161.6 ppm. This is certainly consistent with a bicarbonate structure. We find, for example, NHlHC03 at 162.5 ppm and BaC03 at 169.5 ppm. The observed anisotropies are also consistent with a bicarbonate structure. When the spectrum of adsorbed C02 is excited by 90’ pulses, instead of cross-polarization, the ’same MAS (31)Tanabe, K. In ‘Solid Acids and Bases”; Academic Press: New York, 1970;p 50.

This simple mechanism would indicate that the C02 adsorption occurs with participation of surface-OH groups. This would partly explain why the equilibrium adsorption level of methanol is 4 times bigger than that of C02. Since there are only 3 pmol/m2 of surface-OH groups on Mg032 under our degassing conditions, presumably they will all be taken up in the COz adsorption. There are about 18 pmol/m2 of Mg2+on the (100) plane of MgO surface;32 these will give ample sites where the methanol could adsorb. Therefore, we suggest that both methanol and C02 adsorb onto the Mg2+sites on the MgO surface but C02 requires an additional neighboring surface-OH group. As the degassing temperature is lowered to 300 “C, we would expect the surface-OH concentration to increase, thus the C02 adsorption on MgO’s at room temperature should increase. However, this is the case only for MgO(1) and 42), while MgO(3) shows a decrease. This might indicate nearly complete coverage of the surface by OH (32)Anderson, P. J.; Horlock, R. F.; Oliver, J. F. Trans. Faraday SOC. 1965,61, 2754. Anderson, P. J. In ‘Structure of Metallic Catalysts”; Academic Press: New York, 1975;p 63. (33)Heater, R. K.; Cross, V. R.; Ackerman, J. L.; Waugh, J. S. J. Chem. Phys. 1975,63,3606. (34)Muller, L.;Kumar, A.; Baumann, T.; Ernst, R. Phys. Reu. Lett. 1974,32,1402. (35)Demco, D. E.;Tegenfeldt, J.; Waugh, J. S. Phys. Rev. B 1975,11, 4133. (36)Opella, S. J.; Frey, M. H. J. Am. Chem. SOC.1979,101, 5854. (37)Alemany, L.B.; Grant, D. M.; Alger, T. D.; Pugmire, R. J. J. Am. Chem. SOC.1983,105,6697.

Liang and Gay

596 Langmuir, Vol. 1, No. 5, 1985 Table I. TIHof All Methanol-MgO Systems a n d Selected Samples Treated at Higher Temperatures coverage, pmol/m2 MgO(1) 2.0Y 3.63" 4.31" 5.92 7.04 8.42 MgO(2) 1.22O 2.79" 4.550 5.70 7.73 8.64 MgO(3) 2.99" 4.11" 4.980 6.21 7.07 8.18

TlH, (room temp) 0.55 0.31 0.50 0.15 0.16 0.14 0.39 0.19 0.07 0.04 0.04 0.05 0.82 0.83 0.83 0.68 0.62 0.51

TlH, 8 (170 "C)

T ~ Hs, (320 "C)

(500 "C)

0.56 0.21 0.25 0.20

0.51 0.24 0.19 0.21

0.38 0.10 0.13 0.02

0.12 0.09 0.08

0.18 0.08 0.07

0.07 0.08 0.11

T ~ H8,

"60% 13C-enriched methanol was used. Table 11. TIE,T,,, a n d TcE for Some Selected Methanol-MgO Systems coverage, pmol/m2 2.09 MgOU 3.63 MgO(2) 1.22 2.79 MgO(3) 2.99 4.98

s 0.55 0.31 0.39 0.19 0.82 0.83

TIH,

TI,, ms 12,66 13,380 47 21 25 11

TCH,ms 0.11 0.13 0.12 0.11 0.14 0.10

for sample (3). This would inhibit C02 adsorption if the latter requires adjacent Mg and MgOH sites as proposed above. 2. Measurements of Relaxation Times and CrossPolarization Times. Proton spin-lattice relaxation times (TIH)were measured for all methanol-MgO samples before I3C CP NMR spectra were taken. These data are important, as mentioned before, to determine the optimum repetition period, which is about 1.25T1H T1H were also measured for some selected samples with which higher temperature studies were done, as shown in Table I. In Table 11, TIH,Tlp,and TCH are shown for some selected methanol-MgO systems in which 60% ,%-enriched methanol was used. These were measured at 40-kHz field strength for 'H and 13C. Proton Tlp in general is much shorter than the T1H for the same system, except for one case in MgO(1). For MgO(l), the Tlpdata show nonexponential relaxation which can be resolved into a s u m of two exponentials whose decay times are given in Table 11. Proton relaxation in these systems could reasonably be attributed to proton-proton dipolar coupling and/or to interaction with paramagnetic impurities in the surface. Other relaxation mechanisms do not seem plausible. A proposed relaxation mechanism must account for the experimental facts that TlPis always much shorter than T, and that both increase with coverage, the latter more markedly. The shortness of Tlpsuggests the presence of interactions of relatively long correlation time, since the spin-locking field strength was 40 kHz. This suggests a low-frequency motional process in the adsorbed layer. Relaxation due to paramagnetic impurities is usually characterized by a correlation time of the order of the electron The most likely impurities in MgO are (38) Pfeiffer, H. NMR: Basic Princ. Prog. 1972, 7, 53.

d L

I

loo

ppm

I

I

0

I

Figure 5. 13C CP NMR of 2.79 pmol/m2 of 60% %enriched methanol on MgO(2) at different contact times. RF field = 40 kHz; repetition rate = 2.5 8-l; 512 contacts. All spectra are normalized. Contact times (ps): (a) 30; (b) 100, (c) 150; (d) 200, (e) 250; (f') 500; (9) 750; (h) 1000; (i) 1250; 6)1500; (k) 1750; (1) 2000.

first-row transition ions. These typically have unmeasurably short electron Ti's a t room temperature but extrapolation of a typical low-temperature m e a ~ u r e m e n t ~ ~ yields values of the order of lo+' s for dilute Ni2+in MgO. Such a correlation time is too short to give a proton T1, that is substantially less than T,. In an attempt to further elucidate the situation, we have studied some MgO samples which were doped with 10 ppm of Ni2+. This was done by impregnating MgO with a solution containing the desired amount of Ni2+, drying slowly, and heating to 500 O C . It is likely that Ni2+will not be uniformly incorporated at this temperature. Since the present oxide has =15% of its atoms in the surface layer, we may assumed the surface Ni2+concentration to lie between 10 and 70 ppm. On such a doped oxide, we find that the proton T1 is substantially reduced, typically by a factor of 2.5-3.0, at all coverages. The Tlprelaxation becomes exponential, with a relaxation time close to those found for the undoped oxide (the shorter of the two, for the originally nonexponential cases). These results suggest the following: the increase of l/Tl by an easentially constant fador suggests Tl relaxation may well be dominated by paramagnetic relaxation, propagated by spin diffusion in the adsorbed layer. The rate of relaxation should be proportional to the impurity concentration for such a whereas a constant additive contribution might be expected if relaxation in the undoped samples had been dominated by proton dipoledipole interactions. (39) Lopez, P.; Peecia, J. 'Magnetic Resonance and Related Phenomena"; North-Holland Publishing Co.: Amsterdam, 1975; Proceedings of the 18th Ampere Congress, p 413. (40)Abragam, A. 'The Principles of Nuclear Magnetism"; Oxford University Press: Oxford, UK, 1961; Chapter 9.

Langmuir, Vol. 1, No. 5, 1985 597

Adsorption and Thermal Decomposition of Methanol

A

A

Figure 6. Plot of '% FID signal VB. contact times using data from Figure 5.

a '00

b

~~~

'00

ppm

0

Figure 8. 13Cspectra of methanol adsorbed on MgO(2) at different coverages. RF field = 70 kHz; contact period = 1.5 ms except (c), (d), and (e) which are 2 ms. (a) 1.22 pmol/m2,*repetition rate = 2.56 s-', 6000 contacts, (b) 2.79 pmol/m2,*reptition rate = 6.25 s-l, 5400 contacts; (c) 4.55 pmol/m2,*repetition rate = 14.3 E-', 350 contacts; (d) 5.70pmol/m2,repetition rate = 14.3 s-l, 39600 contacts; (e) 7.73 pmol/m2,repetition rate = 14.3 s-l, 86000 contacts, (fJ8.64 mol/m2,repetition rate = 12.5s-l, 430000 contacts. * indicates 60% 'V-enriched methanol used. All spectra are not normalized for the different number of contacts.

g & ~

ppm

0

Figure 7. 13Cspectra of methanol adsorbed on MgO(1) at different coverages. RF field = 70 IrHz;contact period = 2 ms except (a) which is 1.5 ms. (a) 2.09 pmol/m2,*repetition rate = 1.54 ,'-E 5000 contacts; (b) 3.63 pmol m2,*repetition rate = 2.94 s-', 6000 contacts; (c) 4.31 pmol/m{* repetition rate = 1.82 s-l, 2000 contacts; (d) 5.92 pmol/m2, repetition rate = 5.56 s-l, 285000 contacts; (e) 7.04 pmol/m2, repetition rate = 5.56 s-l, 66000 contacts; (f) 8.42 pmol/m2, repetition rate = 6.25 s-l, 90000 contacts. * indicates 60% '%-enriched methanol used. Spectra are not normalized for different number of contacts. However, the fact that T1, is substantially shorter than Tl and little affected by doping suggests a different mechanism here. The most reasonable explanation is dipole-dipole interactions modulated by low-frequency molecular motions in the adsorbed layer, which would not contribute effectively to Tl. The increase in l/Tlp with adsorbate coverage is probably due to an increasing fraction of the adlayer bound relatively weakly as the surface fills. This would lead to greater average intensity of lowfrequency motion and is consistent with the changes in '% spectrum, discussed below.

The cross-polarization times are short and relatively consistent for all samples. Clearly, this indicates a substantial static component of the intramolecular lH-13C dipolar coupling. Figure 5 shows the 13C CP NMR spectra of a typical methanol-Mg0 system as the contact times are varied. At very short contact time, the 13Cresonance is "split" in the middle showing that some of the adsorbed species for which the methyl rotation axis lies along the magic angle do not receive adequate croas-polarization.N Contact times are varied up to 2 m, at which time we usually do our 13C CP experiments. No dipolar oscillation^^^^^-^^ were observed at short contact times. A plot of intensity vs. contact time is shown in Figure 6. Generally, the contact times used in our experiments (1.5 or 2 ma) are much greater than the longest TCHin all the methanol-Mg0 systems, so that the carbons are crosspolarized to an equal extent. Also, the contact times is shorter than the Tlp, so that the spin-locked proton magnetization does not decay significantly within the contact period. This permits good quantitative 13C CP analysis of the methanol-MgO systems. 3. 13CSpectra of Adsorbed Methanol. Figures 7-9 show spectra of static samples, for various coverages, on each preparation of MgO. In general, for all cases, a broad powder pattern is observed in the region expected for a methoxide species. At low coverages, we used 13Cenriched methanol, which caused the edges of the pattern to be less sharp than previously reported." This is because the 13C-13C dipolar coupling amounts to a few hundred hertz for the enriched samples and gives a line width of this magnitude, convolved with the powder pattern.

598 Langmuir, Vol. 1, No. 5, 1985

Liang and Gay

il

I I 100

ppm

0

I 200

I

I 100 ppm

\ I

I

0

Figure 9. 13Cspectra of methanol adsorbed on MgO(3) at different coverages. RF field = 40 kHz; contact period = 2 ms; repetition rate = 3.33 s-l. (a) 2.99 pmol/m2,*14400 contacts; (b) 4.11 Nmol/m2,*8192 contacts; (c) 4.98 pmol/m2,*8192 contacts; (d) 6.21 pmol/m2, 429000 contacts; (e) 7.07 pmol/m2,435000 contacts; (f) 8.18 pmol/m2, 244000 contacts. * indicates 60% '%-enriched methanol used. All spectra are not normalized for the different number of contacts.

Figure 10. 13C spectra of 4.31 pmol/m2of methanol (60% 13Cenriched) on MgO(1) treated at different temperatures. RF field = 70 kHz; contact period = 2 ms except (a) which is 0.5 ms. (a) Room temperature, repetition rate = 1.82 s-l, 2000 contacts; (b) 170 "C (3 h), repetition rate = 1.43 s-', 7600 contacts; (c) 320 "C (3 h), repetition rate = 1.72 s-l, 5900 contacts; (d) 500 "C (3 h), repetition rate = 4.17 s-l, 38800 contacts. Spectra are not normalized for different number of contacts.

On all samples, a narrower peak appears near 50 ppm at high coverages (cf. Figure 90. This presumably represents the appearance of methanol with considerable rotational freedom once the strong chemisorption sites have been filled. Magic-angle spectra have been obtained for methanol on sample (2) (Figure 13). These show a single peak having a width of 80 Hz. The position of this peak shifts from 49.3 ppm at 1.3 pmol/m2 to 50.3 ppm at 7.6 pmol/m2. All of these data are consistent with only two chemisorbed species on MgO at room temperature-a relatively rigid methoxide and a more mobile species which appears at high coverages. The fairly minor differences among different preparations can be accounted for by varying populations of the two species and by motional differences. A change in number of chemisorption sites with preparation will of course vary the relative proportions of species at the higher coverages. Varying degrees of motion will partially average the powder patterns from the adsorbed layer, leading to the small differences observed. Such motional variation may well result from packing differences in the adsorbed layer if different preparations yield different local geometries for neighboring adsorption sites. 4. '3c NMR Measurements of Adsorbed Methanol Treated at Elevated Temperatures. The 13CCP spectra of the methanol-MgO systems at room temperature and after heating (170,320,500 "C) are shown in Figures 10-12. The spectra shown are those in which 60% 13C-enriched methanol was used. Essentially, the spectra of the adsorbed species remained the same until 300 OC. One striking feature is the gradual sharpening of the methoxide powder pattern as the temperature is raised. This is more pronounced for MgO(2)

and MgO(3) than for MgO(1). This is perhaps due to the "recrystallization" of the adsorbed layer on heating and recooling. Possbly a more uniform layer is achieved by thermal crossing of activation barriers. A similar effect is observed in the magic-angle spectra on MgO(2). The line widths decrease from 80 to 40 Hz on heating to 300 "C and recooling as seen in Figure 13. The spectra of all methanol-MgO systems treated at 500 "C show a low-field resonance with the complete disappearance of the methoxide species. From the position and anisotropy (approximately 100 ppm) of this resonance, it clearly contains sp2carbon and is presumably a carbonate or formate species. The magic-angle spectrum shows a single line at 167.4 ppm, having a width of 110 Hz. With the MAS technique we were able to perform a delayed decoupling experiment.36p37The resonance intensity of this species is reduced to 50% by a decoupler delay of 200 ps. This strongly suggests a proton two bonds distant from carbon.37 Given the chemical shift, bicarbonate or hydrogen-bondedcarbonate is the obvious species. It is most unlikely that this is a formate species. A decoupler delay experiment on ammonium formate shows decay to 50% in 25 and to 5% in 60 ps. For the MgO surface species, the intensity remains >95% after a 60-ps delay. These results clearly exclude the possibility of a surface formate. Since crystalline formates have an anisotropy of about 130 ppm41 and our surface species has a powder pattern 100 ppm wide, the latter clearly cannot be a formate in rapid, nearly isotropic motion. If a formate, it must, at most, have a restricted anisotropic motion, thus C-H di(41) Ackermann, J. C.; Tegenfeldt, J.; Waugh, J. S. J.Am. Chem. SOC. 1974, 96,6843.

Adsorption and Thermal Decomposition of Methanol

Langmuir, Vol. 1, No. 5, 1985 599

I

I

I

200

I

I

ppm

100

I

1

0

Figure 11. 13Cspectra of 3.55 Mmol/m2of methanol (60% 13Cenriched) on MgO(2) at different temperatres. RF field = 70 IrHz; contact period = 1.5 ms. (a) Room temperature, repetition rate = 14.3 8 ,350 contacts; (b) 170 "C (3 h), repetition rate = 6.67 s-l,6000contacts; (c) 330 "C (3 h), repetition rate = 10 s-l, 3000 contacts; (d) 520 "C (3 h), repetition rate = 10 s-l, 26000 contacts. Spectra are not normalized for different number of contacts.

polar coupling must be largely static, and rapid dephasing should be observed, contrary to experiment. We find that both ammonium bicarbonate and the hydrogen-bonded carbonate MgC03.3H20show dipolar dephasing behavior similar to our surface species. Thus, we cannot distinguish between these possibilities. Conventional single-pulse 13CNMR spectroscopy done on some selected samples revealed a wide range of alkanes and alkenes and C02 on MgO when the methanol-Mg0 was heated to 500 "C for 3 h. The resolvable resonances, relative to Me4Si, are -10.6 (CH,), 4.2 (C2H6),11.6, 15.4, 23.6, 110.5,117.5, 120.6, 133.8,142.4,158.3,169.1,and 183.9 (CO). These were not observed in our CP experiments, indicating them to be highly mobile species, probably no more than physisorbed in any case. Gas chromatography and mass spectrometer analysis of the gaseous products (or easily desorbed surface species) reveal hydrocarbons ranging from C1 to C4 (both saturated and unsaturated) and large amounts of H2 and CO.

Conclusion We have demonstrated here that I3C CP NMR is suitable for surface studies involving strongly chemisorbed species (as methoxide) and species having restricted motion. We are able to detect a bicarbonate or H-bonded carbonate species formed from the methoxide species when the sample was heated to 500 "C. We do not find any evidence for formate or methyl carbonate, as proposed by other workers.13 The success of this study lies in the fact that methanol-MgO systems have short cross-polarization times and relatively long proton spin-lattice relaxation times in the rotating frame. Presently, we are extending

I

I 200

1

100

I

I

0

ppm

Figure 12. 13Cspectra of 4.11 Mmol/m2of methanol (60% 13Cenriched) on MgO(3) at different temperatures. RF field = 40 kHz except (a) which is 70 kHz, contact period = 2 ms. (a) Room temerature, repetition rate = 3.33 s-l, 8192 contacts; (b) 180 OC (3 h), repetition rate = 0.83 s-l, 2347 contacts; (c) 320 "C (3 h), repetition rate = 0.83 s-l, 78357 contacts; (d) 500 "C (3 h), fepetition rate = 0.83 s-l, 103OOO contacts. Spectraare not normahzed for different number of contacts.

I I

A I

200

d D

I

I

100

I

1

0

ppm

Figure 13. MAS-CP 13Cspectra: (A) C02(90% I3C enriched) on MgO(2); (B) methanol (60% 13C enriched) on MgO(2); (C) sample (B) after heating to 300 "C; (D)sample (B) after heating to 500 "C.

our studies to the chemisorption of ethanol and isopropyl alcohol on MgO's and chemisorption of alcohols on other oxides. These results will be published shortly. The preliminary results reported here with magic-angle spinning show that this technique will be of great value in elucidating the surface chemistry of more complicated systems.

Acknowledgment. This work was supported by an operating grant from the Natural Science and Engineering Reserach Council of Canada. Registry No. MeOH, 67-56-1; MgO, 1309-48-4.