636
J. Phys. Chem. 1985,89, 636-641 formate-d) is a likely mechanism for reaction of certain neopentyl, butyl, and tert-butyl esters.l2 Our results for ethyl-2,2,2-d3 formate (and larger esters) point to the importance of reactivity at the carbonyl functional group in the intramolecular rearrangement chemistry of ester cation radicals in the solid state. The results for the various methyl formate species indicate the sensitivity of the choice of reaction pathways to deuteration at a specific site. Finally, our results clearly show that the ester cation radicals are unstable and highly reactive. They undergo a remarkable diversity of solid-state reactions, including complexation with the matrix, intramolecular and intermolecular H (or proton) transfer, fragmentation, and intramolecular alkyl attack at the carbonyl.
to the carbonyl oxygen occurs, analogous to reactions 4 and 5, respectively.
Relation to Larger Esters and Conclusions In recent work with a number of esters we," Iwasaki,8 and SymonsIo have separately shown that larger ester radical cations tend to undergo fragmentation reactions such as the McLafferty rearrangement at low temperatures. For example, fragmentation of tert-butyl acetate cation to form isobutene cation and presumably acetic acid was reported to occur at 4 K by Iwasaki and at 77 K by Symons and ourselves. From the results of our own work and that of others, we now believe that at this time the only isolated ester cations known to be stable at 77 K are those of neopentyl esters of formic, acetic, and propionic acids and in these cases the spin density is localized mainly on the neopentyl alkyl group.]' The results found in this work have bearing on the chemistry of larger ester cation radicals. Very recent work in our laboratory indicates that alkyl attack on the carbonyl functional group, with concomitant C-0 or C-C bond scission (as found for ethyl-2,2,2-d3
Acknowledgment. We thank Ffrancon Williams for helpful and encouraging comments. Acknowledgment is made to the Office of Health and Environmental Research of the U.S. Department of Energy and to the donors of the Petroleum Research Fund, administered by the American Chemical Society, for support of this research.
ESR and Electron Spin Echo Modulation Spectroscopic Studies of Molybdenum-Adsorbate Interactions on Supported Mo/SiO, M. Narayana, R. Y. Zhan,+ and Larry Kevan* Department of Chemistry, University of Houston, Houston, Texas 77004 (Received: June 19, 1984)
Interactions of Mo5+on Mo/Si02 prepared by impregnation and hydrogen reduction at high temperatures have been studied with electron spin resonance and electron spin echo modulation (ESEM) methods. Two Mo5+species denoted as Mo(A) and Mo(B) are observed. Mo(A) with gll = 1.865 disappears on adsorption of H20, CH30H, NH3, and CH$N and a new species, Mo(C), is formed. Mo(A) seems to be- the prime site for molecular adsorption. Analyses of ESEM data for samples with adsorbed D20, CD30H, and CH30D indicate coordination of Mo5+in Mo(C) with one hydroxyl group or with one methanol molecule with an Mo-O distance of 0.22 nm. One farther noncoordinated water and methanol are also indicated by the ESEM data. For the samples with adsorbed NH3 or ND3, strong Mo-N interactions are seen with significantquadrupole effects.
Introduction Supported molybdenum catalysts such as Mo/Si02 and Mo/A1203 have received considerable attention in recent years because of their importance in organic molecule oxidation, hydrodesulfurization, and coal hydrogenation processes. The paramagnetic species observable in these catalysts upon reduction and/or other treatments have been the subject of numerous electron spin resonance (ESR) studies '-I7 and have been interpreted as Mo5+ in square-pyramidal (C&), octahedral, or distorted tetrahedral configurations. There is also considerable difference of opinion as to what are the catalytically active sites and the sites of formation of the superoxide anion 0,. Hall and co-workers'*'2 suggested that the molybdenum ion in bismuth and other molybdates as well as on supported catalysts is in tetrahedral arrangement before reduction and stays in the same configuration with a hydroxyl ion replacing one of the ligand oxygens on reduction. Che et al.7389'3and other worker^^^'^-^^ interpreted the Mo5+to be formed by loss of one oxygen from Mo6+06octahedra on reduction resulting in Mo5+05square pyramidal geometries. Kazansky and co-workers14 reported a new species of Mo5+ by photoreduction at 77 K which they claimed to be in a tetrahedral environment. While the spin Hamiltonian parameters of a paramagnetic species do give a certain amount of information regarding its Permanent address: Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, Changchun, The People's Republic of China.
0022-3654/85/2089-0636$01 SO10
geometry, usually it is difficult to obtain unambiguous details about short-range order from continuous wave ESR methods alone. In recent years a powerful pulsed magnetic resonance technique, (1) Masson, J.; Nechtschein, J. Bull. SOC.Chim. Fr. 1968, 3933. (2) Peacock, J. M.; Sharp, M. J.; Parka, A. J.; Ashmore, P. G.; Hocken, A. J. J. Catal. 1969, 15, 379. (3) Seshadri, K. S.; Petrakis, L. J. Phys. Chem. 1970, 74, 4102. (4) Seshadri, K. S.; Massoth, F. E.; Petrakis, L. J. Cafal. 1970, 19, 95. (5) Dufaux, M.; Che, M.; Naccache, C. J. Chim. Phys.-Chim. Biol. 1970, 67, 527. (6) Naccache, C.; Bandiera, J.; Dufaux, M. J. Cafal. 1972, 25, 334. (7) Che, M.; Tench, A. J.; Naccache, C. J . Chem. SOC.,Faraday Trans. 1 1974, 70, 263. (8) a. Burlamacchi, L.; Martini, G.; Ferroni, E. J . Chem. Soc., Faraday Trans. 1 1972, 68, 1586. b. Martini, G. J. Magn. Reson. 1974, 15, 262. (9) Howe, R. F.; Leith, I. R. J. Chem. SOC.,Faraday Trans. 1 1973.69, 1967. (10) Abdo, S.; LoJacono, M.; Clarkson, R. G.; Hall, W. K. J. Cafal. 1975, 36, 330. (1 1) Hall, W. K.; LoJacono, M. Proc. 6th In?. Congr. Catal. 1977, 246. (12) Abdo, S.; Clarkson, R. B.; Hall, W. K. J. Phys. Chem. 1976,80,2431. (13) a. Che, M.; Figueras, F.; Forissier, M.; McAteer, J.; Perrin, M.; Portefaix, J. I.; Praliaud, H. Proc. 6th Int. Congr. Catal. 1977, 261. b. Che, M.; McAteer, J. C.; Tench, A. J. J. Chem. SOC.,Faraday Trans. 1 1978, 7 4 , 2378. c. Lunsford, J. H. Coral. Reu. 1973, 8, 135. (14) Pershin, A. N.; Shelimov, B. N.; Kazansky, V. B. Kine?. Kafal. 1979, 20, 1298. (15) Petrakis, L.; Meyer, P. L.; Debies, T. P. J. Phys. Chem. 1980, 84, 1020. (16) Fricke, R.; Hanke, W.; Ohlman, G. J . Catal. 1983, 79, 1. (17) Machiels, C. J.; Sleight, A. W. J . Catal. 1982, 76, 238.
0 1985 American Chemical Societv
Mo-Adsorbate Interactions on Mo/SiO,
The Journal of Physical Chemistry, Vol. 89, No. 4, 1985 637
electron spin echo modulation (ESEM) spectroscopy,’s-20has been developed to give complementary information to that obtained from ESR. This technique can reveal weak dipolar interactions between the paramagnetic species and the surrounding magnetic nuclei which cannot typically be resolved by continuous wave ESR methods. With ESEM analysis we have been able to study the adsorbate interactions of Cu/Si02, V/Si02, and of several paramagnetic ions, and radicals in zeolites and other disordered media.2’-25 In this paper we present ESEM studies of Mo/SiO, prepared by impregnation and adsorbed with various inorganic and organic adsorbates. Both for water and methanol adsorbates only one molecule seems to be present in the first coordination sphere of Mo5+.
Experimental Section Silica gel from Fisher Scientific (Grade 950,60-200 mesh, 700 m2/g) was used after thorough washing with HCl to remove impurity Fe3+. The supported molybdenum systems were prepared by impregnating the silica gel with aqueous solutions of ammonium paramolybdate. The material was dried at 393 K for 4 h and afterward calcined at 773 K in flowing air for 24 h. The molybdenum content in these samples was -2 wt.%. For ESR and ESEM measurements the catalyst was placed in 3-mm 0.d. Suprasil quartz tubes and dehydrated in vacuo to a residual pressure of torr at 773 K for 0.5 h. At the same temperature the catalyst was reduced with 100 torr of H2 (Union Carbide 99.99% pure) for 1 h and then evacuated at 773 K for about 0.5 h. To study adsorbate interactions, the samples were cooled to ambient temperature and exposed to D 2 0 (Aldrich Gold Label 99.99%) CH30D, CD30H, CD3CN, and ND3 (Stohler Isotopes 99.9% D). The liquids were previously vacuum distilled over activated MgS04 and 3A zeolite. N H 3 and ND3 were purified by a freezepumpthaw method followed by vacuum distillation. The typical adsorption time was about 1 h. ESR spectra of the reduced catalysts with and without various adsorbates were recorded with a modified Varian E4 spectrometer at room temperature and at 77 K. ESEM spectra were obtained with a home-built s p e c t r ~ m e t e at r ~4.2 ~ ~K. ~~ Theory and Analysis The theory and analysis of electron spin echoes and the accompanying nuclear modulations have been well described in literature.1s-20 When resonant microwave pulses are applied in suitable sequences to a paramagnetic spin system spontaneous or stimulated microwave echoes are generated due to the reformation of macroscopic magnetization. In a two-pulse sequence, the echo occurs T after the second pulse, T being the interval between the first and second pulse. In a three-pulse sequence, the echo occurs after a time of T T from the second pulse where T i s the time interval between the second and third pulses. Scanning of T in a two-pulse experiment or T i n a three-pulse experiment results in the echo decay envelope. When the electron spins are excited by the microwave pulses, through hyperfine interaction the nearby nuclei are also excited and the echo decay envelopes are often modulated by the precessional frequencies of these interacting nuclei. Such modulations can be analyzed to yield information about the distance between the spin system and the interacting nuclei, the number of such nuclei, and the isotropic hyperfine coupling when the spin is partly delocalized onto these nuclei. In
+
(18) Mims, W. B. Phys. Rev. B 1972, 5, 2409. (19) Salikov, K. M.; Semenov, A. G.; Tsvetkov, Yu. D. “Electron Spin Echoes and Their Applications”; Science: Novosibirsk, U S S R . , 1976. (20) Kevan, L. In ‘Time Domain Electron Spin Resonance”; Kevan, L., Schwartz, R. N., Eds.; Wiley-Interscience: New York, 1979: Chapter 8 . (21) a. Ichikawa, T.; Kevan, L. J. Am. Chem. SOC.1981, 103, 5355. b. Ichikawa, T.; Yoshida, H.; Kevan, L. J. Chem. Phys. 1981, 75, 2485. (22) Narayana, M.; Narasimhan, C. S.; Kevan, L. J. Caral. 1983, 79, 237. (23) Narayana, P. A.; Kevan, L. J. Magn. Reson. 1982, 46, 84. (24) Janakiraman, R.; Kevan, L. J. Phys. Chem. 1982,86, 2727. (25) Kevan, L.; Narayana, M. In “Intrazeolite Chemistry”; Stucky, G., Ed.; American Chemical Society: Washington, DC, 1983; ACS Symp. Ser. No. 218, p 283. (26) Ichikawa, T.; Kevan, L.; Narayana, P. A. J. Phys. Chem. 1979,83, 3378. (27) Narayana, P. A.; Kevan, L. Photochem. Photobiol. 1983, 37, 105.
DPPH
1 W
200 0
H
Figure 1. ESR Spectra a t 77 K of Mo/SiO, (a) after H2(100 torr) reduction a t 773 K for 1 h, (b) after adsorption of D20(23 torr) for 1 h on the reduced sample, (c) after adsorption of CD,OH (100 torr) for 1 h on the reduced sample. Note the greatly reduced intensity of the high-field g,,line after adsorption of either adsorbate.
most cases when the interaction is primarily dipolar one observes the free precession Larmor frequency of the interacting nucleus and its second harmonic in a two-pulse experiment and only the free precession frequency in a three-pulse experiment. If the then its quadrupole interaction interacting nucleus has I > also plays an important role whenever the site symmetry at the nucleus is low enough to permit a sizeable electrical field grad i e r ~ t . ~ * -The ~ ~ presence of a nonzero nuclear quadrupole interaction (NQI) usually affects the two-pulse ESEM spectra conspicuously through rapid damping of the second harmonic or 2wf component. When the quadrupole interaction becomes significant compared to the dipolar interaction the free precession frequency of the nucleus is shifted depending upon the magnitude of the quadrupole interaction. When the NQI is nonzero but still small compared to the dipolar interaction, it affects three-pulse spectra only at longer values of the swept pulse interval time, T . Dikanov et al.” derived general expressions for any arbitrary value of nuclear spin interacting with an electron spin S = ’/,. These expressions consider the dipolar and hyperfine (contact) interactions exactly in the limit where the nuclear quadrupole interactions are zero and the hyperfine interactions are smaller than the dipolar interactions. More recently Shubin and Dikanov30 obtained expressions using a first-order perturbation approach including axially symmetric nuclear quadrupole interactions. However, these expressions are unsatisfactory when the electron-nuclear distances are less than 0.3 nm.33 Most of the data presented in this paper were obtained by using the exact expressions in the zero NQI a p p r o ~ i m a t i o n .For ~ ~ data where this was not adequate we tried to obtain better fits to the experimental spectra with the first-order perturbation approach of Shubin and Dikanov30 to include N Q I but the fits were not any better. Presummably this is due to the fact that the nuclear-electron distances are below the limits mentioned above so that the dipolar ~
~
~
~~~~~
(28) Mims, W. B. Phys. Rev. B 1972, B6, 3563. (29) Dikanov, S. A,; Shubin, A. A.; Parmon, V. N. J. Magn. Reson. 1981, 42, 474. (30) Shubin, A. A.; Dikanov, S. A. J. Magn. Reson. 1983, 52, 1. (31) Narayana, P. A.; Kevan, L. J. Magn. Reson. 1977, 26, 437. (32) Narayana, P. A.; Kevan, L. Magn. Reson. Rev. 1983, 7, 239. (33) Romanelli, M.; Narayana, M.; Kevan, L. J Chem. Phys. 1984.80, 4044.
638 The Journal of Physical Chemistry, Vol. 89, No. 4, 1985 TABLE I: ESR Parameters of Mo5+ in Mo/Si02 Systems‘ matrix sample color prep method Mo/SiO, dark gray H, reduction, 773 K Mo/Si02/D20 blue gray adsorbed, RT‘ Mo/Si02/CD30H blue gray adsorbed, R T Mo/Si02/ND3 blue gray adsorbed, R T Mo/Si02/CD3CN
blue gray
adsorbed, R T
Mo(CO)6/Si02
pale blue blue gray gray blue blue
evacuation, R T evacuation, 673 K H2 reduction, 973 K hydrolysis evacuation, 773 K H, reduction, 773 K C O reduction, 873 K D, reduction, 77 K, UV
MoClS/Si02 Mo/SiO, Mo/Si02
Narayana et al.
gLb
glI(”
d2’
1.95 1 1.936 1.937 1.956d 1.928 1.947d 1.929 1.934 1.948 1.955 1.936 1.941 1.940 1.961 1.982d 1.929 1.961
1.895 1.893 1.896 1.900
1.865
1.887 1.895 1.884 1.884 1.894
G
97 95 96
ref this work this work this work this work
99
this work
All,
82
9 9 9 16 16 13b 13a 14
98
14
1.862
80 1.865
1.882 1.891 1.811
98 1.861
1.891
uEstimated errors in gvalues are f0.003 and the estimated error in A,, for 95*97Mo is *3 G . bNote that g, is measured at the derivative peak of the g, component rather than more correctly at the derivative crossover point. This was done to compare with the previous literature which is reported in this way. ‘RT = room temperature. dRhombic g with g, split into g, and g,.
interaction terms were ~versimplified.~~ By using fast Fourier transform (FFT) methods the time domain data obtained in the two- and three-pulse experiments can be converted into frequency domain s p e ~ t r a . ~From ~ . ~these ~ it is possible to determine fairly accurately the magnitude of the isotropic hyperfine coupling if any is present and also the NQI parameter^.^^ When the N Q I becomes larger than the dipolar interaction, additional frequency components show up in the F?X spectra. However, if the NQI is nonzero but less than about 10% of the dipolar interaction it is difficult to detect it unambiguously from the FFT spectra. In such instances one can approximate the small NQI by using small frequency shifts in the expressions for the ENDOR transition^^^^^^^^' in time domain simulations.
Results ESR. N o ESR spectrum is seen in impregnated Mo/Si02 after drying at 393 K or after calcining at 773 K in flowing air. The samples remain white. Upon reduction with H2 at 773 K, the color changes from white to dark gray and a strong ESR signal is seen (Figure la). N o difference is seen between the spectra at 293 and 77 K, indicating the absence of motional averaging and a long spin-lattice relaxation time for the paramagnetic species. Most of the ESEM experiments were done with the dark gray sample (773 K reduction) after exposing it to various adsorbates. In the reduced catalysts with no adsorbate a strong echo signal is seen with no nuclear modulation. On exposure to D 2 0 or C D 3 0 H the sample color changed initially to white gray and then to blue gray. Evacuation at room temperature did not restore the white color or the dark gray color of the calcined and reduced samples. The ESR spectra of the samples with D 2 0 and C D 3 0 H adsorbates are shown in Figure 1, b and c, respectively. The most conspicuous difference with reference to the spectrum of the reduced sample is the disapat gll(2)= 1.865. The gl,(l)peak pearance of the high-field g11(2) does not change with these adsorbates but the g, peak decreases appreciably (see Table I). In Figure 2, the ESR spectra of the reduced samples obtained after exposure to ammonia and CD3CN (Figure 2, a and b, respectively) are shown. In contrast to water and methanol, these two adsorbates change the ESR spectrum to that typical of rhombic symmetry with three g values. In all the spectra in Figures 1 and 2 prominent hyperfine lines ( I = 5 / 2 ) are seen at low field. Weak corresponding due to 95*97M~ hyperfine lines are seen to high field at increased gain as indicated. An approximate value of the hyperfine coupling is obtained from the outer hyperfine lines and is included in Table I. Also given in Table I are the g values reported by Howe and Leith9 for molybdenum hexacarbonyl/SiO,, by Fricke et al. l6 for (34) Dikanov, S.A.; Tsvetkov, Yu. D.; Bowman, M. K.; Astashkin, A. V. Chem. Phys. Lett. 1982, 90, 149.
MolSlOp
‘
200G
’
-H
Vt ’ 93
Figure 2. ESR Spectra at 77 K of Mo/Si02 reduced at 773 K for 1 h in H2(100 torr) and adsorbed with (a) ND3 (100 torr) for 1 h and (b) CD3CN (-50 torr) for 1 h.
hydrolyzed MoClS/SiOz, by Che et a1.I3for Mo/Si02 prepared by impregnation and reduction by H2 and by CO, and by Pershin et aLi4for Mo/Si02 prepared by impregnation and 77 K reduction by D2 with ultraviolet light. ESEM Results. Mo/SiOz samples reduced in H2 at 773 K show an ESE signal but no modulation, not even proton modulation, could be observed. This indicates the absence of any hydrogen containing ligands in the first coordination shell. In Figure 3, the experimental and simulated three-pulse ESEM spectra are compared for a Mo/Si02 sample reduced at 773 K and adsorbed with DzO at room temperature. The best fit data obtained clearly indicates that there are two kinds of deuterons interacting with the paramagnetic ion, one at 0.29 nm with a small isotropic hyperfine coupling of 0.1 MHz and two farther deuterons at 0.36 nm with zero isotropic hyperfine coupling. The uniqueness of these parameters was checked and confirmed by simulating three-pulse spectra at different magnetic field values and comparing with the corresponding experimental data. While the fits obtained for all the three-pulse ESEM data were excellent, the fits for the two-pulse data were not as good. This is illustrated in Figure 4. The difference between the calculated and experimental data after T = 2 ps is significant. This is partly due to neglect of the nuclear quadrupole interaction in the expressions
The Journal of Physical Chemistry, Vol. 89, No. 4, 1985 639
Mc-Adsorbate Interactions on Mo/Si02
r
MolSi021D,0 3 PULSE ESE 1 r, = 0.29nm a 1 = 0.1 MHz
nl = n2 = 2 r 2 = O.36nm a 2 = 0.0 MHz
_-__
I
I
0 0
1
2
1
EXPT
0
-
4
3
MolSi02lD20 FFT OF 2 PULSE ESE H =3330 G
5
0
5
10
15
20
FREQUENCY, MHz
T, p s
Figure 3. Calculated and experimental three-pulse ESEM spectra of Mo/Si02 reduced at 773 K in H2 (100 torr) for 1 h and adsorbed with D 2 0 (23 torr) for 1 h at room temperature. The decay function used in the best fit was g ( T ) = exp(2.195 - 0.03T 0.0016p). The spike at -0.05 us is an artifact due to two-pulse interference.
Figure 5. Frequency spectra obtained by fast Fourier transform of the time domain data shown in Figure 4. The data tapering parameter was 10%.
+
nl
=
MolSiOgl CH30D 2 PULSE ESE n=1 r = 0.29nm a l s o = 0.2 MHz
MolSiOglD20 2 PULSE ESE 1 rl = 0.29nm a1 = O.1MHz
n2= 2
r2
= 0.36nm
-__-
____
a 2 = 0.OMHz
O
0
L
L
- 2 0.5
1.5
1
r,
Figure 4. Calculated and experimental two-pulse ESEM spectra of Mo/Si02 reduced at 773 K in Hz(100 torr) for 1 h and adsorbed with D20(23 torr) for 1 h at room temperature. The decay function used in the best fit was g ( 7 ) = exp(2.45 - 0.4s + 0 . 0 0 5 ~-~0 . 0 0 2 ~ ~ )The . significant differences observed in the 2w component depths after T = 2 ps are partly due to neglect of the nuclear quadrupole interaction in the Hamiltonian.
used for the simulation of the nuclear modulations. Some simulations were made with the first-order perturbation expressions of Shubin and D i k a n ~ v .These ~ ~ expressions include the nuclear quadrupole interaction but such simulations did not show any better fit to the experimental data. The failure of these first-order perturbation expressions to yield a better fit in spite of the inclusion of nuclear quadrupole interaction terms could be due to the over simplification of the dipolar interaction terms.33 Figure 5 illustrates the frequency domain spectrum obtained by FFT of the time domain data in Figure 4. The strong and relatively narrow peak at 2.10 MHz is close to the deuterium free precession frequency, 2.18 MHz in a magnetic field of 3330 G . A small splitting seen in the second harmonic peak with shoulders at 4.20 and 4.50 MHz indicates that the paramagnetic molybdenum interacts with at least two types of deuterons in accord with the conclusions from the simulations of the time domain data shown in Figures 3 and 4. This splitting in the second harmonic was observed in all the samples reduced at 773 or 973 K with D 2 0 adsorbate. The origin of the third relatively weak peak at 6.7 MHz is not clear a t this time but was reproducible in all the samples with adsorbed D 2 0 .
EXPT
- CALC
EXPT CALC
2
/is
Figure 6. Calculated and experimental two-pulse ESEM spectra of Mo/Si02 reduced at 773 K in H2(100 torr) for 1 h with adsorbed CH,OD (100 torr) for 1 h at room temperature. The additional structure seen in the experimental spectrum between 7 = 0.25 and 0.75 ps is due to modulation by interacting protons of the methyl group. The decay function used in the best fit was g ( r ) = exp(2.625 - 0.987 + 0.08i2).
Mo ISi 0 2 I C D30 H 2 PULSE ESE
nl = 3 rl = 0.36 nm n 2 = 3 r2= 0 . 4 5 nm
____
0
1
0
I
I
EXPT CALC
- i - I L I
2
1
r, ps
Figure 7. Calculated and experimental two-pulse ESEM spectra of Mo/Si02 reduced at 773 K in H 2 (100 torr) for 1 h with adsorbed CD,OH (100 torr) for 1 h at room temperature. Note the addition of a second shell of deuterons to improve the uniqueness of the fit. The decay function used in the best fit was g(7) = exp(2.90 - 1.927 + 0 . 2 8 ~ ~ - 0.01~3).
Narayana et al.
The Journal of Physical Chemistry, Vol. 89, No. 4, 1985
640
2 20
4 4 5 MHz
MoIS102IND,, NH, FFT OF 2 PULSE ESE H = 3380 G
MolSiOp
/\\! 11
2 1
0
0
5
10
15
__
I
i
NH3
._ ..
20
F R E Q U E N C Y , MHz
Figure 8. Frequency spectra obtained by fast Fourier transformation of the two-pulse time domain data for Mo/Si02 reduced at 773 K in H2 (100 torr) for 1 h with adsorbed NH3 (100 torr) for 1 h. The deuteron frequencies are at 2.20 and 4.45 MHz and the nitrogen frequencies are at 3.1, 5.3, and 7.1 MHz.
The calculated and experimental two-pulse ESEM spectra for Mo/Si02 with adsorbed C H 3 0 D and C D 3 0 H are shown in Figures 6 and 7. The agreement between the experimental and calculated data is excellent. Here it should be noted that the microwave pulse widths were deliberately chosen to reduce the interference of proton modulation in the samples with adsorbed CH,OD and C D 3 0 H . The proton modulations are shallower because the modulation depth is proportional to the nuclear spin and have smaller periods because of the higher Larmor frequency. Consequently, an important part of the data is lost during the dead time of the spectrometer which makes analysis of proton modulation less reliable in this case (see Figure 6). For the sample with adsorbed C H 3 0 D both two-pulse and three-pulse ESEM simulations show that the paramagnetic molybdenum interacts with only one deuteron at 0.29 nm with a small isotropic hyperfine coupling of 0.2 MHz. For the samples with adsorbed CD30H, the simulations indicate that Mo interacts with one group of methyl deuterons at 0.36 nm and with a farther group of methyl deuterons at 0.45 nm, with both groups having zero isotropic hyperfine coupling. In Figure 8, the frequency domain spectra obtained by FFT of the time domain data for Mo/Si02 samples with adsorbed NH, or ND, are compared. From this comparison the peaks at 2.20 and 4.45 MHz in the spectrum for ND, can be assigned to interacting deuteriums, the deuterium free precession frequency and the second harmonic in a magnetic field of 3380 G being 2.21 and 4.42 MHz, respectively. The other three peaks present with both adsorbates at 3.1, 5.3, and 7.1 MHz are assigned to Mo-N interactions. Because of such strong interaction with the nitrogen nucleus, it is difficult to simulate the two-pulse ESEM spectra since the zero quadrupole approximation breaks down. Figure 9 is the ESR spectrum observed on adsorbing about 2 torr of high-purity oxygen (99.995%) on the reduced samples at room temperature. The intensity of the signal in the g, region observed in the reduced sample decreases only by about 10% on oxygen adsorption but the signal intensity of the 02-radical spectrum observed after such adsorption is significantly higher by about a factor of 100. Also it should be noted that unlike the other adsorbates, the high-field peak of gil(2) = 1.865 does not disappear on adsorption of oxygen. The radical thus observed is in the literature'3c with gl, g2, and g, values of identified as 022.018, 2.012 and 2.004, respectively. In some samples reduced at 973 K, an 02-sepctrum with a slightly different g, value of 2.029 was observed but its reproducibility was poor. This species is probably due to oxygen adsorbed on Si4+ of the support.'3c Discussion ESR Data. To present our ESR results in the proper perspective we will briefly summarize some of the ESR data available in the literature on supported paramagnetic Mo species. Howe
Figure 9. ESR spectrum at 77 K of Mo/Si02 reduced at 773 K in H2 (100 torr) for 1 h and adsorbed with 2 torr of O2at room temperature for 1 min. Note that the gIl2= 1.865 species does not disappear in contrast to the case for adsorption with D20,CD30H,ND,, and CD3CN in Figures 1 and 2.
and Leith9 carried out an extensive ESR study of Mo/Si02 prepared by grinding M o ( C O ) ~with silica and activating the mixture by evacuation at different temperatures. They distinguished three different Mo species which can be described by their gvalues shown in Table I. A species denoted as Mo(C) with g, = 1.934 seen in mildly activated and rehydrated samples was assigned to Mo5+ in distorted octahedral or square-pyramidal geometries; it is implied that Mo(C) has some hydroxyl or water ligands in its first coordination shell. Species Mo(B) with g, = 1.948 observed on activation at higher temperatures was suggested to still have square-pyramidal geometry, but with oxide ions replacing hydroxyl or water ligands in the first coordination sphere of Mo5+. On H2 reduction at 973 K of the activated samples, Mo(B) is seen together with a new species, Mo(A), with gSl,= 1.862. No specific coordination geometry was assigned to Mo(A) and this species was found to disappear by outgassing the sample at 773 K. Che et observed only what appears to be a Mo(B) species in impregnated Mo/Si02 samples upon high-temperature reduction with H2. However, by using C O instead of H2 as the reducing agent they observed simultaneous formation of Mo(A) and presumably Mo(B) although the g values differ somewhat. They found Mo(A) to be reactive with adsorbed oxygen leading to the formation of OF. For high-temperature reduction with H 2 our results show Mo(A) and Mo(B) formation which is in reasonable agreement with the results of Howe and Leith.9 However, we find that Mo(A) does not disappear by degassing at 773 K as Howe and Leith9 report. The lack of proton modulation on the echo signal of H2 reduced Mo/Si02 indicates that Mo(B) and presumably Mo(A) do not have hydroxyl or water ligands. This is consistent with the conclusion of Howe and Leith.9 Our new results involve the effect of several adsorbates on Mo(B) and Mo(A) and the associated ESEM studies. First we note that Mo(A) disappears with all adsorbates (HzO, C H 3 0 H , NH,, CH,CN) tried. Secondly, the g, value shifts to smaller values consistent with Mo(C) formation. It appears that Mo(A) reacts with all adsorbates to form Mo(C); some reaction of Mo(B) to form Mo(C) may also occur. It thus seems likely that Mo(A) is coordinatively unsaturated and reacts with adsorbates to form a more stable octahedrally coordinated species as has been suggested9 for Mo(C). For H20 and CH,OH adsorbates the ESEM results indicate only one directly coordinated molecule, so a square-pyramidal structure seems probable for Mo(A). Figure 9 shows that Mo(A) is not reactive toward 02,in contrast to the results of Che et al.," but that some other species in the Mo/Si02 surface can donate an electron to O2to form 0,. Hall and co-workerse" have suggested that electron-donating Mo centers are most likely in highly symmetric tetrahedral coordi-
Mo-Adsorbate Interactions on Mo/Si02 nation and thus are not observable by ESR owing to very fast magnetic relaxation via a low-lying excited state. Pershin et al.14 observed a different Mo5+ species on S i 0 2 upon irradiation with ultraviolet light at 77 K in a hydrogen atmosphere with gll = 1.811. They assigned this to tetrahedral coordination. Thus we suggest that Mo(A) is not tetrahedrally coordinated. Adsorption of NH, or CH3CN leads to lower symmetry to the g, region with two peaks assignable to rhombic symmetry. The gll(*)of Mo(A) disappears as in the case of H 2 0 or C H 3 0 H adsorption. The complex structure in the gIlregion could be due to different MoS+having different number of adsorbate molecules in the coordination sphere because of incomplete complexation. However, aging of the samples in the NH, or CH3CN atmosphere did not change the spectra. Thus we conclude that MoS+ site symmetry becomes rhombic on adsorption of N H 3 or CH3CN. ESEM Data. In the samples with adsorbed D 2 0 the best-fit simulations of the ESEM data indicate that Mo(C) interacts with one deuteron at 0.29 nm and two farther deuterons at 0.36 nm. The FFT frequency spectra also indicate that at least two types of deuterons interact with Mo'+. We suggest that there is one hydroxyl group in the first coordination sphere of Mo(C); the two further deuterons are assigned to an uncoordinated water molecule. If tetrahedral orbitals are assumed on the oxygens of OD, the Mo-D distance of 0.29 nm translates into a M o - 0 distance of 0.22 nm. For Mo6+-06 distorted octahedra the Mo-O distances are reported35to be -0.195 nm in-plane and 0.168 and 0.232 nm axially with the short bond apparently being a double bond. Thus if MoS+in Mo(A) has square-pyramidal geometry with an oxygen in the apical position missing, it can form Mo(C) by coordination to one adsorbate ligand (L) with a MoS+-L distance less than 0.232 nm since the ionic radius of MoS+is expected to be larger than that of Mo6+. The two more distant deuterons at 0.36 nm correspond to a Mo5+-0 distance greater than that required for direct coordination. Thus they can be assigned to a nearby water molecule. The ESEM data for C H 3 0 D shows one interacting deuteron at 0.29 nm while for CD,OH it shows three interacting deuterons at 0.36 nm. For tetrahedral orbitals on the oxygens of methanol and the methanol molecular dipole directed along the Mo-0 direction both of these distances translate into a Mo-0 distance of 0.22 nm, the same as found for hydroxyl coordination. This is consistent with one intact methanol molecule coordinating to Mo(C). The ESEM data with C D 3 0 H adsorbate also indicates three farther interacting deuterons at 0.45 nm. These are too far for direct coordination and can be assigned to a nearby methanol molecule. The second shell methanol molecule was not observed by ESEM for C H 3 0 D adsorbate because the influence of one deuteron at a longer distance is much weaker than that of three. G r ~ f has f ~ ~reported infrared studies of methanol on polycrystalline MOO, at 373 K indicating that methanol adsorbs weakly at one site as the undissociated molecule and strongly at another site as the methoxy group. Since we adsorbed methanol at room temperature we believe that methanol only adsorbed as the undissociated molecule. Although Mo/SiO, only coordinates one adsorbed methanol, ESEM data for other metal species on Si02 indicates that Cu/Si02 coordinates two methanol mole(35) Allcock, H. R.; Bissel, E. C.; Shawl, E. T. fnorg. Chem. 1973, 12, 2963. (36) Groff, R. P. J. Caral. 1984, 36, 215.
The Journal of Physical Chemistry, Vol. 89, No. 4, 1985 641 cules21band that V/Si02 coordinates three methanol molecule^.^' Adsorption of NH, or ND, gives rise to a complicated ESEM spectrum for the Mo/Si02 samples. The frequency spectra obtained by FFT of the time domain ESEM data show both nitrogen and deuterium interactions for ND, and only nitrogen interactions for NH3. The peaks at 2.20 and 4.45 MHz are at the deuterium free precession frequency and its twofold multiple for Ho = 3380 G so these peaks are assigned to a Mo-D interaction. Since the frequency is not shifted from the free precession value there is no significant delocalization of the unpaired spin from Mo5+onto the protons or deuterons of the adsorbed ammonia. G r ~ f f ' s , ~ infrared data for adsorbed NH, on MOO, and the earlier data of Belokopytov et al.,* indicate that ammonia adsorbs very strongly, most probably via a coordinate bond from nitrogen to the Mo species. The free precession frequency of 14N at Ho = 3380 G is 1.0 MHz. Unfortunately, due to the removal of signal decay from the time domain data, reliable information cannot be obtained from our FFT spectra for 0 to 1.O MHz. However, the strong frequency components observed at 3.1, 5.3, and 7.1 MHz indicate that the nuclear quadrupole interaction of the nitrogen nucleus is significant relative to the dipolar interaction. It is significant that similar nitrogen frequency components were observed by ESEM for oxidized V/Si02 with adsorbed ND,.,' Conclusion
Two types of Mo5+ denoted as Mo(A) and Mo(B), observable by ESR were generated by reducing impregnated Mo/Si02 samples at 773 K with H2. Species Mo(A) disappears on adsorption of water, methanol, ammonia, and acetonitrile and a new species Mo(C) is formed. Adsorption of O2onto reduced samples results in an intense 02-radical in accordance with previous reports but does not affect the Mo(A) intensity. Species Mo(B) is much less reactive with adsorbates. Simulation of ESEM data indicates that Mo5+ in Mo(C) interacts with one nearby deuteron at 0.29 nm and two farther deuterons at 0.36 nm in the samples with adsorbed D20. For samples with adsorbed C H 3 0 D , the ESEM analysis yields one deuteron at 0.29 nm while the analysis for CD30H adsorbate gives three deuterons at 0.36 nm and three more deuterons at 0.45 nm. These data are consistent with a model of MoS+interacting with one hydroxyl group or with one methanol molecule in the first coordination sphere with a Mo-0 distance of 0.22 nm. One further noncoordinated water and methanol are also indicated by the ESEM data. ESEM spectra could not be simulated for NH, and ND, adsorbates because of strong Mo-N interactions. The frequency spectra obtained by fast Fourier transformation of the time domain data indicate that there is no spin delocalization onto the ammonia protons (deuterons) and that the quadrupole interaction of 14N plays a significant role relative to the dipolar interaction.
Acknowledgment. We thank the Robert A. Welch Foundation, the National Science Foundation, and the Energy Laboratory of the University of Houston for support of this research. Registry No. Mo, 7439-98-7; HzO, 7732-18-5; CH,OH, 67-56-1; NHJ, 7664-41-7; CH3CN, 75-05-8; D2,7782-39-0. (37) Narayana, M.; Narasimhan, C. S.; Kevan, L. J. Chem. Soc., Faraday Trans 1, in press. (38) Belokopytov, Yu. V.; Kholyavenko, K. M.; Gerei, S. V. J. Cural. 1979, 60, 1 .