J. Chem. SOC., Faraday Trans. I, 1982, 78, 1043-1050 Use of Hydrogen Atoms for the Low-temperature Reduction of Oxides B Y MICHEL CHE,* BEGONA CANOSA? AND AGU s T IN R. G ON z A LE Z-E L I PET Laboratoire de Chimie des Solides, ER 133, CNRS, Universite P. et M. Curie, 4 Place Jussieu, 75230 Paris Cedex 05, France Received 6th April, 1981 The interaction of hydrogen atoms with the surface of supported (MOO, or V,O, on SiO,) or bulk oxides (TiO,) kept in liquid nitrogen has been studied by e.p.r. after the oxides were pretreated in oxygen at high temperatures. It is shown that reduction occurs as evidenced by the appearance of e.p.r. signals assigned to (MO=O),+ (gL = 1.952, gI1 = 1.890), (V=O)z+ (gL = 1.982, gll = 1.938, Al = 64 G, All = 190 G) and Ti3+ (sl = 1.976, gll = 1.950).These paramagnetic centres are found to be inert towards oxygen, showing that they are located below the surface, in contrast to the result; obtained after thermal reduction in uacuo or in hydrogen. The presence of F centres (sli = 2.0015, gl = 2.0000), observed in the supported oxides only, is related to the support, as shown by experiments performed with SiO, alone. The formation of reduced transition metal ions and of other centres (F centres, O;, OzHo) is discussed on the basis of the reaction of Ho atoms with 0,- oxide ions. The most widespread method of reduction of oxides is to heat them in uacuo or in a reducing atmosphere (hydrogen or carbon monoxide).l Some other methods have also been used, for instance low-temperature irradiation in vacuo or in a reducing atmosphere with U.V.light2 or y-rays.l? Recently, it has been shown by ferromagnetic resonance that hydrogen atoms can be used to produce nickel particles of small diameter by reduction of Ni2+ cations within the framework of exchanged Nix zeolites, due to the small size of the hydrogen atom.4 The present work reports the use of hydrogen atoms for the reduction at low temperatures of supported (MOO, or V,O, on SiO,) or unsupported (TiO,) transition metal oxides, as monitored by e.p.r. There are several reports on the interaction of atoms with oxide surfaces. One of the first examples is the production of a blue colour when MOO, was exposed to a stream of hydrogen atoms. This colour reaction has been used to obtain the rates of hydrogen atom reactions and a similar effect has been reported for WO,.,q6 Smith and Tench' were able to produce surface F centres on high surface area MgO bombarded with hydrogen atoms produced by a microwave discharge.They also studied the reaction of such hydrogen atoms with alcohols preadsorbed on Mg0.8 Transition metal oxides have also been considered. Thus, the interaction of hydrogen atoms with TiO, has been followed by conductivity meas~rements.~ Oxygen atoms have also been prepared and their interaction with various surfaces (zeolite 4A,1° MgO,ll Ti0212) investigated. In relation to the present work, Wintruff et al.13 described the reactivity toward oxygen of surface defects produced by the interaction of a hydrogen plasma with SiO,. The present report has a direct implication for hydrogen spillover studies.In t Present address: Institut fur Physikalische Chemie, Sophienstrasse 1 l,8000%lunchen, West Germany. 10431044 REDUCTION BY HYDROGEN ATOMS particular, if the active and migrating species involved in spillover is the neutral atom, it should behave like the hydrogen atom produced from the microwave discharge. We have thus restricted our choice to the oxides M00,,14 V,o5l5 and Ti0,,16 whose reduction by spilled-over hydrogen using Pd or Pt metal has been investigated in detail. EXPERIMENTAL The unsupported oxide was a TiO, prepared by the flame method (anatase, Degussa, P25). The supported oxides (MOO, and V,O, on SiO,) were prepared by the grafting method by reacting surface silanol groups with MoCl, dissolved in chloroform' or directly with vapour phase VCl,.17 Prior to reduction with hydrogen atoms, the samples were heated first in U ~ C U O for one hour and then in oxygen for one hour at 400 OC for TiO, and 500 "C for MOO, and V,O, on SiO,.The catalysts had a white colour after this oxidizing treatment, which was carried out in the same cell where the reduction was taking place. 'C FIG. 1 .-Experimental set-up for the reduction of oxides with hydrogen atoms. A, e.p.r. tube; B, microwave cavity; C, liquid-nitrogen dewar; D, microvalve; E, Pirani gauge; F, stopcocks; G, sintered disc. The experimental set-up in the present work is a modified version of the one described earlier17* l8 and is represented schematically in fig. 1. The hydrogen had a purity of 99.95 % and was used after passing through a trap at 77 K.Its flow could be controlled by measuring the pressure with the aid of a gauge placed in E. The cell could be separated and isolated from the rest of the system by means of the stopcocks F. The catalyst was maintained in the liquid- nitrogen bath C during the reduction on the sintered disc G. The origin of time for the various reduction periods was taken when the microwave discharge was switched on. It was produced with a Microtron 200 microwave power generator mark 111 (Electro Medical Supplies, Wantage, UK) equipped with a microwave discharge cavity B tuned at 2450MHz. The apparatus was made of Pyrex except for the silica tubing passing through the microwave cavity. Typical operating conditions were: pure H, flow (4 x lop2 dm3 min-l) under 1 Torr pressure with microwave discharge power of 150 W.After reduction by Ho atoms, the catalyst was transferred to an e.p.r. tube A which was finally sealed off for e.p.r. observation. The entire process was performed within a polystyrene vessel filled with liquid nitrogen. For most experiments, e.p.r. spectra were recorded at 77 K on a Varian spectrometer (model E3, X band) with 100 kHz modulation. The spectrometer was also equipped with a 77-513 K variable temperature accessory (E-257). The g values are measured relative to a DPPH sample (g = 2.0036). The magnetic field increases from left to right in the figures.M. CHE, B. CANOSA A N D A. R. GONZALEZ-ELIPE 1045 RESULTS After 3 min exposure of TiO, to hydrogen atoms, the solid had a brown colour associated with an e.p.r.spectrum at 77 K (fig. 2). Two kinds of species could be detected, one with g, = 1.976, gll = 1.950 and the other with g, = 2.000, g , = 2.008 and g, hardly resolved. The first signal is very similar to that observed after thermal reduction of TiO,, and thus assigned to Ti3+ ions. The second is similar to that detected after oxygen photoadsorption on hydroxylated TiO, surfaceslg and assigned to an O,Ho radical. These two species (Ti", O,Ho) were not stable at room temperature. After exposure of MoO,/SiO, (1 h) and V,O,/SiO, (10 min) to hydrogen atoms, the solids turned brown and gave e.p.r. spectra. ' I Y 2.000 FIG. 2.-E.p.r. spectrum (X band, 77 K) of TiO, (anatase) maintained in liquid nitrogen and reduced by hydrogen atoms for 3 min.The e.p.r. spectrum obtained at 77 K for the MoO,/SiO, sample is represented in fig. 3. It is composed of a signal with g, = 1.952, gll = 1.890, similar to that observed after thermal reduction of M00,/Si0,~* and assigned to (Mo=O),+. Another signal can also be identified with the following g tensor components: g, = 2.018, g , = 2.010 and g, = 2.0045, which correspond to the parameters reported for 0; absorbed onto Mo0,/Si0,.21 The last signal which can be identified is very narrow and almost symmetrical with gll = 2.001 5 and g, = 2.0000. It is saturated easily with microwave power above 4 mW either at 77 K or at room temperature. The g values of this species, close to the free electron g value, and its facility to become saturated with low microwave power make it likely to be assigned to an F centre,,, which is in line with previous work on MgO where F centres were produced at the surface of the solid by interaction with Ho atoms.' After heating the sample at room temperature, the 0; species was no longer observed at 77 K or room temperature, in contrast to the (Mo=O),+ species and the F centre which did not change in either its g values or its intensity.The e.p.r. spectrum for the V,O,/SiO, system is shown in fig. 4. A signal with parametersg, = 1.982,g11 = 1.938, A , = 64 G, A , , = 190 G, similar to those obtained after thermal reduction of V,O,/SiO,, can be identified and assigned to (V=O)2+.23 F centres similar to those observed for MoO,/SiO, could also be found in the spectrum. The spectrum recorded at 77 K did not change after warming the sample to room temperature.1046 REDUCTION BY HYDROGEN ATOMS FIG.9 F centre 3.-E.p.r. spectrum (X band, 77 K) of MoO,/SiO, maintained in liquid nitrogen and reduced by hydrogen atoms for 1 h. 2.0045’ 1 g,! = 1.938 IDPPH n Y F cent re FIG. 4.-E.p.r. spectrum (X band, 77 K) of V,O,/SiO, maintained in liquid nitrogen and reduced by hydrogen atoms for 10 min. Note that on exposure to oxygen at 77 or 300 K, the three oxides reduced by this method did not lead to the formation of superoxide 0; ions, as normally observed for samples reduced by the thermal method. This indicates that the reduced transition metal ions were located below the surface and were not accessible to oxygen at 77 or 300 K. It also demonstrates that the coordination sphere of the cation is complete and does not change on adsorption of oxygen.This situation is similar to that observed for (V=O)2+ cations.24 When the bombardment of the oxides by hydrogen atoms was performed at room temperature, the same results as previously were obtained except for the centres which were not stable at room temperature (Ti3+, 02H0, 0;).M. CHE, B. CANOSA A N D A. R. GONZALEZ-ELIPE 1047 Experiments performed with the silica support alone (Degussa, 300 m2 g-l) led to the formation of an e.p.r. spectrum at 77 K characterised by gl = 2.0000 and gll = 2.0015 with a peak-to-peak line width AHpp N 4 G. This e.p.r. signal, which did not disappear on warming to room temperature and was easy to saturate on increasing the microwave power, is assigned to F centres.Similar results have been obtained by Wintruff et aZ.13 in the case of SiO,. DISCUSSION The small concentration of Ho atoms detected by their e.p.r. spectra for the solids reduced and maintained throughout at 77 K shows that the Ho atoms produced by the microwave discharge have mainly reacted with surface 0,- ions followirig the reaction Ho + 02- -+ OH- + e-. (1) ( 2 ) The electrons are then trapped by transition metal ions according to the process e- + Mn+ -+ M(n-1)+ as shown by the e.p.r. spectra of Ti3+, (Mo=O)~+ and (VZO)~+ species. The inertia of these reduced cations is expected since the pretreatment 'in oxygen' is believed to fill any vacancy in the coordination sphere of surface transition metal ions with adsorbed oxygen. In this sense, the case of MoO,/SiO, is very helpful since it has been shown earlier that reduction of such catalysts in the proper conditions could produce two kinds of (Mo=O)~+ species characterised by gl = 1.958, gI1 = 1.856 and gl = 1.941, gll = 1.885' assigned to bulk octahedrally hexacoordinated and surface pentacoordinated (Mo=O)~+ species, respectively.20 The former was found to be inert on exposure to oxygen in contrast to the latter which led to the formation of 0; species. Note that Ho atoms have been shown to oxidize Ti3+ ions in acidic organic according to the reactions (3) H0-Ti1I1+H+ -+ H,+TiIV.(4) HO +Ti111 + HO-TiIII In the present work, it is shown that Ti3+ ions are octahedrally coordinated and not reactive eliminating reaction (3) as a possibility.Furthermore, process (4) requires the reaction between two ' mobile' species H-Ti111 and H+ which is not possible in the present work since one is dealing with adsorbed species at 77 K and since Ti3+ ions are finally detected. The presence of F centres observed only in the case of supported oxides is to be associated with the support, since experiments performed on SiO, alone indicate the presence of F centres. This is consistent with: (i) The absence of hyperfine structure in the F centre signal for V205/Si0,. This indicates that the trapped electron is not interacting with V nuclei ( I = S, 100% natural abundance) and thus not located in the supported V,05 oxide but rather in the support SiO, oxide. (ii) The fact that transition metal ions are better traps than anion vacancies, if any, in the transition metal oxides (MOO,, V,O,).In contrast, in the support oxide (SiO,) the anion vacancies are more efficient traps than Si4+ which are not known to be reduced in normal conditions. Accordingly, F centres were not observed for TiO,, where the only electron traps are the Ti4+ ions.1048 REDUCTION BY HYDROGEN ATOMS The presence in some cases of 0; or O,HO radicals can be explained following the ( 5 ) (6) reactions : 0,+e- -+ 0; 0, + Ho -+ O,HO. The two radicals 0; and O,HO are not independent as shown by the reaction O,HO -+ 0; + H+. (7) In the case of a hydroxylated TiO, surface, it has been shownlg that reaction (7) was displaced to the left. Thus, according to reactions (1) or (7), an increase in the degree of hydroxylation is likely even if, at the beginning of the reduction by Ho atoms, the oxide surfaces can be considered as almost free of OH-.The hydrogen atom method appears attractive since reduction can occur at low temperatures as indicated by the present results. From available thermodynamic data, it can be calculated that the two following reductions are exothermic: 2 TiO,(s) + 2 H ' (g) -+ Ti,O,(s) + H,0(1) AH2,, K = - 84.4 kcal m0l-l v2°5(s) + * (g) v2°4(s) + H20(1) AHZg8~ = - 143.2 kcal m01-l (9) where (s), (g) and (1) refer to solid, gas and liquid, respectively. The oxides Ti,O, and V,04 have been chosen to correspond to the reduction state deduced from the e.p.r. data. The previous enthalpy variations have been obtained using Hess's law from the enthalpy changes of the following reactions : 2 TiO,(s) -+ Ti,O,(s) +i O,(g) AH298 = + 88.1 1 kcal mo1-l 26 a v2°5(s) '2O4(') + i O2(g) = +29.3 kcal mol-l 26b H,(g) + f 02(g) -+ H20(1) AHsg8 K = - 68.3 1 kcal m0l-l 26c In reactions (8) and (9), the formation of liquid H,O has been considered.However, the addition of a term including the enthalpy of adsorption on the surface would not change the reasoning. As a matter of fact, for the reaction 02-(s) + H,O --+ 20H-(s) (14) the enthalpy value estimated from desorption studies is - 107 2 kJ mol-l for TiO,,' while a similar value of - 117 kJ mot1 can be obtained from immersion studies.28 For MOO,, although this calculation cannot be made because of the lack of data for Mo205, a reasonable value can be obtained by comparison with data concerningM.CHE, B. CANOSA AND A. R. GONZALEZ-ELIPE 1049 (16) leading to 2 W03(s) + 2 H’ (8) -+ W205(s) + H20(1) = - 109.7 kcal mol-l. Measurements of the temperature by means of a thermocouple located just below the sintered disc indicate that, during the reduction, the temperature increases to a few degrees above 77 K. The ‘local’ temperature where the reduction actually takes place is, however, difficult to measure, although attempts are being made. Note also the absence of reactivity of the transition metal ions. As outlined above, this is due to the pretreatment in oxygen at high temperature. It is likely that if the reduction is performed after a pretreatment in uacuo, the exposed transition metal ions should be obtained with higher reactivity, This has been observed in the case of MgO pretreated in uacuo.l1 Experiments are presently being performed along this line. Finally, our results indicate that hydrogen atoms produced by a microwave discharge are able to produce isolated reduced transition metal ions at low temperature. It is thus not surprising that the final state of reduction is different from that obtained at substantially higher temperatures with spilled-over hydrogen which was able to reduce MOO, to M0,0,,~9~ CrO, to Cr20530 and TiO, to Ti,O,.ls In the latter case, one would not expect, for complete reduction, to detect any e.p.r. spectra because of dipolar interaction between neighbouring Mo5+, Cr5+ or Ti3+ in the reduced oxides. The fact that we have also used supported oxides (MOO, and V205 on SO,) is also to be considered.It has been shown that supported MOO, is less reducible than bulk We are thus investigating, for the sake of comparison, the reduction of bulk MOO,, V205 and CrO, by hydrogen atoms produced by a microwave discharge. We thank B. Morin for recording some of the e.p.r. spectra. A. R. G-E. acknowledges a grant from the CSIC, Madrid, Spain. Partial support of this work from the ‘Ministkre des Universites’ (grant 78 C 527 E) is also acknowledged. M. Che, F. Figueras, M. Forissier, J. C. McAteer, M. Perrin, J. L. Portefaix and H. Praliaud, Proc. VZth Znt. Congr. Catal. (The Chemical Society, London, 1977), vol. I, p. 251. P. Mbriaudeau, M. Che, P. C. Gravelle and S. J. Teichner, Bull. SOC. Chim., 1971, 13. M. Che, M. Dufaux and C.Naccache, Tagung Hoch-frequenzspektroskopie (Karl-Marx Universitat, Leipzig, 1969), Erganzugband, p. 10. M. Che, M. Richard and D. Olivier, J. Chem. SOC., Faraday Trans. 1, 1980, 76, 1526. T. H. Johnson, J. Franklin Znst., 1929, 207, 629. H. W. Melville and J. C. Robb, Proc. R. SOC. London, Ser. A, 1949, 196, 445. D. R. Smith and A. J. Tench, Chem. Commun., 1968, 1 113. D. R. Smith and A. J. Tench, Can. J. Chem., 1969,47, 1381. P. Dumont and P. de Montgolfier, J. Chim. Phys., 1972, 69, 16. lo P. Svejda and D. Hermerschmidt, Ber. Bunsenges. Phys. Chem., 1976, 80, 491. l1 P. Svejda, R. Haul, D. Mihelcic and R. N. Schindler, Ber. Bunsenges. Phys. Chern., 1975, 79, 71. l2 P. Svejda, W. Hartmann and R. Haul, Ber. Bunsenges. Phys. Chem., 1976, 80, 1327. l3 W. Wintruff, R.Herrling and H. J. Tiller, Chem. Phys. Lett., 1976, 38, 524. l5 M. A. Ibanez and G. C. Bond, unpublished results. l6 R. T. K. Baker, E. B. Prestridge and R. L. Garten, J. Catal., 1979, 56, 390; 1979,59, 293. l7 B. N. Shelimov, C. Naccache and M. Che, J. Catal., 1975, 37, 279. l8 D. Olivier, M. Richard, L. Bonneviot and M. Che, Growth and Properties of Metal Clusters, ed. J. l9 A. R. Gonzalez-Elipe, G. Munuera and J. Soria. J. Chem. SOC., Faraday Trans. 1 , 1979, 75, 748. 2o M. Che, M. Fournier and J. P. Launay, J. Chem. Phys., 1979, 71, 1954. 21 M. Che, A. J. Tench and C. Naccache, J. Chem. SOC., Faraday Trans. I , 1974, 70, 263. 22 B. Henderson and A. K. Garrison, Adu. Phys., 1973, 22, 423. 23 L. L. Van Reijen and P. Cossee, Discuss. Faraday SOC., 1966, 41, 277. 24 V. B. Kazansky, V. A. Shvets, M. Ya. Kon, V. V. Nikisha and B. N. Shelimov, Catalysis, ed. J. ,G. C. Bond and J. B. P. Tripathi, J. Less-Common Met., 1974, 36, 31. Bourdon (Elsevier, Amsterdam, 1980), p. 193. Hightower (North Holland, Amsterdam, 1973), vol. 11, p. 1423.1050 REDUCTION BY HYDROGEN ATOMS 25 D. Behar and A. Samuni, Chem. Phys. Lett., 1973, 22, 105. 26 I. Barin and 0. Knacke, Thermochemical Properties of Inorganic Substances (Springer-Verlag, Berlin, 27 G. Munuera and F. S. Stone, Discuss. Faraday Soc., 1971, 52, 205. 2* C. M. Hollabaugh and J. J. Chessick, J. Phys. Chem., 1961, 65, 109. 29 Selected Values of Chemical Thermodynamic Properties (Department of Commerce, Washington 30 J. B. P. Tripathy and G. C. Bond, J . Indian Chem. Soc., 1978, 55, 950. 31 J. Masson and J. Nechtschein, Bull. SOC. Chim., 1968, 3933. 1973), (a) p. 784, (6) p. 830, (c) p. 323. D.C., 1961), p. 296. (PAPER 1/547)