J. Chem. SOC., Faraday Trans. I, 1986, 82, 273-280 Electron Spin Resonance Studies of Free and Supported 12-Heteropoly Acids Part 2.-Reduction of H,[PMo,,O,,] - xH,O and H,[SiMo,,O,,] xH,O and Oxygen Adsorption Rolf Fricke and Gerhard Ohlmann Central Institute .for Physical Chemistry, Academy of Sciences of the G.D.R., DDR- 1 199 Berlin-Adlershof, German Democratic Republic Hydrogen and carbon monoxide reductions of SiMo,, and PMo,, heteropoly acids (HPA) have been studied, the various Mo5+ signals detected being identified as arising from stages in the dehydration and destruction of the HPA in a similar manner to vacuum heat treatment. Oxygen adsorption studies revealed the formation of 0; radicals on the supported HPA, prereduced at high temperatures, a phase in which the Keggin structure is already destroyed.In contrast, HPA in which the Keggin structure is preserved proved unable to stabilise 0; species. The stability of the supported HPA was shown to be markedly dependent on the nature of the support. Changes in the structure of heteropoly acids during dehydration in air and in vacuum have been identifiedl through an analysis of the sequence of e.s.r. signals arising from different stages in the dehydration process. The dehydration of the HPA is apparently accompanied by the reduction of the molybdenum ions and eventual destruction of the Keggin anion structure. This electron spin resonance (e.s.r.) study continues these investigations and evaluates the differences and similarities in behaviour of free and supported HPA during hydrogen or carbon monoxide reduction and in their responses to oxygen adsorption. Experimental General details of materials preparation and e.s.r.techniques are given in Part 1.' Reduction with hydrogen or carbon monoxide was carried out at 30 Torr? pressure in situ within e.s.r. tubes under static conditions and avoiding subsequent exposure to the atmosphere. For the oxygen adsorption studies three catalysts were compared : SiMo,,/SiO, and SiMol,/Al,O,, each containing 5 % HPA, and an Mo/SiO, monolayer catalyst (4.3% M o ) ; ~ all samples contained no phosphorus. The samples were prereduced with CO at different temperatures (623, 673, 723 and 773 K) and evacuated for 10 min at the temperature chosen. Molecular oxygen was adsorbed on the prereduced catalyst at room temperature and the excess was pumped off for several minutes.For a given temperature the oxygen pressure was varied (5, 30 and 100 Torr). t 1 Torr = 101 325/760 Pa. 273274 E.S.R. Studies of Heteropoly Acids 293 373 473 573 673 773 8 73 f I I I I I ]PMO,~ - HPA I I I I - - I co H2 f I 290 373 473 573 673 773 873 T/K Fig. 1. Schematic representation of Mo5+ e.s.r. signals obtained after reduction of PMo,, samples with H, or CO. (Circles denote spectra shown separately; for the g values see table 1.) Results Reduction in H, or CO The resonance signals observed after reduction in hydrogen or carbon monoxide are shown in fig. 1. (Most are described in detail in Part 1 .)l Some additional remarks may be helpful. (a) In a manner similar to that observed for dehydrated HPA (fig.4 in Part 1) the spectra of PMo,,-HPA reduced at 573 K show a superposition of two signals, one of which has identical parameters to signal F.I The second signal has a maximum intensity when measured at room temperature [fig. 2(a)] and shows an enhanced intensity after further evacuation of the sample at the reduction temperature [fig. 2(b)]. The g value assessed at the peak maximum (g = 1.944) suggests an assignment as signal E.l (b) A PMo,,-HPA sample whose Keggin anion structure was destroyed by heating in air at 773 K was reduced in a stepwise manner beginning at room temperature. After each reduction step the e.s.r. spectrum remained identical to that obtained immediately after the 773 K air treatment. The sequence of signals characterising the reduction states (fig.1) could not be observed. ( c ) The marked variation in line shape observed for PMo,,-HPA and PMo,,/SiO, samples reduced between 623 and 773 K (fig. 3) is a result of the superposition of threeR . Fricke and G. Ohlmann 275 Fig. 2. PMo,,-HPA, reduced with hydrogen at 573 K for 2 h (a) without further treatment and (b) after additional evaluation of sample a for 1 h at 573 K. (Dashed lines show the spectra registered at 295 K; full lines show those at 77 K.) Fig. 3. PMo,,/SiO,, reduced with hydrogen for 1 hat (a) 623, (b) 673 and (c) 723 K. [q.s.r. = 77 K (full lines), 293 K (dashed lines).] pairs of spectra. The spectral variation is a function of the temperature dependence of individual spectra of species formed at the same reduction temperature and also a function of the large variation in their line shapes on increasing the reduction temperature.The most remarkable resonance in this series is a single line (signal G) with g = 1.923, peak-to-peak width AHpp = 40 G [fig. 3 ( c ) ] , which becomes slightly asymmetric at 773 K. ( d ) The PMo,,/Al,O, sample behaves differently from both the unsupported and silica-supported PMo,, samples (fig. l), as found during the dehydration studies., (e) A change in the reduction medium (hydrogen or carbon monoxide) had no significant effect on the type of Mo5+ e.s.r. signal observed, with the exception that at high reduction temperatures the gross changes in spectra experienced for hydrogen reduction (fig. 3 ) were not observed for the samples reduced with CO.Oxygen Adsorption Adsorption of oxygen on unsupported SiMo,,-HPA produced no oxygen radicals detectable by e.s.r. spectroscopy at either room temperature or 77 K, after varying the pressure of oxygen and the reduction temperature between 673 and 773 K. In contrast, the SiMo,,/SiO, sample showed an e.s.r. signal attributable to an oxygen radical (gl = 2.015, g, = 2.009 and g, = 2.004), the resonance intensity depending upon the experimental conditions. The signal becomes broadened in the presence of an excess276 E.S.R. Studies of Heteropoly Acids 91 1 20 G - ' g3 Fig. 4. SiMo,,/SiO, (5% HPA), 0; signal recorded after prereduction at 773 K with 30 Torr CO and adsorption of 30 Torr 0, (the excess of 0, has been pumped off). (c.s.r. = 77 K.) I 5 O L," 1 I I I I I I 1 I a I l I I 673 773 672 773 673 773 reduction temperature/K Fig.5. Intensity I of the 0; radical signal plotted against temperature of prereduction with CO: (a) SiMo,,/SiO,, (b) SiMo,,/Al,O, and (c) Mo/SiO,. Different curves represent changes in the pressure of adsorbed 0,: x 5; 0, 30 and 0, 100 Torr. Note different calibration for (c). of oxygen, but is well resolved after evacuation (fig. 4). In the case of SiMo,,/Al,O, a similar signal was observed for which g, = 2.016, g, = 2.010 and g , = 2.005. Both sets of parameters clearly identify the radical as O,., Quantitative comparisons of resonance intensities were obtained for the three samples studied under identical conditions (fig. 5). The results show that the two silica-supported samples have the same tendency [fig.5(a) and (c)] to exhibit a marked increase in the 0; concentration when the samples are prereduced at higher temperatures. A markedR. Fricke and G . Ohlmann 277 influence of oxygen pressure was only observed for SiMo,,/SiO,. In contrast, for SiMo,,/Al,O, the 0; concentration was generally low and almost independent of the prereduction conditions or the oxygen pressure [fig. 5 (b)]. The formation of 0; species was also observed for PMo,,/SiO, samples. In all cases the Mo5+ signal observed after prereduction did not change during oxygen adsorption. The formation of 0- radicals, which sometimes accompanies the appearance of 0; species on transition metal catalysts, was not observed for these heteropoly acid samples. Discussion H, or CO Reduction The Mo5+ e.s.r.signals observed for samples after reduction in hydrogen or carbon monoxide are similar in some respects to those obtained during vacuum dehydration of HPA., Thus reduction also causes a transition temperature at ca. 673 K, associated previously1 with signal B and with irreversible destruction of the Keggin-anion structure. The intensities of resonances produced after reduction were, however, 5-10 times greater than after vacuum dehydration. The colour changes of samples also occurred in the same order: yellow-green --+ brown-green --+ blue --+ black (black was not observed for the vacuum-treated HPA). However, H,- or CO-reduced HPA had already changed colour to blue at ca. 550-600 K, 150 K lower than was observed for vacuum-treated HPA.l This transition temperature compares favourably with t.p.r. results which show that hydrogen uptake begins at ca.550-600 The Mo5+ spectra obtained appear similar to several reported by Konishi et aL6 after various stages of reduction at 523 K in a hydrogen atmosphere, although g values for the species were not given. On the other hand, the Mo5+ resonance observed by Otake et al.’ for a PMo,,-HPA sample, reduced at 553 K in hydrogen, could not be reproduced even for reduction in a flow system. Infrared spectra have shown that the ternary Mo-0 bond of the MOO, octahedron is relatively stable during the reduction of K,PMo,,O,, with hydrogen.8 In contrast, the intensities of Mo-0-Mo infrared bands at 800 and 870cm-l have been shown to decrease as the reduction process progressed.This phenomenon has been confirmed for the PMo,,-HPA sample discussed in these s t ~ d i e s . ~ It is therefore suggested that signal F, observed after reduction in hydrogen (or after vacuum treatment) is due to Keggin anions which have lost some of their bridging Mo-0-Mo oxygens. The large Ag value (gl-gll = 0.106) suggests a relatively large distortion of the MOO, octahedra, while the Keggin-anion structure is not irreversibly destroyed. This is in accord with the X-ray diffraction analysis by Eguchi8 of K,PMol,O,,, reduced at 673 K in hydrogen, reported to have the same crystal structure before and after reduction. In addition, the absence of signal F when reducing PMo,,-HPA, already destroyed by air treatment at 773 K, confirms the above assignment of this signal.An increase in the reduction temperature between 623 and 773 K produced marked changes in the e.s.r. spectra, shown in fig. 3. The greater the degree of reduction of the sample, the smaller the linewidth of the relevant signal. This suggests that as the anion structure is destroyed during the reduction process, the exchange interactions between neighbouring Mo5+ ions become more probable, leading to the decrease in linewidth. As depicted in fig. 3, the interaction depends upon the measurement temperature, producing a variation in e.s.r. linewidth. The temperature dependence of the spectra observed for the sample reduced at 623 K [fig. 3 (a)] cannot be explained on this basis alone. Furthermore, spin-lattice relaxation effects can be excluded because the e.s.r.spectral intensities vary with temperature approximately according to the Curie law. It is therefore assumed that the structure278 E.S.R. Studies of Heteropoly Acids Table 1. Mo5+ signals obtained after various treatments of PMo,,-HPA with a probable assignment to the HPA structure signal g l gii origin A* A B C D E F G H 1.957 1.948 1.930 (44. O)u 1.940 1.944 1.958 (45.6)a 1.966 1.95 1 1.93 1 1.864 1.873 (93.6)b 1.895 1.917? 1.852 (101.6)b 1.893 1.923 electron hopping Keggin anion dehydrated but undestroyed irreversible destruction of the Keggin-anion final destruction product (MOO,) structure reduced Al-0-Mo 'phase' (?) loss of constitution water, not irreversibly strong distortion, loss of bridging oxygen destroyed destroyed anion structure, interacting Mo5+ unknown a A l ; A,, (in G).produced on extraction of bridging oxygens in the first reduction step and terminal oxygens in the second step is stable, and can be changed easily by decreasing the measurement temperature to 77 K. The g values of the Mo5+ resonances observed under oxidisingl and reducing conditions, together with their proposed assignments, are listed in table 1 . It must, however, be clearly understood that the appearance of one or more e.s.r. signals does not alone constitute conclusive evidence for the presence of an HPA. Oxygen Adsorption Oxygen radicals were not observed after oxygen adsorption on the unsupported SiMo,,-HPA samples. Variation in the conditions of prereduction with CO or during the adsorption of oxygen did not change the result.In contrast, the supported HPA is able to stabilise 0; species in high concentrations, provided that the samples have been prereduced at 723 K or above (fig. 5). In the region of prereduction temperature in which the Keggin-anion structure is at least partly maintained, 0; species were again not detected. One may conclude that the distorted but undestroyed Keggin anion is unable to form, or at least to stabilise, oxygen radicals. Two mechanisms may be responsible for this behaviour. (a) It is accepted3 that, for supported transition-metal catalysts, the transfer of an electron from the metal atom to the adsorbed oxygen molecule may proceed according to the following scheme: M(n-l)+ + O,(ads) + Mn+ + O;(ads) requiring a special coordination symmetry around M which is probably tetrahedral.Low-temperature prereduction does not obviously lead to such coordinatively unsaturated adsorption sites of the Keggin anion at which molecular oxygen could be adsorbed and transformed into 0; species. The similar behaviour of the SiMo,,/SiO, and the monolayer Mo/SiO, catalysts with respect to oxygen adsorption would appear to confirm this mechanism. (b) If oxygen were adsorbed on the preserved Keggin anion it could be quickly incorporated into the lattice according to the following scheme : O,(gas) 4 O;(ads) -+ 0-(ads) -+ 02-(lattice)R. Fricke and G. Ohlmann 279 without stabilisation of 0- or 0;. This rapid electron transfer may well be an important mechanism for the unsupported HPA, as also noted by Akimoto et ai.,l0 but it has limited applications to the case of supported HPA.Dismissing a discussion of the properties of oxygen radicals, which are irrelevant in this context, it is possible to conclude in general that the detection of the 0; radicals is closely linked to the structure of the HPA. Thus it is suggested that the existence of 0; may be taken as evidence for the destruction of the Keggin-anion structure. For alumina-supported HPA the concentration of 0; species is relatively low and almost independent of the experimental conditions. It is therefore assumed that on SiMol,/A1203 the radicals originate from small parts of the sample which characterise neither the Keggin structure nor the Al-0-Mo ‘phase’ suggested in Part 1 (see table 1 of the present paper).This phase is obviously unable to form or stabilise oxygen species under the conditions used in these studies. Influence of the Support It is now possible to draw two conclusions regarding the influence of the supports on HPA from the results presented here and in Part 1.l (a) Under almost all of the heat-treatment conditions studied, the aerosil-supported HPA behaves in a manner very similar to that of the unsupported, powdered HPA. In accordance with other studiesll it would appear that the HPA is not chemisorbed by the aerosil, which serves merely as an inert support, leaving the HPA undistorted. After having lost almost all of its water of crystallisation, the supported HPA becomes more strongly chemisorbed on the aerosil surface, with a consequent increased interaction between the surface and the HPA.This interaction accelerates the destruction of the Keggin anion on the support surface, as shown by the results in fig. 3 of ref. (1) and concluded from the results of d.t.a. and u.v.-visible spectroscopic measurements.11 This, however, is in contradiction to the proposals of Chumachenko et a1.,12 who have suggested that the silica support should stabilise the HPA structure during thermal treatment. (b) In contrast, for alumina-supported HPA, the e.s.r. spectra of PMo12/Al,03 samples show no analogies with PMo12-HPA or with PMo12/Si0,, especially when comparing the temperatures at which characteristic signals appear. Typical transitions between resonances were obtained in some cases, but under different conditions of heat treatment.Thus signal C clearly dominates the air-treated samples, whereas signal D is characteristic for high-temperature reduced samples. At low temperatures signal A* indicates an electron-hopping process. Jerschke~itzl~ has observed that impregnation of alumina with up to 13.5 % HPA, immediately leads to a strong interaction between HPA and the support. In this context the appearance of signal A*l is taken as evidence that, assuming the sample is not heated at high temperatures, the Keggin structure is preserved after impregnation. Under oxidising conditions the HPA may be easily destroyed at low temperatures, as shown by the appearance of signal C. At elevated temperatures a new HPA ‘phase’, represented by signal D, is formed. As signal D is formed exclusively on the alumina-supported material, it must be assigned to the formation of a reduced Al-0-Mo phase.Evidence to support this supposition is afforded by the results from the oxygen-adsorption experiments, which again indicate a strong interaction between the alumina and the HPA. As this conclusion is in disagreement with the results of Eguchi et aZ.14 for reduced but unsupported PMl,-HPA, further studies of the origin of signal D would appear to be necessary. We thank Dr A. Ellison, Humberside College of Higher Education, Hull, for valuable discussions and kind support. Thanks are also due to Dr H-G, Jerschkewitz for advice and for supplying the HPA samples and Dr U. Ewert, Center for Scientific Instruments, for making his computer program COMPAR available. The technical assistance of Mr U. Marx is gratefully acknowledged.280 E.S.R. Studies of Heteropoly Acids References 1 (Part 1) R. Fricke and G. Ohlmann, J. Chem. SOC. Faraday Trans. 1, 1986,82, 263. 2 R. Fricke, W. Hanke and G . Ohlmann, J. Catal., 1983, 79, 1. 3 M. Che and A. J. Tench, Adv. Catal., 1983, 32, 1. 4 K. Katamura, T. Nakamura, K. Sakata, M. Misono and Y. Yoneda, Chem. Lett., 1981, 89. 5 S. Yoshida, H. Niiyama and E. Echigoya, J. Phys. Chem., 1982, 86, 3150. 6 Y. Konishi, K. Sakata, M. Misono and Y. Yoneda, J. Catal., 1982, 77, 169. 7 M. Otake, Y. Komiyama and T. Otaki, J. Phys. Chem., 1973, 77, 2896. 8 K. Eguchi, Y. Toyazawa, K. Furuta, N. Yamazoe and T. Seiyama, Chem. Lett., 1981, 1253. 9 E. Schreier, unpublished results. 10 M. Akimoto, Y. Tsuchida, K. Sat0 and E. Echigoya, J . Catal., 1981, 72, 83:. 11 H-G. Jerschkewitz, E. Alsdorf, H. Fichtner, W. Hanke, K. Jancke and G. Ohlmann, 2. Anorg. Allg. 12 N. N. Chumachenko, T. M. Yurieva, D. V. Tarasova and G. I. Aleshina, React. Kinet. Catal. Lett., 13 H-G. Jerschkewitz, unpublished results. 14 K. Eguchi, N. Yamazoe and T. Seiyama, Chern. Lett., 1982, 1341. Chem., 1985, 526, 73. 1980, 14, 87. Paper 5/963; Received 7th June, 1985