J. Chem. Soc., Faraday Trans. I, 1987, 83 (lo), 3115-3128 Electron Spin Resonance Studies of Free and Supported 12-Heteropoly Acids Part 6.-The Investigation of Reduced H,(SiW,,O,,) a x H,O and Ag,(SiW,,O,,) * x H 2 0 and the Effects of Oxygen Adsorption Rolf Fricke, Hans-Georg Jerschkewitz and Gerhard Ohlmann* Central Institute of Physical Chemistry, Academy of Sciences of the GDR, DDR- 1199 Berlin- Adlershof, German Democratic Republic A number of W5+ e.s.r. signals have been observed when reducing 12- tungstosilicic acid and its silver salt with hydrogen or methanol up to 873 K. By means of various adsorption measurements and by comparison with results of molybdenum-containing heteropoly acids, a more detailed conclusion concerning the state of the samples could be drawn.The adsorption of oxygen, leading to the formation of 0; radicals, was shown to proceed on decomposed Keggin anions only. A strong interaction between the acid or silver salt and alumina was found to promote de- composition. Samples used as catalysts in the conversion of methanol showed only a narrow signal due to paramagnetic coke residues, possibly because the catalyst, partly reduced in the course of the reaction, is immediately reoxidized in air. Compared to methanol, the use of hydrogen facilitates reduction of the acid and the silver salt. During the last decade an increasing number of papers has been published, showing that heteropoly acids (abbreviated as HPA) lend themselves as catalysts for various catalytic reactions. Among these HPAs those containing molybdenum and/or tungsten and some of their salts have shown the most promising catalytic properties.'-' For this reason investigations have been extended to the study of properties which are related to the catalytic activity, such as for example, dehydration, thermal stability, acidity etc., applying a great variety of physical methods, including i.r.'-ll and PAS-F.t.i.r.12 spectroscopies, d.t.a./t.g.,?l3 t.p.d.and t.p.r.14-16 and, to a limited extent, e.s.r. spectroscopy. 7-19 In addition, it has been shown that tungsten-containing HPAs and their silver or ammonium salts are capable of transforming methanol to hydrocarbons3* '* 20-22. As far as e.s.r. spectroscopy is concerned, tungsten is an unfavourable element. Compared to molybdenum the six-valent state, which is diamagnetic and therefore not detectable by e.s.r., is very stable,,? 7 9 l9 i.e.its reducibility in hydrogen or methanol, at temperatures where catalytic reactions proceed (below 773 K), is relatively low. In addition, it has been that the spin-lattice relaxation time for 5d and 4d ions (W5+ is 5dl) is relatively short compared to 3d ions, so that observation of their e.s.r. signals is possibly difficult except at low temperatures. This may well be the reason why only few observations of W5+ e.s.r. signals have been reported in the literature : thus for 1Ztungstophosphoric or 12-silicic acids, the work of Prados and Pope17 and of Saidkhanov et aL6 represent the only known examples, both reporting W5+ signals from reduced HPAs in solution. Of course the reduction of HPAs is not just merely a demanding test of the e.s.r.method, as reduced HPAs can favourably increase the activity and stability of catalysts 31 1531 16 E.S.R. of 12-Heteropoly Acids in the conversion of methan01.~~~~ l1 The present contribution continues a series of e.s.r. studies on H,+,(PV,Mo,,.,O,,) (n = 0.. .3) heteropoly acids,lg' ,** 25 with tungsten- containing samples. Experimental Preparation Dodecatungstosilicic acid has been prepared from the appropriate solution of sodium silicate, sodium tungstate and hydrochloric acid according to the method well known in the literature.26 The product was cleaned by twofold extraction with ether followed by crystallization of the acid from aqueous solution. The water content was determined analytically to be 20-22 mol/mol.The silver salt [Ag,(SiW,,O,,)] was prepared by dissolving the stoichiometric amount of silver carbonate in a weak aqueous solution of the acid. After 24 h storage of the solution in a refrigerator the fine crystalline precipitate was separated by filtration. The supports used for the present investigations were Degussa silica, Aerosil-50 (Ox 50), specific surface area S = 50 m2 8-l: and Degussa Alumina-C, S = 100 m2 8-l. The preparation of the supported catalysts has been carried out in the same way as described for the PMo12/support sample^,'^ except that both supports were dehydrated for 3 h at 873 K before impregnation with the HPA. Only the alumina-supported silver salt has been prepared under different conditions. To obtain these samples, silver carbonate was dissolved at 333-343 K in an aqueous solution of the acid.The maximum amount of the solution was chosen to be that which could still be re-adsorbed by the alumina. The limited solubility of the silver salt required prompt working utilizing super- saturation conditions. Catalysts containing up to 13.5 wt. % W could be obtained in this way in one step. Higher concentrations required sample drying and repetition of the procedures described above. The samples are designated as SiW12-HPA and AggSiW12 for the unsupported HPA and its silver salt, respectively. The supported catalysts are additionally annotated by the support used (e.g. SiW12/Si02) and the concentration of the active component is given when necessary. E.S.R. Measurements The reduction of catalysts was carried out under static conditions in the e.s.r.sample tube. In general, 30 Torr (1 Torr = 101 325/760 Pa) of H, or 100 Torr of CH,OH were used for reduction. Owing to the observed instability of signals under vacuum the catalysts were not evacuated after reduction. The time of reduction was 1 h at each temperature, at 50 or 100 K intervals between 293 and 773 K (and 873 K in a few cases). Exposure of the reduced catalysts to the atmosphere was avoided. Oxygen adsorption (30 Torr) at room temperature was performed on catalysts pre- reduced with hydrogen at the temperatures given in the text and the excess oxygen was subsequently pumped off. E.s.r. measurements were performed at 77 K and room temperature with a ZWG ERS 220 spectrometer operating at X-band.An n.m.r. marker and a DPPH sample (g = 2.0036) were used for magnetic field calibration (the field values are given as 1 G = lo-' T) and for calculation of the signal intensity. The main signals obtained (signals 3-5) were computer-simulated using the program cOMPAR. 27R. Fricke, H-G. Jerschkewitz and G. Ohlmann I DPPH 1 31 17 g, 9 2 % m ;-,..'-t, 1 a , a . I . ; , . , ' .. I , , I I , ' I * I ( . I , b , ' 0 Fig. 1. E.s.r. Results Sample Treatment in Air After preparation, neither the HPA nor the Ag salt showed any e.s.r. signal. The samples were colourless or pale yellow,' sometimes also tending to greyness (Ag salt). They retained this colour when treated in air up to 773 K and no sample showed an e.s.r. signal, whether unsupported or supported on SiO, or A1,0,.This behaviour is different from that of PV,Mo,,., - HPAs, which showed a Mo5+ ( n = 0) or V4+ ( n = 1-3) signal immediately after preparation, the intensity of which did not decrease to zero even after treatment at 773 K in air or ~ x y g e n . ' ~ ~ ~ ~ Clearly, the well known stability of the six-valent state of tungsten in these conditions means that, in contrast to the PV,Mo,,., samples, the dehydration process for the current tungsten sample series could not be studied using e.s.r. spectroscopy. Sample Reduction with Hydrogen or Methanol The justification for studying the reduced state of the SiW,,-HPA and its Ag-salt is at least twofold: (a) it is known from X-ray and i.r. investigations that the Keggin anion structure is, at the most, only slightly influenced by reduction;2v28 (b) catalytic investigations have shown that pre-reduction of the silver salt influences the activity as well as the stability of the catalyst.'*'' The reduction which was performed, usually up to 773 K and in very few cases also at 873 K, changed the sample colour from white-pale yellow to blue-greyish, dark blue and black at higher temperatures.A number of e.s.r. signals could be observed, whose shape and parameters depended upon the conditions of reduction and the sample composition. The first signal, which appeared after a brief reduction at 323 K of the SiW,,-HPA, is signal 1 (fig. 1) with the following parameters: g , = 1.944, g,, = 1.901, A , = 41.8 G, A , , = 91.2 G. The hyperfine structure was usually not well resolved but could, however, be evaluated in a few cases which are not shown here in detail.In all of the cases investigated, signal 1 was observed31 18 E.S.R. of 12-Heteropoly Acids Fig. 2. E.s.r. spectra of SiW,,-HPA, reduced in H, at 573 K as a function of the time of reduction q.s.r. = 77 K. (a) 5 min, (b) 1 h, (c) 4 h. to be temperature-independent and was usually the first e.s.r. signal to appear when increasing the reduction temperature. A second signal, designated as signal 2 and also shown in fig. 1, was always of low intensity and was absent at room temperature. It was probably present for all samples, but was not resolved for the supported samples. The g values for signal 2 were calculated as: g, = 1.830, g, = 1.802, and g, = 1.786, being very similar to a low-temperature W5+ signal of reduced polytungstates described by Prados and Pope.” Fig.2 shows the sequence of e.s.r. spectra taken after reduction (H,) at 573 K as a function of time. Signal 1 is still present, but a new signal (signal 5 ) now appears, of comparable intensity. It is characterized by a g, value of 1.771. The g,, value could not be fixed owing to variation of the resonance position, depending upon time and conditions (see amplified dotted lines of the signals in the high-field part of the spectra). Details of this complication are shown for the spectrum recorded after 5 min reduction. Although the parallel component is clearly resolved, showing two different positions at g,, = 1.557 (5*) and g,, = 1.608 (9, the perpendicular component presents itself still as a single line.In a few cases, however, a shoulder was observed, or even an indication of splitting of the perpendicular component. Therefore, a reasonable explanation is that signal 5 is a superposition of signals, with g, values very close to each other (with these peculiarities in mind the signals will still be designated by signal 5, further on in the text). After 4 h reduction several new lines could be observed, one of which was identified asR. Fricke, H-G. Jerschkewitz and G. Ohlmann 31 19 Fig. 3. E m . spectra of SiW,,-HPA, reduced in H, for 1 h at: (a) 673 K, (b) 773 K, ( c ) 873 K. c.s.r, = 77 K (dotted lines show the room-temperature spectra). a new signal 3, which is better shown in fig. 3 and 4. Signal 5, which is the most characteristic signal for samples not supported on A1,0,, is sometimes visible also at room temperature; its intensity is variable, however, as is the intensity ratio 1(77 K):Z(295 K).Increase of the reduction temperature above 573 K, up to 873 K, produced the set of characteristic spectra shown in fig. 3. After reduction at 673 K, when the sample colour is already black, a better-resolved signal 3 is observed, together with signal 5 (or 5*). The resonance at g = 1.504 is not assumed to be a component of signal 5, as it occurs independently of signal 5, although always with low intensity. A sharp and narrow signal at g = 2.004, signal 9, could also be observed increasing in intensity as the temperature increased. After 873 K reduction, both signals 3 and 5 / ( 5 * ) disappeared almost completely and a signal at g,, = 1.907 and g , = 1.678 (signal 7) characterises the high- temperature reduced state.Signal 7 is undetectable, however, at room temperature [fig. 3(c)]. Two variations of the results described above are shown in fig. 5, which presents the spectra obtained after 573 K reduction with CH,OH. Fig. 5(a) and (b) show a superposition of the signals 1, 3 and 5, already identified from fig. 2. However, on increasing the time of reduction up to 6 h two new additional signals appeared : a narrow line at ca. g = 2.00 (signal 9*) and a similarly narrow, axial signal with g1 = 1.744 and g,, = 1.607 (signal 6). As shown by the broken line in fig. 5(c), signal 6 is strongly temperature-dependent and is not visible at room temperature.3120 E.S.R.of 12-Heteropoly Acids Fig. 4. E.s.r. spectra of an Ag,SiW,,/Al,O, catalyst (20 YO W) reduced with H,, 1 h at: (a) 373 K, (b) 473 K, (c) 573 K, ( d ) 673 K, (e) 773 K. q.s.r. = 77 K (dotted spectra were taken at room temperature). The signals described so far were observed more or less unchanged for all samples, provided that the active component was not supported on Al,O,. Fig, 4 gives the general view for the Ag,SiW,,/Al,O, catalyst taken as an example. At low temperature the picture is similar to the free acid or Ag salt (including the silica-supported samples), showing signals 1 and 3 superimposed. After 773 K reduction, however, a superposition of signal 3 with a new signal is obtained, designated as signal 4, with g , = 1.783 and g,, = 1.676.In all experiments, signal 4 was only observed for alumina-supported samples, regardless of whether the acid or the silver salt had been reduced. In contrast to signals 2 and 3 signal 4 could be observed at both low temperature and at room temperature (see the dotted line in fig. 4). For reasons of completeness, it should additionally be noted that another signal, not separately shown in one of the present figures, appeared at ca. g = 4.2 (signal 8) when studying the alumina-supported samples. This signal is well known from the literature and is usually assigned to Fe3+ ions in tetrahedral coordination, caused by impurities in the support. Adsorption of Oxygen These experiments were undertaken because recent results obtained from supported SiMo,,-HPAlg and PV,MO,,-,-HPAS~~ have shown that, if taken with necessaryR.Fricke, H-G. Jerschkewitz and G. Ohlmann 3121 Fig. 5. E.s.r. spectra of a SiW,,/SiO, catalyst (20% W) reduced with CH,OH at 573 K as a function of time of reduction, q+s-r. = 77 K (dotted spectrum is taken at room temperature). (a) 5 min, (b) 1 h, (c) 6 h. DPPH g1 I I g3 Fig. 6. 0; e.s.r. spectrum of an SiW,,/SiO, catalyst (20% W) (prereduced at 723 K with H, and 25 Torr 0, adsorbed at room temperature), q.s.r. = 295 K.3122 A 7 / / / E.S.R. of 12-Heteropoly Acids A 12 n c U .r( d 2 8 - g - U n 4 4 - - B !"i I '\ I \ I \ I \ \ I ' I 1 \ I \ I \, ( b ) I I 373 573 773 TIK Fig. 7. A, Amount of coke [I(C'), in arbitrary units] us. the time of methanol conversion derived from: (a) Ag,SiW,,/A1,03 (20% W), e.s.r. signal intensity I; (b) Ag,SiW,,/SiO, (20% W), e.s.r.signal intensity I; (c) sample (a), chemical determination of carbon contents. B, e.s.r. signal intensity I(C') of a paramagnetic coke signal us. the temperature of reoxidation (1 h at each temperature). The catalysts have been used for 1 h in the conversion of methanol at 573 K. (a) Ag,SiW12/A1,0, (20% W), (b) Ag,SiW,,/SiO, (20% W), (c) WO,/SiO, (20% W), (d) WO,+Ag,/SiO, (20 % W), (e) WO,/Al,O, (10 % W), (f) Ag,SiW12, unsupported. caution, the results can give information not only on the properties of oxygen radicals formed under special conditions but also on the state and structure of the Keggin anion. Neither unsupported SiW ,,-HPA nor its silver salt ever generated stable oxygen radicals (0- or 02).This observation is in accordance with previous results on other unsupported H P A S ' ' ~ ~ ~ and supports the conclusions drawn in these papers that the inability of the HPAs to generate stable oxygen radicals is probably a result of fast electron recombination in the bulk of the Keggin anion. In contrast to the unsupported samples, a radical signal was observed after oxygen was adsorbed on supported catalysts, pre-reduced at 723-773 K in hydrogen (fig. 6). Comparison of the parameters of this signal (8, = 2.026, g , = 2.014, g , = 2.005) with the literature data2' allows this signal to be assigned to 0; oxygen radicals stabilized at W6+ sites. Adsorption of H, on a sample showing this signal caused no change, showing that 0- radicals, which react very quickly with H,, were not present on the surface of the catalyst.The Ag,(SiW,,O,] Catalyst after Use in the Methanol Conversion E.s.r. spectra were recorded for both the unsupported and the supported silver salt after being used in the conversion of methanol to alkenes at 573 K.* After reaction, the catalysts were stored in air for some hours. For all of the samples studied, only signalM rn 2 6 1 s cj % 9 8 - I I I 9 8 q 8 I € I I 9 8 I (M % oz) 'O~IV/"M!S'~V (M % OZ) zO!S/"M!S'~V "M!S 'SV (M O h 0 I ) cOzW/zlM!S ( M %OZ> zO!S/zTM?S VdH-"M!S (M % 02) E ~ Z ~ ~ / Z K ~ ~ ~ ' ~ ~ (M % 02-S) zO!S/zlM!S'%' zKM!S'SV (M % 0 I) 'OzIV/"M!S (M YoOZ) ZO!S/ZKM!S VdH-"MIS €11 EL9 u s ElP €L€3124 E.S.R. of 12-Heteropoly Acids 9* could be observed, the signal intensity depending upon the time of reaction [fig.7(a)]. The conditions under which signal 9* appeared, whether after reduction with methanol or after use in methanol conversion, clearly indicate that this signal originates from paramagnetic coke residues on the catalyst. Obviously, this signal does not represent all the coke present on the catalyst because, as shown in fig. 7(b), treatment in air at temperatures up to 773 K led to very pronounced changes in the signal intensity. Other supported tungsten catalysts containing WO, or a mixture of WO, and Ag,O were treated in the same way as the silver salt, and the following conclusions can be derived [fig. 7(b)]: (a) the most intense coke signals were found on silica-supported samples, showing maximum intensity at a reoxidation temperature of ca. 573 K ; (b) the silver component is responsible for the promotion of the coke formation, represented by signal 9*, because W/SiO, and W/Al,O, catalysts studied for comparison showed no signal.The absence of any signal for the unsupported Ag,SiW,, salt can be explained both by the absence of silica and the low surface area, which does not allow deposition of great absolute quantities of coke on its surface. Discussion Owing to the lack of any e.s.r. signal under oxidative conditions, it is impossible to discuss the dehydration process of SiW,,-HPA as has been recently accomplished for the PV,Mo,,-,-HPAs (n = 0-3).19* 24 In addition, the present studies show that the reduction of tungsten proceeds at temperatures above 473 K, i.e.temperatures where large amounts of the water of crystallization have already e s ~ a p e d . ~ , ~ For a better understanding of the state of the HPA under the conditions described in this paper it is necessary to elucidate the origin of the e.s.r. signals summarized in table 1 in terms of the type of sample and the temperature of reduction. Signal 1 A signal similar to signal 1 was found by Saidkhanov et aL6 for a SiW,,-HPA in solution and by other authors for various tungsten-containing catalysts,30’ 31 and is unanimously attributed to W5+ ions. It is, however, proposed from the present results, that this signal originates from Mo5+ impurities rather than from W5+ species, for the following reasons. The g-values for this signal are relatively too high to be attributed to the W5+ species.More conclusively, a detailed analysis of the hyperfine structure, clearly shows 6 h.f.s. lines for each signal component (h.f.s. separation is shown in table 2), whereas there should only be 2 h.f.s. lines per signal component for lE3W, for which I = 1/2 with 14.28 % natural abundance. Indeed, the presence of molybdenum in all samples, owing to Mo impurities in the WO,, has been proved by emission spectral analysis. HCl was adsorbed onto several samples in an attempt to produce ligand exchange which would be reflected in g-value inversion from g , > g,l to g, < g,l.32 No changes were, however, induced in signal 1, showing that the Mo5+ impurities were fully incorporated in the Keggin anion structure, rather than existing at the ‘surface’.Signal 2 In an early paper, Prados and Pope17 obtained a very similar signal, with slightly higher g, and g, values, from various frozen solutions of one-electron, electrochemically reduced pofytungstates, and attributed the signal to electrons trapped on rhombic tungsten centres. Owing to the observed similarity of the g values and the conditions of generation (one-electron redu~tion’~ and low-temperature reduction in our case) this signal is assumed to originate from the same type of tungsten ions. The signal was foundR. Fricke, H-G. Jerschkewitz and G . Ohlmann 3125 Table 2. E.s.r. signals observed after reduction (H, or CH,OH) of SiW,,-HPA and its silver salt (unsupported and supported on SiO, or A1,0,) with the possible assignment to the HPA structure as discussed in the paper signal g , gll origin 1 1.944 1.901 impurity of Mo5+ ions 2 1.830 1.802 1.786 W5+ (electrons trapped on rhomb.W centres") 3b 1.838 1.743 W5+ of Keggin anions (KA) fragments (incompletely 4b 1.783 (1.676) W5+-O-Al ' phase ' 5' 1.771 W5+, KA strongly distorted 5** 1.771 :%I} but undestroyed 6 1.744 1.607 unknown 7 1.678 1.907 W5+ of destroyed KA 8 ca. 4.2 Fe3+ impurity of A1,0, support 9 2.00 lattice defect of KA 9* 2.00 paramagnetic coke residue (41.8)" (91.2)" crystallized regions) " H.f.s. splitting. reduction). Spectra ~imulated.,~ Parallel components derived from fig. 2 (5 min to be of low intensity at 77 K but, owing to the strong temperature-dependence of this signal,17 it can be expected that its intensity would increase drastically at temperatures lower than 77 K.Signal 3 This signal is especially well characterised by its g-values and by the behaviour of the signal on the adsorption of different species. The signal disappeared when oxygen or air was adsorbed, suggesting that the appropriate species were surface ions. The signal which was only observed at low temperature (77 K), disappeared when the sample was evacuated after reduction. Adsorption of H,O, CH,OH, 0, or air regenerated the original signal and for the last two gases, an 0; signal was superimposed. (However, observation of the signal requires removal of the excess oxygen.) To our knowledge this signal has not been described before, perhaps because of its instability. According to the g-values, the signal can be attributed to a dl species, i.e.W5+ ions in the present case. From the temperature-dependence of the signal and the sample behaviour under vacuum it can be deduced that the spin-lattice relaxation time is short, pointing to a rather symmetric coordination sphere which is further enhanced on sample evacuation (and, vice versa, decreased by adsorption). One may therefore conclude that signal 3 originates from low (possibly tetrahedral) coordination W5+ surface ions, which were not incorporated in the Keggin anion structure and which probably belong to incompletely crystallized regions of the HPA. Signal 4 Because this signal was only observed on alumina-supported samples, it may be concluded that it is the special interaction between the HPA (or its silver salt) and A1,0, which defines the signal.It is well known from e.s.r. investigations of PV,MO,,~,/A~,O,~~~~* or ESCA studies of WO,/AI,O, catalyst^,^^,^^ that the inter- 103-23126 E.S.R. of 12-Heteropoly Acids action of the active components with the alumina support is rather strong, giving rise to facile decomposition of the Keggin anion s t r u c t ~ r e . ~ ~ ~ 24 Analogously, this is also suggested here for the reduced SiW,,/Al,O, sample (or the Ag-salt/Al,O,), especially as infrared studies have shown that destruction of the Keggin anion under oxidizing conditions is already complete at 613 K (and at ca. 660 K, respectively). l3 It is reasonable to expect formation of A14(SiW1204,,)3 salts under the present conditions, because both the SiW,,/Al,O, and the aluminium salt showed similar properties in the catalytic conversion of methanol as well as in i.r.and Raman ~pectroscopy.~~ On the other hand, for WO,/Al,O, samples, Salvati et aZ.,, did not find any evidence for the appropriate compound Al(W0,). E.s.r. measurements of the unsupported salt as well as silica-supported A14(SiW1204,,), showed signal 5 rather than 4 after stepwise reduction in hydrogen up to 773 K. Therefore, signal 4 is, with the necessary caution, attributed to a ' W5+-O-Al phase' from the destroyed SiW,,-HPA and not to the formation of the aluminium salt. Signal 5 This is the most characteristic signal of the free SiWl,-HPA and its silver salt, as well as of the silica-supported catalysts. There is sufficient information available concerning the stability of the Keggin anion structure of SiW,,-HPAs and of PW,,-HPAs, to show that the anion structure of these unsupported acids is stable up to ca.773-823 K.,, '9 16$ l3 It is possible to conclude that signal 5 is, in addition to signal 2, a further W5+ signal representing the undestroyed Keggin anion, which appears at conditions of increased reduction (i.e. at higher temperatures than signal 2). The coordination sphere is rather stable because adsorption of H,O, CH,OH or 0, did not influence signal shape or intensity. Prolonged evacuation led to decreased intensity, observed especially at room temperature, which was, however, reversed after the re-adsorption of H,O, CH,OH or 0,. The formation of 0; radicals was not observed for samples showing signal 5.It is suggested that these W5+ species are coordinatively saturated under the conditions specified, with approximately distorted octahedral or square-pyramidal symmetry, because oxygen is unable to enter the coordination sphere to broaden the signal or to form oxygen radicals. The changes observed during evacuation can be attributed to smaller structural variations of the W5+ coordination state which influence the spin-lattice relaxation time to a certain degree. A similar W5+ signal (8, = 1.767, g , , = 1 S89) described by Abdrakhmanov et al.35 has no relationship to the HPA structure because it originates from [WOF5I2- species in frozen solution. It should be mentioned that there exists some similarity between the signal 5 described here and a Mo5+ signal (signal F) observed for reduced PMo,,-HPA, the latter being attributed to highly distorted but undestroyed Keggin ani0ns.l' To explain the appearance of the different g values (signals 5 and 5*), further investigations will have to be made.Signals 6 and 7 The origin of these signals, and especially of signal 6, is not fully clear. However, because signal 7 could only be obtained after reduction at temperatures as high as 773-873 K, it is suggested that it arises from W5+ ions in a destruction product of the Keggin anion. No further information is available at the moment.R. Fricke, H-G. Jerschkewitz and G. Ohlmann 3 127 Signals 9 and 9* Both signals have approximately the same g value of 2.00, their lines are very narrow and were obviously caused by radicals.Signal 9 is only observed after high-temperature reduction with hydrogen and can be attributed to paramagnetic lattice defects of the Keggin anion or to its destruction products. It is well known from the literature36 that reduction of catalysts with hydrocarbons generally produces coke, which is often paramagnetic and shows an e.s.r. signal. This is also the case for the HPAs reduced by methanol or following use in the catalytic conversion of methanol, thereby generating signal 9*. It is possible that signal 9 may be superimposed but cannot be distinguished, while the different origins of the signals are obvious. Conclusions The present e.s.r. studies have shown that, besides the W5+ signal already described by Prados and Pope,17 additional characteristic W5+ signals could be observed and described, representing the state of the SiW,,-HPA (table 2).The conditions of formation of the 0, oxygen radicals allow some further conclusions. As shown in more detail for the supported SiMo,,-HPA,ls the undestroyed Keggin anion is unable to stabilize oxygen radicals. This property of HPAs can be confirmed here also for SiW,, -HPA, supporting the conclusion above, that signals 3 and 4 represent only anion fragments, whereas signal 5 represents the distorted but undestroyed Keggin anion. For reasons of completeness it should also be mentioned that, in addition to the signals summarized in table 1, several peaks or shoulders have also been observed [see for example, fig. 2 (4 h spectrum) and fig. 3(a)] which obviously also arise from paramagnetic tungsten species, It was impossible, however, to correlate meaningfully the appearance of these signals with the type of sample or with the conditions of measurement and therefore they have not been discussed in this paper.The appearance of W5+ signals shows that reduction with hydrogen starts at ca. 473 K. This is a lower temperature than is usually derived from t.p.r. meas~rements.~* l3? 1 4 7 l6 Comparing these results, one has however to take into consideration that e.s.r. spectroscopy is more sensitive than t.p.r., but inferences from both methods coincide at higher reduction temperatures. In accordance with recent t.p.r. studies13 it was found that the reducibility of the HPAs with methanol is lower than with hydrogen.Although the types of signals were the same in both cases (apart from signal 9*), they appeared, from a rough estimation, at a temperature ca. 100 K lower when hydrogen was used. Reduction was also observed to be more facile for the silver salt compared to the acid, an effect which has also been observed by t.p.r.'ll3 Though it is known that the Agf in the salt is easily reduced to metallic silver, no signal was observed which could originate from paramagnetic silver species or, with respect to tungsten, which could be interpreted as a specific influence of the silver on one of the W5+ signals. In connection with catalysis it is undoubtedly a disadvantage that the only signal which was observed after the catalytic reaction was that arising from paramagnetic coke residues.Akimoto et al.' have found that a partly reduced PW,,-HPA is immediately reoxidized by oxygen at room temperature and therefore it seems possible that reoxidation could be a probable explanation for the absence of any W5+ signal, especially as the present reduction studies with methanol have shown that reduction takes place at 573 K.3128 E.S.R. of 12-Heteropoly Acids The authors thank Dr Alan Ellison (Humberside College of Higher Education Hull) for kind support. Thanks are also due to Dr U. Ewert (Center for Scientific Instruments) for making his simulation program COMPAR available, Dr R. Matschat, for carrying out the emission spectral analysis, and K. Menning for providing a literature analysis of this series. The technical assistance of Mrs W.Ebert, I. Otremba, and Mr U. Marx is gratefully acknowledged. References 1 G. A. Tsigdinos, Top. Curr. Chem., 1978, 76, 1. 2 M. Misono, Proc. Climax 4th Int. Conf. Chem. Uses of Molybdenum, ed. H. F. Barry and P. C. H. 3 Y. Ono, T. Mori and T. Keii, Proc. 7th Int. Congr. Catal., Tokyo 1980 (Kodansha, Tokyo 1981), part 4 H. Hayashi and J. B. Moffat, J. Catal., 1982, 77, 473. 5 Y. Izumi, R. Hasabe and K. Urabe, J. Catal., 1983, 84, 402. 6 S. S. Saidkhanov, A. I. Kokorin, E. N. Savinov, A. I. Volkov and V. N. Parmon, J . Mol. Catal., 1983, 7 M. Akimoto, H. Ikeda, A. Okabe and E. Ekhigoya, J. Catal., 1984, 89, 196. 8 H. Ehwald, W. Fiebig, H.-G. Jerschkewitz, G. Lischke, B. Parlitz, E. Schreier, P. Reich and G. 9 M. B. Varfolomeev, V. V. Burljajev, T. A. Toporenskaja, H.-J.Lunk, W. Wilde and W. Hilmer, Z. 10 M. Misono, M. Mizuno, K. Katamura, A. Kasai, Y. Konishi, I. Sakata, T. Okuhara and Y. Yoneda, 11 T. Baba, H. Watanabe and Y. Ono, J. Phys. Chem., 1983,87, 2406. 12 J. B. Moffat and J. G. Highfield, Catal. Energy Scene, ed. S . Kaliaguine and A. Mahay (Elsevier, 13 H-G. Jerschkewitz, B. Parlitz and E. Schreier, unpublished results. 14 S. Yoshida, H. Niiyama and E. Echigoya, J. Phys. Chem., 1982,86, 3150. 15 B. K. Hodnett and J. B. Moffat, J. Catal., 1984, 88, 253. 16 B. K. Hodnett and J. B. Moffat, J. Catal., 1985, 91, 93. 17 R. A. Prados and M. T. Pope, Znorg. Chem., 1976, 15, 2547. 18 V. F. Cuvaev, T. A. Karpukhina, G. K. Mailieva, K. 1. Popov and V. I. Spicyn, Zzv. Akad. Nauk, 19 R. Fricke and G. Ohlmann, J. Chem. Soc., Faraday Trans. I , 1986, 82, 263; 273. 20 T. Baba, J. Sakai, H. Watanabe and Y. Ono, Bull. Chem. Soc. Jpn, 1982, 55, 2555. 21 H. Hayashi and J. B. Moffat, J. Catal., 1983, 83, 192. 22 H. Ehwald, H-G. Jerschkewitz, G. Lischke and G. Ohlmann, Chem. Stosow., 1984, 28, 195. 23 J. E. Wertz and J. R. Bolton, Electron..Spin Resonance (McGraw-Hill, New York, 1972). 24 R. Fricke, H-G. Jerschkewitz and G. Ohlmann, J. Chem. Soc., Faraday Trans. I , 1986, 82, 3479. 25 F. Ritschl and R. Fricke, J. Chem. Soc., Faraday Trans. I , 1987, 83, 1049. 26 G. Brauer, Handbuch der Praparativen Anorganischen Chemie, 2. Auflage (F. Enke-Verlag, Stuttgart, 27 U. Ewert, Thesis (Humboldt University Berlin, 1979). 28 M. T. Pope, Heteropoly and Isopoly Oxometalates (Springer-Verlag, Berlin, 1983). 29 M. Che and A. J. Tench, Adv. Catal., 1983, 32, 1. 30 A. J. A. Konings, W. L. J. Brentjens, D. C. Koningsberger and V. H. J. de Beer, J. Catal., 1981, 67, 31 S. Sivasankar and A. V. Ramaswamy, 6th Nut. Symp. Rec. Adv. Catal. Catal. React. Eng., Pune (India) 32 R. Fricke, W. Hanke and G. Ohlmann, J. Catal., 1983, 79, 1. 33 L. Salvati Jr, L. E. Makovsky, J. M. Stencel, F. R. Brown and D. M. Hercules, J. Phys. Chem., 1981, 34 I. E. Wachs, C. C. Chersich and J. H. Hardenbergh, Appl. Catal., 1985, 13, 335. 35 R. S. Abdrakhmanov, N. S. Garifyanov, E. I. Semyonova, Z. Strukt. Khim., 1968, 9, 530. 36 J. M. Peacock, M. J. Sharp, A. J. Parker, P. G. Ashmore and J. A. Hockey, J. Catal., 1969, 15, 379. Mitchell (Climax Molybdenum Corp., Ann Arbor, 1982), p. 289. B, p. 1414. 21, 365. Ohlmann, Appl. Catal., submitted. Anorg. Allg. Chem., 1981, 472, 185. Bull. Chem. Soc. Jpn, 1982, 55,400. Amsterdam, 1984), p. 77. SSSR, Ser. Khim., 1981, 4, 717. 1962), p. 1489. 145. 1983, p. 95. 85, 3700. Paper 6/1814; Received 10th September, 1986