J. CHEM. SOC. FARADAY TRANS., 1994, 90(18), 2663-2670 EPR and NMR Studies of Amorphous Aluminium Boratest Simion Simon, Andre van der Pol, Edward J. Reijerse, Arno P. M. Kentgens, Geert J. van Moorsel and Engbert de Boer Department of Molecular Spectroscopy, Faculty of Science, University of Nijmegen , Toernooiveld 6525 ED Nijmegen, The Netherlands Amorphous aluminium borates, Al,,, -x,B,,O, with 0 d x d 0.5,prepared from mixtures of aluminium nitrate, boric acid and glycerol, have been studied by EPR and MASNMR as a function of composition and heat- treatment temperature (Tt < 860 "C). EPR studies showed the presence of physisorbed NO,, NO and 0, mol-ecules, produced by decomposition reactions during the thermal treatment. The 0, molecules in the gaseous state were observed in a narrow temperature interval around 60 K and in the condensed phase at low tem- perature (t20K).The D value for condensed 0, amounts to 109 GHz, significantly lower than the value for 'free' O,, which is 119 GHz. Above 20 K the NO, molecules in all samples rotate rapidly (2lo7 Hz) about an axis parallel to the interatomic oxygen-xygen direction ; this mobility decreases with increasing heat-treatment temperature. Some EPR lines were tentatively ascribed to pairs or clusters of the abovementioned paramagne- tic molecules. ,'At MASNMR studies showed the presence of six-, five- and four-coordinate Al atoms, their relative concen- trations being strongly dependent on the thermal history and composition of the samples. The fractions of tetra-and penta-coordinated Al atoms were maximum at heat-treatment temperatures between 300 and 600 "Cand decreased considerably after the samples were exposed to air.Therefore the low coordinated Al atoms are predominantly located at the surface. The decreased mobility of NO, molecules, at high treatment temperatures, indicates that NO, interacts strong- ly with the pore surfacewhen it contains a large fraction of four- and five-coordinate Al ions. Aluminas are extensively used as the supporting material in catalytic reactions. They are acid-base catalysts with a high surface area.' Successful efforts have been made to prepare amorphous aluminas that exhibit a zeolite-type porosity. The pore configuration and dimensions depend on composition, preparation method and heat treatment.'., The addition of typical glass-forming components such as SiO,, P,O, and B,O,, increases the stability of aluminas and leads also to new proper tie^.^-^ The incorporation of transition-metal ions or rare-earth metal elements gives these materials interesting optical, magnetic and catalytic proper- ties.'-' ' Amorphous aluminium borates with high surface areas can be prepared by a sol-gel method from solutions of aluminium salts and boric acid using ammonium hydroxide or methanol as precipitant.Recently a new method has been developed which also results in an amorphous material having a high surface area. This method involves the low- temperature thermal decomposition of aluminium nitrate and boric acid sustained by the simultaneous oxidation of a suit- able organic agent. Materials prepared this way were studied by thermal analysis methods, X-ray diffraction and FTIR spectro~copy.~~~In this paper we report EPR and NMR studies of aluminium borates prepared according to this method.EPR studies of aluminium borates revealed the pres- ence of physisorbed NO,, NO and 0, molecules which are produced by decomposition reactions during the thermal treatment. The mobility of the NO, molecules, as reflected in the EPR spectra, was strongly dependent on the measure- ment temperature, the heat-treatment temperature and the sample composition. At low temperatures, EPR spectra of gaseous 0, as well as for 0, in the condensed phase were observed. 27Al MASNMR studies revealed three signals at 6, 30 and 60 ppm, corresponding to six-, five- and four-coordinate Al, respectively.Their relative intensities were strongly depen- dent on the composition and the thermal history of the samples. The structural information obtained from 27Al ?This paper was presented at the 27th International ESR Con-ference at the University of Wales, Cardiff, 21st-25th March, 1994. MASNMR is used to explain the different strength of the interactions between the identified paramagnetic gaseous species and the active sites of the pore surfaces, developed during the thermal treatment. Experimental Aluminium borate samples were prepared with composition Al,~l~x~B,x03with x = 0, 0.1, 0.2, 0.3, 0.4 and 0.5.To a stoi- chiometric mixture of Al(NO,), * 9H,O and H,BO,, glycerol was added as an organic reducing agent (10 wt.% in all samples) and a small amount of distilled water. After the components had dissolved a single liquid phase was formed at room temperature. The clear solutions were heated to 95°C and after ca. 2 h spongy, bulky solid samples were obtained. At the end of the heating procedure decomposition reactions took place as apparent from emission of gaseous products. The conversion that takes place during the synthe- sis can be summarized as follows : 2[Al(NO,), .9H20] -+Al,O, + 6N02 + $0,+ l8H,O (1) 6NO, +6N0 + 30, (2) 2H3B03+B2034-3H20 (3) glycerol oxidation (4) EPR and NMR measurements were carried out on samples treated for 30 min at various temperatures (see Fig.1). For this procedure the solid material was crushed and placed in a cylindrical furnace for heating in the open air. Immediately after they had been heated the samples were sealed in quartz tubes for EPR measurements or placed in air-tight glass bottles for NMR measurements. Just before the beginning of the NMR measurements the samples were rapidly transferred to the spinners in order to keep hydration effects to a minimum. The solid samples were white for 100 d 17;l"C < 150, yellow-green for 150 < TJ"C < 200, yellow-brown for 200 d TJC < 300, and white for ?; =-300°C. The samples are denoted ABx-T,, where x refers to the boron content and y to the treatment temperature (y = TJlOO).2664 800 600 Y 400 2 4 6 8 tlh Fig. 1 Heat treatment diagram. The heating temperatures are denoted T,, where y = TJ100. EPR spectra were recorded on powdered samples on a Bruker ESP-380 X-band spectrometer at 5.8-300 K and at a static field between 0.05 and 13 kG. The average microwave frequency of the experiments was 9.3 GHz. MASNMR measurements were carried out at room tem- perature on a Bruker AM-500 spectrometer equipped with a solid-state accessory, using a home-built probe head equipped with a Jakobsen 5 mm MAS assembly. Usually 1 ps pulse excitations were applied and spinning speeds up to 14 kHz were employed. Spectra are referenced with respect to an external AI(NO3)3 solution [AI(H,0),3 '1.Results EPR on Aluminium Borate EPR spectra representative for various stages of the synthesis process are shown in Fig. 2-7. All samples exhibited a rela- tively small EPR line at ca. 1550 G with g = 4.23 originating from Fe3+ impurities present in Al(NO,), . This signal can be used as an internal standard for estimating the relative inten- sities of the other EPR signals. NO2 A characteristic feature present in the EPR spectra of all samples is a number of lines in the g x2.0 region extending over ca. 150 G. The highest intensity and best resolution was attained at low measurement temperatures. Fig. 3 and 4 illus-trate the EPR spectrum of this signal as a function of tem-perature on a more expanded scale.For samples with y -c 1.5 the lines became practically undetectable at measurement T, > 150 K (Fig. 3), but for samples with y 23 they were clearly observed even at room temperature (Fig. 4). These spectra can be ascribed to NO, and are well described in the In Fig. 3 the nitrogen hyperfine EPR lines are labelled with x, y or z, where y runs parallel to the inter- atomic oxygen-oxygen direction. The spectra show a clear temperature dependence. At low temperatures a powder-like spectrum is observed. Going to higher temperatures the x and z components merge, while the y components keep their positions. From this behaviour it can be concluded that the NO, molecules above 20 K rotate rapidly around an axis J. CHEM. SOC. FARADAY TRANS., 1994, VOL.90 62K 58 K 1 A* 12 3 4 5 6 7 8 9101112 ~/103G Fig. 2 EPR spectra of AB,.,-T, at: 12 K, 55 K, 58 K,62 K parallel to the y axis (2lo7 Hz), as has been observed pre- viously.'s~'8~'9For the sample AB,.,-T, this leads eventually to an isotropic spectrum (T 2 125 K), whereas for the sample AB,.,-T, even at 300 K the rotation is still anisotropic. Thus the mobility of NO, molecules depends on the thermal history of the samples. In the Discussion we will further elaborate on this. In Table 1 the magnetic parameters are listed together with those obtained for NO, adsorbed on similar systems and for NO, in the gas phase. From the simi- larity of the values in Table 1 it can be concluded that the adsorbed NO, is not greatly distorted by the aluminium borate matrix.-125K 3250 3300 3350 3400 HIG Fig.3 Temperature dependence of the NO, EPR spectrum f AB,,,-TI. The assignment of hyperfine lines is shown. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 n 3250 3300 3350 3400 HIG Fig. 4 Temperature dependence of the NO, EPR spectrum of ABo.,-T, In a relatively small temperature interval (50-65 K) a beauti- ful multiline spectrum was observed between 5 and 12.5 kG (Fig. 2 and 7). Comparing this multiline signal with that observed for 0,in the gaseous state20v21 it can be inferred that this signal originates from 0, molecules in gaseous form probably present inside the pores of the sample. Above ca. 65 K this spectrum broadens beyond detection (see Fig. 2).At low temperatures it disappears and is replaced by a new strong signal at cu. 11.7 kG (Fig. 2 and 7). Apparently decreasing the temperature causes the O2 molecules to con- dense on the surfaces of the pores and then gives rise to the well known signal at ca. 11.7 kG,22-25characteristic of 0,in the condensed phase. That the signal is due to 0, molecules produced by the decomposition reactions (1) and (2) is proven by the experiment illustrated in Fig. 5. In Fig. 5(a)the EPR spectrum is shown for the as-prepared sample AB,.,- T1.5.A very strong signal is observed at 11.7 kG. The quartz tube was then opened and connected to a vacuum system. After evacuation of the sample at room temperature the sample tube was sealed and subsequently the EPR spectrum was measured.A dramatic decrease of the signal intensity at 11.7 kGwas observed [see Fig. 5(b)].A further short heat r 1 4 I I 200 G II 123456789101112 ~~103G Fig. 5 EPR spectra of AB,,,-T,,5 at 14 K: (a) before evacuation, (b) after evacuation at room temperature and (c) after a new short (<1 min) heat treatment at 150"C treatment of the sample in the closed tube (<1 min) at the same temperature (150 "C) resulted in an enhanced signal [Fig. 5(c)]. This enhancement must be due to 0, molecules produced by the decomposition reactions that occur during the short heat treatment. The EPR spectrum of condensed triplet 0, has been analysed by using the following spin Hamiltonian H,= D[S,2 -fs(s+ l)] + E(S; -S;) + BB, -g s and with the aid of the EPR simulation program MAGRES.28 An excellent fit was obtained using the follow- ing set of parameters: gx = g,, = 2.02, gz = 0.673, D = 109.3 GHz, E = 0.075 GHz, Lorentzian linewidth = 200 G.The dotted line in Fig. 7 shows the simulation. In passing we note that a perfect fit for the signal at 11.7 kG could be obtained only by taking a non-zero asymmetry parameter E Table 1 EPR parameters of NO, molecules adsorbed on surface of various oxide matrices oxide matrix 9, g: 9, Ad MHz A:l MHz AZl MHz Ad MHz ref. - MgO (93 K) ZnO (77 K) 2.005 2.007 1.9915 1.994 2.002 2.003 1.9995 2.001 148 146.1 137 132 189 181.1 158 153.1 16 19 silica gel (77 K) 2.004 1.9907 2.004 Vycor (4.8 K) 2.005 1.9913 2.0017 zeolites (77 K) 2.0043 1.9922 2.0017 aluminium borates 2.00 15 1.9985 2.009 (20 K)NO, (gas) (293 K) 2.0062 1.9910 2.00 19 x = y is the axis parallel with the interatomic oxygen-oxygen direction.1.9996 165.7 137 165.7 156.1 18 1.9994 140.3 128.2 183.5 150.7 19 1.9993 143.3 137.2 189.1 156.5 17 2.0066 144.5 133.0 185 154.1 -a 1.9997 128.0 126.0 184.9 146.5 13 Present work. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 300 K 4 5 6 7 8 9101112 1 ~~~ ~ 4 5 6 7 8 9 1011 12 H/103 G Fig. 6 EPR spectra for ABO.3-T,,, (a)and AB,.,-T,., (b)at 14 K and the feature at ca. 5 kG could be reproduced only using an axial g tensor. In Table 2 the values of the parameters are listed together with those obtained in other matrices.The D values for trapped 0, molecules are significantly smaller than that for 'free' 0, , 119 GHz.,' NO The presence of NO in the samples is also revealed in the EPR spectra. The EPR spectra are shown for ABo.,-T'.5 [Fig. qa)] and AB0.5-TI.5 [Fig. 6(b)] measured at 14 K. The latter spectrum reveals an asymmetrical signal with a large tail at high field, characteristic of NO, at g x 2.0 without con- Table 2 EPR parameters of 0, molecules trapped in different matrices ~~ ~ AH1 Dl El matrix 9x g,, g: G GHz GHz ref. N, solid 2.02 2.02 0.7 40 108 -23 (<20 K) 2.0 2.0 2.0 25 108.3 -24 2.0 2.0 2.0 -107.1 -25 CO solid 2.02 2.02 0.7 80 108 -23 (<20 K) 2.0 2.0 2.0 400 108.5 -24 KBrO, 1.999 2.015 1.996 -108.9 1.3 26 crystals (26 K)NaClO, 2.004 2.006 2.003 -112.2 0.135 27 crystals (4.2 K)aluminium 2.02 borates ((20 K) * Present work.x = 2.02 0.673 320 109.3 0.075 -= z is the molecular axis. 111l1011111 123456789101112 H/103 G Fig. 7 Temperature dependence of the EPR spectra of AB,.,-T,. The dotted line is the simulated signal for trapped 0, molecules using parameters listed in Table 2. tamination from other signals. In Fig. 6(a) the NO signal is superimposed on the NOz signal, as is the case in Fig. 5(c) (see insert). Close inspection of the spectra of other samples always showed a contribution from NO. In Table 3 the mag- netic parameters are tabulated together with those of physi- sorbed NO molecules on y-al~mina,,~ ~ilica-magnesia,~~ Mg0I6 and zeolite^.'^'^^ Our values, estimated directly from the EPR spectra [Fig.5(c) and 61, are in accordance with those measured in similar systems. Other Paramagnetic Species In the EPR spectra of samples heat treated between 150 and 200°C we observed EPR signals that we could not identify. As an illustration of this we refer to Fig. 7, where spectra are shown for the sample AB,.,-T,. Broad lines are observed below ca. 5 kG. As can be seen from the figure, some lines shift to lower magnetic field values with decreasing tem- perature. It is suggested that these signals arise from pairs or clusters formed from paramagnetic species with S = 1/2. In Fig. 5(a) there are some features at 5 and 8 kG which might also be due to paramagnetic dimers or clusters.In some cases (at high NO concentrations) a well defined line at g = 4 was observed [see peak in Fig. qb), indicated by an arrow], which might be a half-field signal from paramagnetic NO dimers. Peaks due to dimers of NO, could be discerned in the full- field region of the NO, spectrum, especially at high NOz concentration, as was observed by Schaafsma and Komman- deur." Finally, in the spectra of all samples in the region around g x 2.0, especially at low temperature, weak signals were J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 3 EPR parameters of NO molecules adsorbed on surface of various matrices (materials) 41AXl AYI oxide matrix 9, s: MHz MHz MHz ref.y-alumina (77 K) 1.996 1.996 1.96 -95 -29 silica-magnesia 1.996 1.996 1.91-1.95 -85 -29 (77 K)MgO (77 K) 1.995 1.995 1.91 -93 -16 zeolites 1.986-2.00 1.978-1.998 1.83-1.93 -85-95 -30, 31 (77 K) --aaluminium borates 1.985 1.985 1.81-1.91 -85 (14 K) 'Present work. x = z is the molecular axis. observed superimposed on the strong signals of NOz and NO. They are possibly related to some crystal defects or to organic and/or inorganic radicals produced during the syn- thesis. NMR on Aluminium Borates It was expected that changes in coordination of aluminium would be manifest clearly in 27Al MASNMR spectra. This was indeed the case. The effect of heat-treatment temperature on the shape of the 27Al MASNMR spectra is shown in Fig.8 for samples without boron (spinning side bands are marked by asterisks). The dependence of the spectra on sample com- position is illustrated in Fig. 9 for samples with different n T8.6 T6 T4 T2 Tl .5 250 200 150 100 50 0 -50 -100 -150 6 Fig. 8 27Al MASNMR spectra of AB, samples at room tem-perature showing the effect of heat-treatment temperature. The spin- ning side bands are marked with asterisks and are outside the range of chemical shifts of the three A1 resonances. boron contents and the same heating temperature (T = 400°C). In all spectra three resonance peaks can be discerned with different intensities. The resonances at 0-8 and 58-64 ppm are unanimously accepted to originate from octa- hedrally and tetrahedrally coordinated Al, re~pectively.~'-~~ The third resonance at ca. 30 ppm is consistent with the chemical shift of penta-coordinated Al, as previously observed in NMR studies of crystalline materials with well defined penta-coordinated A133,3 or in studies of disordered matrices, as gels and glasses, containing A1 Fig.8 shows that mainly between 200 and 300°C penta-coordinated and tetra-coordinated A1 are formed at the expense of octahedrally coordinated Al. Above 800 "C the spectrum corresponds to the NMR spectrum of y-alumina, consisting of only tetrahedrally and octahedrally coordinated Al. Fig. 9 shows that the amount of penta-coordinated A1 increases, whereas the fraction of tetra-coordinated A1 decreases with increasing boron concentration.0.1 0 I I I ! I 1 I I I I 250 200 150 100 50 0 -50 -100 -150 -200 6 Fig. 9 27Al MASNMR spectra of AB,-T, samples showing the effect of the boron conten;; the value of^x k shown on the spectra. The spinning side bands are marked with asterisks and are outside the range of chemical shifts of the three A1 resonances. I\\ * * I I I I I 250 200 150 100 50 0 -50 -100 -150 6 Fig. 10 *'A1 MASNMR spectra for the AB,-T, sample: (a) as pre- pared, (b)after 1 month and (c) after 2 months storage in air, and (d) after a new heat treatment at the same temperature (400°C). The spinning side bands are marked with asterisks and are outside the range of chemical shifts of the three A1 resonances. For samples with y d 7.5 the 27Al NMR spectra changed in time when these samples were exposed to air.This effect, also dependent on the boron content, is illustrated in Fig. 10 for the sample AB,-T,. When, after the first measurement [Fig. lqa)], the sample is exposed to air, changes are observed caused by absorption of H20 molecules from the air [Fig. lqb), (c)]. The peak intensities of tetra- and penta- coordinated A1 decrease, whereas the intensity of the peak due to octahedrally coordinated A1 increases. After renewed heat treatment at the same temperature (400°C) a similar 27Al NMR spectrum was obtained to that measured for the as-prepared sample [Fig. lqa)]. The same effect was observed recently for other aluminium oxide matrices.32 Discussion NMR During sol-gel synthesis the materials pass through several stages.'* In the starting solutions at room temperature the principal process is the hydrolysis reaction, during which the majority of the present cations become coordinated to hydroxy groups and water molecules.By increasing the tem- perature, condensation reactions evolve with the formation of M-0-M bonds and the production of water. These con- densation reactions proceed during the drying of gels by suc- cessive heat treatments. At the end of the synthetic process amorphous porous xerogels are obtained. Our experiments shed more light on what happens specifi- cally during the synthesis of the xerogels. Up to a heat-treatment temperature of 200 "C decomposi-tion reactions occur and glycerol is partially oxidized.The 27Al MASNMR spectra recorded for samples taken at heat- J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 treatment temperatures of 150 and 200°C hardly differ (see Fig. 8). A dramatic change in the local structure of the xerogels takes place between 200 and 300"C,as is nicely illustrated by the 27Al MASNMR spectra in Fig. 8. From these spectra it may be concluded that in the heat-treatment temperature range 300-600°C four- and five-coordinate A1 are formed at the expense of six-coordinated Al. This is more clearly demonstrated in Fig. 11 where the change in the relative peak areas is plotted against the heat-treatment temperature. The relative intensities of the NMR peaks were determined by deconvolution of the spectra, assuming that the line shapes can be fitted with Gaussians.Since the electric field gradient is different at each of the three sites, the peaks have different and asymmetrical line shapes. These effects were taken into account by allowing the lines for each site to be a linear com- bination of one, two or three Gaussians with different widths and positions. In this way good fits were obtained. Fig. 9 shows that four- and five-coordinate A1 are also present in the boron-containing samples. The possibility that the resonance peak at ca. 30 ppm is due to aluminium coordi- nated to boron atoms in the second coordination sphere6 can be ruled out because this resonance was also observed in pure alumina samples. Moreover, it was observed in 29Si NMR of boron-containing MFI zeolites that boron present in the second coordination sphere of Si had no effect on the Si chemical shift.39 Therefore, we do not expect any effect on the A1 chemical shifts either.From Fig. 9 it can furthermore be inferred that the boron ions prefer four- instead of six-coordination (see spectrum of sample AB,,,). The decrease in the number of aluminium atoms in the four- and five-coordinate ion sites in samples exposed to air (so-called air-equilibrated gels), suggests that low-coordinate A1 atoms are predominantly located at the surface of the xerogels. It is known that these four- and five-coordinate A1 atoms at the surface are associated with the catalytically active sites.'*40 The partial rehydroxylation that takes place when the sample is exposed to air is the principal factor involved in the different results reported in the literature for similar material^.^.^ The difference in preparation procedure seems to be of minor importance.Not only is the local structure changed during the synthe- sis of xerogels but also the internal porosity. By increasing 0.8 \ 0.6 c4-0 Y s z* 0.4 0.2 :\ 0.0 100 300 500 700 900 TJC Fig. 11 Fraction of (V) hexa-, (0)penta- and (0)tetra-coordinated A1 ions as a function of heat treatment temperature (TJ.The dotted lines are guides to the eye. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 the heat-treatment temperature the dimensions of the pores diminish, the surface area decreases and the density of the samples increases (skeletal densification).After 3 h heat treat- ment at 860°Cthe NMR spectra of the samples (Fig. 8) are identical to that of y-alumina, proving that the amor- phous alumina xerogels had been transformed into poly- crystalline alumina. EPR The EPR results offer us new, important information about the evolution of catalytically active sites. The gaseous para- magnetic species that are produced by the decomposition reactions interact with these active sites before they change their initial configuration by structural relaxation or as a result of interactions with other non-paramagnetic molecules such as water. As indicated above, the mobility of physisorbed NO, mol-ecules depends on the heat-treatment temperature. Their mobility, determined by the strength of their interactions with the surfaces of the xerogels, is revealed by the tem- perature dependence of the EPR spectra shown in Fig.3 and Fig. 4. NO, molecules in samples treated at low temperatures have a higher mobility than in samples treated at higher tem- peratures. For instance in the sample AB,.,-T, the NO, mol- ecules at 125 K rotate isotropically, whereas in the sample AB,.,-T, at the same temperature the NO, molecules rotate anisotropically around an axis parallel to the interatomic axis between the oxygen atoms (denoted as the y axis). For samples with y Q 1.5 the mobility of the NO, molecules is comparable to the mobility observed for NO, adsorbed on Vycor glassIg or on zeolite^,^' but for samples with y > 2 the mobility of NO, is lower.At high treatment temperatures the samples become to a great extent dehydroxylated and accordingly the interaction between the NO, molecules and the surface becomes stronger, thus restricting their mobility. For the sample AB,.,-T, the NO, molecules move aniso- tropically even at room temperature (see Fig. 4). As shown in Table 1, the magnetic parameters of NO, in aluminium borate xerogels are comparable to the values observed for similar systems. Hence the adsorbed NO, mol-ecules are not greatly distorted by the aluminium borate matrix. Pietrzak and concluded from this that the majority of the NO, molecules dimerize at low temperature and are surrounded by several N204 molecules which have condensed on the walls of the zeolites studied by them, thus shielding the NO, molecule from strong cation interactions.The same conclusion may be drawn for the system investi- gated by us. Fig. 6 shows that on increasing the boron content the 0, EPR signal decreases. This can be rational- ized as follows. When the boron content increases less 0, will be produced by the decomposition of Al(NO,), because its concentration decreases as the boron content increases. The 0, molecules produced will be partly used in the oxidation of glycerol, the concentration of which is the same in all samples. Thus the 0, signal intensity observed will decrease as the boron content increases. This is not contradicted by the experiment described in Fig.5, where we see that the 0, EPR signal increases after a renewed heating procedure. We must remember that in this experiment the sample has been placed in a closed evacuated tube, so that the decomposition products can escape into the empty space above the sample, especially the non-polar 0, molecules. The polar molecules such as NO are preferentially fixed on coordinatively unsaturated active sites (A13+,B3+), developed during the heating process. Thus the 0,molecules that have escaped cannot be consumed in the pyrolysis of glycerol. After cooling the sample to 14 K the 0, molecules will condense on the surfaces, and will give rise to an enhanced EPR signal. At heating temperatures below 200°C we observed in the EPR spectra signals that we believe to arise from species with S 2 1.From the 27Al NMR spectra it can be concluded that below 200°C only a small number of low-coordinate A1 species are present (see Fig. 8). At the beginning they are not uniformly distributed over the surface area of the pores, but will appear in zones in which the hydroxylation process was most effective. From structural studies it was found that one oxygen anion vacancy can create as many as three five- coordinate A1 sites.37 The polar paramagnetic species, espe- cially the very reactive NO molecules, will react with these active sites and form paramagnetic clusters with S 2 1. The formation of these clusters is also favoured by the high gas pressure in the pores, which are closed at this temperature.The sharp signal at g = 4, indicated by an arrow in Fig. 6, may arise from NO dimers. Pairs of NO, molecules can also be formed, especially when the concentration of NO, is high.15 We found evidence for this in our spectra, measured below 20 K. Note that we found no evidence in our experiments of the presence of 0,-.When 0, is adsorbed on 'clean' surfaces of activated oxide materials, usually an electron is transferred to 02.41In our experiments the surfaces will be preferentially covered by polar molecules such as NO, N,O,, NO,, OH-or H,O, so that the 0, molecules are physisorbed on top of them and no electron transfer to 0, takes place. The appearance of the multiline spectrum of 0, was a surprising result.The multiline spectrum arises through the coupling of the rotational angular momentum, which is quenched in the liquid or solid phase, with the electronic spin and orbital angular momentum.21 This spectrum could be observed only over a narrow temperature interval of 15"C.If the 0, concentration is high, collisional broadening will occur. On cooling the sample, the concentration of 0, decreases by condensation of 0, molecules on the surfaces of the pores. A point is then reached where the concentration of 0, is large enough and the relaxation times long enough to make detection of the EPR spectrum of gaseous 0, possible. Further lowering of the temperature leads to total conden- sation and to the disappearance of the gaseous EPR spec-trum and to the appearance of the EPR spectrum of 0, in the condensed phase.The zero-field splitting (D) of 0, in the gaseous state is 119 GHz.,' For 0, in our system the D value amounts to 109 GHz. The reduction of the value of D has been attributed to torsional oscillation of the 0, molecules in a potential well provided by the matrix.42 Table 2 shows that the same effect has been observed for 0, trapped in other matrices. Conclusion Our EPR studies on aluminium borate materials show that the paramagnetic products of decomposition reactions in sol-gel processing of amorphous xerogels can be used as EPR probes for the study of the active sites developed during the synthesis on the surface of the pores.The strength of the interactions between the active sites and the paramagnetic products are reflected in the shape of the EPR spectra of the adsorbed molecules. The 27Al MASNMR spectra reflect directly the changes in A1 coordination during the heating procedure. In the tem- perature range from 200 to 300°C four- and five-coordinate A1 species are formed, which, on exposure to air, are partially transformed again to six-coordinate A1 by absorption of water present in the air. This observation stresses the impor- tance of working under well defined conditions in cases where 2670 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 the materials studied have high surface area. Some conflicting results in the literature may be ascribed to air-equilibration of the samples.Finally, the simultaneous use of NMR and EPR techniques is shown to be a powerful tool for the study of the active sites 15 16 17 18 19 T. J. Schaafsma and J. Kommandeur, Mol. Phys., 1968,14,517. J. H. Lunsford, J. Colloid Interface Sci., 1968,26, 355. T. M. Pietrzak and D. E. Wood, J. Chem. Phys., 1970,53,2454. M. Iwaizumi, S. Kubota and T. Isobe, Bull. Chem. SOC. Jpn., 1971,44,3227. M. Shiotani and J. H. Freed, J. Phys. Chem., 1981,85,3873. in amorphous and crystalline materials. 20 21 R. Beringer and J. G. Castle, Phys. Rev., 1951,81, 82. M. Tinkham and M. W. P. Standberg, Phys. Rev., 1955,97,951. The authors thank Mr. A. A. K. Klaassen, Mr. G. E. Janssen and Mrs. G. H. Nachtegaal for their skilful experimental assistance.22 23 G. M. Graham, J. S. M. Harvey and H. Kiefte, J. Chem. Phys., 1969,52,2235. R. Simoneau, J. S. M. Harvey and G. M. Graham, J. Chem. Phys., 1971,54,4819. 24 H. Kon, J. Am. Chem. SOC., 1973,%, 1045. References 25 26 S. Hirokawa, J. Phys. SOC. Jpn., 1974,37, 897. J. R. Byberg, Chem. Phys. Lett., 1978,57, 579. 1 2 3 4 5 6 H. Knozinger, in Catalysis by Acids and Buses, ed. B. Imelik, C. Naccache, Y. Condurier, B. Tarit and J. C. Vedrine, Elsevier, Amsterdam, 1985, p. 11 1. G. Pajonk, M. Repellin and S. J. Techner, Bull. SOC. Chim. Fr., 1976, 1333. G. Tournier, M. Lacroix-Repellin, G. M. Pajonk and S. T. Teichner, Preparation of Catalysts IV, ed. B. Delmon, P. Grange, P. A. Jacobs and G. Poncelet, Elsevier, Amsterdam, 1987, p. 333.S. F. Mitchel, G. Marcelin and J. G. Goodwin, J. Catul., 1987, 105,521. D. L. Cocke, E. D. Johnson and R. P. Merrill, Catal. Rev. Sci. Eng., 1984,26, 163. K. P. Peil, L. G. Galya and G. Marcelin, J. Catal., 1989, 115, 441. 27 28 29 30 31 32 33 34 35 36 N. Bjerre, J. Chem. Phys., 1982,76, 3347. C. P. Keijzers, E. J. Reijerse, P. Stam, M. F. Dumont and M. C. M. Gribau, J. Chem. SOC., Faraday Trans. 1,1987,83,3469. J. H. Lunsford, J. Catal., 1969, 14, 379. J. H. Lunsford, J. Phys. Chem., 1970,71, 1519. P. H. Kasai and R. J. Bishop Jr., J. Am. Chem. SOC., 1972, 94, 5560. M. E. Smith, Appl. Magn. Reson., 1993,4, l., L. B. Alemany and G. W. Kirker, J. Am. Chem. SOC., 1986, 108, 6158. D. Massiot, A. Kahn-Harari, D. Michel, D. Muller and F. Tau-lelle, Magn. Reson. Chem., 1990,28, S82.M. E. Smith and S. Steuernagel, Solid State NMR, 1992,l 175. A. D. Irwin, J. S. Holmgren and J. Jonas, J. Muter. Sci., 1988,23, 2908. 7 8 9 10 11 F. Abbattista, A. Delmastro, G. Gozzelino, D. Mazza, M. Vallino, G. Busca and V. Lorenzelli, J. Chem. SOC., Faraday Trans., 1990,86,3653. A. Delmastro, G. Gozzelino, D. Mazza, M. Vallino, G. Busca and V. Lorenzelli, J. Chem. SOC., Faraday Trans., 1992,88,2065. S. Tanabe, K. Hirao, N. Soga and T. Hanada, J. Solid State Chem., 1992,97,481. R. Bechara, A. Aboukais and J. P. Bonnelle, J. Chem. SOC., Faraday Trans., 1993,89, 1257. J. G. Darab and R. K. MacCrone, Phys. Chem. Glasses, 1991,32, 191. 37 38 39 40 41 T. E. Wood, A. R. Siedle, J. R. Hill, R. P. Skarjune, C. J. Good- brake, Muter. Res. SOC. Symp. Proc., 1990,180,97. C. J. Brinker, in Glass: Science and Technology, ed. D. R. Uhlmann and N. J. Kreidl, Academic Press, New York, 1990, vol. 4A, p. 169. R. de Ruiter, A. P. M. Kentgens, J. Grootendorst, J. C. Jansen and H. van Bekkum, Zeolites, 1993,13, 128. H. Yong, B. Coster, F. R. Chen, J. G. Davis and J. J. Fripiat, New Frontiers in Catalysis, ed. L. Guczi, F. Solymosi and P. Tktenyi, Elsevier, Amsterdam, 1993, p. 1159. M. Che and A. J. Tench, Adv. Catal., 1983,32, 1. 12 S. Ikoma, K. Kawakita and H. Yokoi, J. Non. Cryst. Solids, 1990,122,183. 42 H. Meyer, M. C. M. OBrien and J. H. van Vleck, Proc. R. SOC. London, Ser. A, 1957,243,414. 13 G. R. Bird, J. C. Bird, A, W. Jache, J. A. Hodgson, R. F. Curl, A. C. Kunkle, J. W. Bransford, J. Rastrup-Anderson and J. Rosenthal, J. Chem. Phys., 1964,40,3378. 14 T. J. Schaafsma, G. A. van der Velde and J. Kommandeur, Mol. Phys., 1968, 14, 501. Paper 4/02417F; Received 25th April, 1994