J . Chern. SOC., Faraday Trans. I, 1985, 81, 2043-2051 Decomposition of N20 on Fe203/A1,03 Catalysts Relationships between Physicochemical and Catalytic Properties BY PHILIPPOS POMONIS Chemistry Department, University of Ioannina, Ioannina, Greece AND DIMITRIS VATTIS AND ALEXIS LYCOURGHIOTIS* Chemistry Department, University of Patras, Patras, Greece AND CHRISTOS KORDULIS Universite Catholique de Louvain, Groupe de Physicochimie Minerale et de Catalyse, 1348 Louvain La Neuve, Belgium Received 20th September, 1984 The influence of the iron(II1) content and the calcination temperature on the dispersion of supported Fe3+ species on alumina, on the semiconducting properties and on the catalytic activity of Fe,O,/Al,O, catalysts has been studied, with the decomposition of N,O into N, and 0, being used as a probe reaction.Above 250 "C iron(1II) is located on the surface of the carrier as a-Fe,O, and 'strongly associated iron(II1)' (denoted by Fe3+-S). The ratio of the amounts of these species remains constant from X = 0.172 to X = 0.615 mmol Fe3+ per g of alumina and then changes favouring Fe3+-S. The transformation of a-Fe,O, into Fe3+-S is not accompanied by any detectable change in the Fe3+ dispersion but causes an increase in the activation energy of conduction. A rise in the calcination temperature and iron(II1) content brings about an increase in the size of supported a-Fe,O, crystals shown by a decrease in the dispersion of Fe;+. it has been demonstrated that a-Fe,O, and Fe3+-S exhibit similar activity. This increases iinearly with the dispersion of the Fe3+ ions.Moreover, no relationship was found to exist between the semiconducting properties of the catalysts estimated by the activation energy of conduction and the catalytic activity. X-ray photoelectron spectroscopy was used to estimate the dispersion of the supported iron(III), conductivity experiments were performed to determine the activation energy of conduction and catalytic tests were carried out to determine catalytic activity. The catalytic properties of supported metal oxide catalysts are generally determined from the following physicochemical characteristics : (a) the kind of the species which the active ions form on the surface of the carrier, (b) the semiconducting properties of the solids estimated from the activation energy of conduction and (c) the dispersion of the active phase on the carrier surface.On which of rhe above the catalytic activity depends is one of the most important problems in catdysis. A research program was recently initiated in our laboratory to solve this problem in the case of the iron(Ir1)-supported alumina catalysts. Although this system had been extensively studied by various methods, especially Mossbauer spectroscopy, a close relationship between the catalytic properties and the above physicochemical characteristics has not been established. Thus we have studied'. by various techniques the influence of the calcination temperature and of the iron(m) content on the textural and structural features of both the active phase and the support. For the carrier it was found that a progressive increase in the calcination temperature brings about the removal of the physically adsorbed H,O followed by 20432044 DECOMPOSITION OF N20 ON Fe,O,/Al,O, an irreversible transformation of the a-Al,[OOH], to alumina.2 For the supported iron(1n) phase it was found that a rise in the calcination temperature from 90 to 250 "C causes a decomposition of the Fe(NO,), 9H,O followed by progressive dehydration of the supported a-Fe,O, - xH20 resulting from that decomposition.An additional rise in the calcination temperature from 250 to 320 "C causes a decrease in the amount of the supported a-Fe,O,; this takes place irrespective of the iron(m) content. Further increases in the calcination temperature do not change the concentration of the supported a-Fe,O,. Two alternative explanations have been proposed for this effect :2 (a) dilution of the Fe3+ into the bulk of the support or (b) formation of a strongly associated, and thus difficult to reduce, iron(Ir1) species on the surface of the carrier.The size of a-Fe,O, crystals formed at temperatures > 320 "C varies with the iron(II1) content. The samples calcined at the maximum calcination temperature, i.e. 665 "C, give the following results: (a) a-Fe,O, crystals were not detected for the specimens with iron content < 0.388 mmol Fe3+ per g of the support, (b) the formation of very small crystalls of a-Fe20, not detectable by X-ray spectroscopy was inferred from diffuse reflectance spectra and magnetic measurements on the samples with iron content in the range 0.388-0.974 mmol Fe3+ per g of support and (c) a-Fe,O, crystals detectable by X-ray spectroscopy were identified in the sample containing the maximum iron(rI1) content, i.e.0.974 mmol Fe3+ per g of support. In the present study we have attempted (a) to study the influence of the iron(m) content and calcination temperature on the state of dispersion of Fe3+ ions on the carrier surface as well as the semiconducting properties of the solid catalysts, (b) to elucidate the solid-state process responsible for the decrease in the amount of the supported a-Fe,O, when calcination occurs at 250-320 "C and (c) to relate the physicochemical characteristics mentioned above of samples calcined at temperatures > 320 "C with their catalytic properties.The series containing samples calcined at 665 "C was selected as typical and the decomposition of N,O was used as a model X-ray photoelectron spectroscopy (X.P.S.), conductivity experiments and catalytic tests under atmospheric pressure were the methods of investigation used. EXPERIMENTAL PREPARATION OF THE SPECIMENS Details of the method of preparation of the various specimens have been reported elsewhere.2 The specimens studied here are referred to as Fe-X-Y, where X denotes the iron(Ir1) content as mmol Fe3+ per g of alumina and Y denotes the calcination temperature ("C). CATALYTIC TESTS The catalytic tests were carried out in a flow system under atmospheric pressure. A catalyst sample of surface area 42.65 m2 was placed on a perforated-glass bed of volume of ca.0.35 cm3 and depth ca. 0.2 cm. An He+N20 (2: 1) gas mixture was passed through the bed at a rate of 180 & 5 cm3 min-'. The contact time was thus 0.12 s. A Varian 3700 gas chromatograph equipped with a thermal-conductivity detector was used to analyse both reactants and products. Two gas valves with 1 cm3 loops were used for sampling. The column (0.5 m long by 8 in? internal diameter) was of stainless steel and filled with 5A molecular sieve. Preliminary tests showed that no decomposition took place up to 700 "C in the absence of catalyst. The temperature range examined was 500-650 "C. In all the experiments the sequence of temperatures examined was 600 -+ 550 -+ 500 -+ 525 -+ 575 + 625 -+ 650 "C. t 1 in = 2 . 5 4 ~ 10+ m.P. POMONIS, D. VATTIS, A.LYCOURGHIOTIS AND CH. KORDULIS 2045 X-RAY PHOTOELECTRON SPECTROSCOPY X-ray photoelectron spectra were obtained using a Vacuum Generators ESCA 3 spectrometer equipped with an aluminium anode (A1 Ka = 1486.6 eV) which operates at 20 mA and 1.4 kV. The residual pressure inside the spectrometer was ca. Torr.? A signal averager (Tracor Northern 1710) was used to improve the signal-to-noise ratio. Binding energies were referenced to the C 1s line at 285 eV. The photoelectrons counted per unit time, i.e. the X.P.S. signal intensities, are represented by the areas under the corresponding peaks. The X.P.S. relative intensity measurements concern the ratios of the intensities of two peaks, I , and I,, associated with the supported Fe3+ and the carrier, respectively. Several models have been proposed to relate the X.P.S.intensity ratios to the dispersion of the deposited phase.12-ls The following equations have been derived,” based on these models, provided that the active phase is present as cubic crystallites on the surface of the carriers: (1 a-c) where Im/Is = IFezp/IA12p, n,/nm is the atomic ratio Al/Fe in the specimen, D,, is the dispersion of the iron(m) and C, C, and C, are proportionality constants. Eqn (1 a-c) were derived assuming that the length of the edge of the cubic crystallites of the deposited phase is much greater than, almost equal to and much smaller than the inelastic mean free path of the photoelectrons, respectively. From eqn (1) one can see that the quantity Im/Is [or (Im/&) (ns/n,) when n,/nm is not constant] is an increasing function of DF,, and therefore it can be used for an estimation of the dispersion of iron(m) on the surface of the carrier. Note that quantitative analysis based on X.P.S.intensity measurements relative to a series of specimens on a porous support is valid only when no variation in the repartition of the deposited phase takes place. By reparation we mean the relative amount of species deposited in the inner parts of the elementary catalyst particle and those located at the mouth of the pores or on the external surface of the particle.16 CONDUCTIVITY MEASUREMENTS The conductivity experiments were carried out in a system similar to that described in ref. (1 1). A pellet of the catalyst was prepared by compressing 400 mg of the sample under a pressure of 10 toncm-2$ for good electrical contact between the catalyst particles.The pellet was clamped between two platinum electrodes, which were connected to a 1.5 V cell. The current passing through the pellet was measured with a Hewlett-Packard volt-ammeter having a range from 10-l2 to A. The electrical measurements were performed under atmospheric pressure since the catalysts were examined for the N,O decomposition under similar conditions. The current i passing through the sample at different temperatures is given by i = VSa/l, where V is the voltage applied, S is the area of the electrode (ca. 0.5 cm2), 1 is the thickness of the pellet (ca. 0.3 cm) and CJ is the specific conductance related to the temperature by the relation applied to semiconductors, CJ = oo exp ( - E,/RT).Plots of log i against 1 /T were used to calculate the activation energy of conduction Eg. RESULTS CATALYTIC TESTS The degree of conversion at each temperature is given in table 1. These data have been normalised to a catalyst containing 1 mmol of supported Fe3+. The variation of activity with the amount of the supported iron(n1) is shown in fig. 1 for three different temperatures. Inspection of table 1 and fig. 1 shows a continuous drop in the activity with increasing iron (111) content. t 1 Torr z 133.3 Pa. 1 I ton cm-2 = 98.07 x lo6 Pa.2046 DECOMPOSITION OF N20 ON Fe,O,/Al,O, 8 - A Y ." .- t; 6 - 0 x - m ." c.l - *-' 4 - 2 - Table 1. Degree of conversiona for the decomposition of N,O temperature of reaction/OC FeX- Y 500 525 550 575 600 625 650 Fe-0.00&665 - (0.050) (0.080) (0.120) (0.220) (0.360) (0.530) Fe-0.172-665 0.878 1.316 2.487 3.949 6.377 8.864 11.116 Fe-0.245-665 0.756 1.418 2.317 3.735 4.964 6.902 8.037 Fe-0.388-665 0.419 0.805 1.449 2.319 2.834 4.348 5.314 Fe-0.615-665 0.408 0.730 1.332 1.933 2.578 3.437 - Fe-0.974-665 0.266 0.503 0.961 1.448 1.833 2.498 - a Nomalised to a catalyst containing 1 mmol of supported Fe3+, except values in parentheses.O L 1 I 1 I 1 I L 1 0.1 0.3 0.5 0.7 0.9 iron(rr1) content, X Fig. 1. Variation of catalytic activity of Fe-X-665 with iron(m) content for the decomposition of N,O at three temperatures: A, 525; e, 575 and 0, 625 "C. X-RAY PHOTOELECTRON SPECTROSCOPY The values of the binding energies, determined for the Fe 2p1,2, Fe 2p,,, and Fe 1s electrons, remain practically constant irrespective of the iron(m) content or calcination temperature.Fig. 2(a) and 3(a) illustrate the variation in the dispersion of the supported iron(rI1) with its concentration in the sample and calcination temperature, respectively. Note that the dispersion decreases as the iron(Ir1) content and calcination temperature increase. The decrease of (IFe 2p/IA1 2 p ) (nAl/nFe) with X is almost exponential, while the rate of the decrease of the dispersion with calcination temperature remains practically constant in the range 200-500 "C and then increases.P. POMONIS, D. VATTIS, A. LYCOURGHIOTIS AND CH. KORDULIS 2047 I 1 1 I I I , 3.0 n E + ._ ;iij 2.0 g W % G 9. a i - 4 1.0 0 .o 0.2 0.4 0.6 0.8 1 .o iron(II1) content, X Fig.2. Variation of (a) the dispersion of Fe3+, (b) the amount of the supported a-Fe,O, and ( c ) the activation energy of conduction with the iron(r1r) content for Fe-X-665. 4 3 n P N +: --. 3 0 2 1 Fig. 3. Variation of (a) the dispersion of Fe3+ and (b) the amount of the supported a-Fe,O, with the calcination temperature for Fe-0.974- Y.2048 DECOMPOSITION OF N,O ON Fe,O,/Al,O, 100 200 300 A00 'Fe 2 p nAl IA12p Fe Fig. 4. Variation of the catalytic activity of Fe-X-665 with the dispersion of Fe3+ for the decomposition of N,O at three temperatures: A, 525; 0, 575 and 0, 625 "C. The variations in the temperature-programmed-reduction signals, which are pro- portional to the amount of supported a-Fe,O,, with the nominal iron(II1) content and the calcination temperature are shown in fig. 2(b) and 3(b), respectively.Note that the values of the t.p.r. signal taken from the ref. (2) correspond to an amount of sample containing 2.2 mg of iron(II1). The variation of the activity with the dispersion of Fe3+ is shown in fig. 4 for three different temperatures. CONDUCTIVITY MEASUREMENTS The variation of the activation energy of conduction, Eg, determined for the Fe-X-665 catalysts with iron(II1) content is illustrated in fig. 2 (c). DISCUSSION INFLUENCE OF CALCINATION TEMPERATURE ON THE DISPERSION OF IRON(III) AND THE CONCENTRATION OF SUPPORTED Fe3+ SPECIES As stated above, the decrease in the amount of the supported a-Fe,O, observed in the range 250-320 "C is attributed to one of the following processes: (a) the high dilution of Fe3+ into the bulk of the support or (b) the formation of a strongly associated, and thus difficult to reduce, Fe3+ species on the carrier surface.Evidence as to which process actually occurs can be obtained by comparison of curves (a) and (b) of fig. 3. Such a comparison clearly demonstrates that the change in the concentration of a-Fe203 is not correlated with the change in the dispersion of Fe3+.P. POMONIS, D. VATTIS, A. LYCOURGHIOTIS AND CH. KORDULIS OL - Fe,O, Fe3'-S 2049 Fig. 5. Iron(rn) species present on the surface of the support for calcination temperatures > 320 "C. The absence of such a correlation in the range 250-320 "C suggests that process (a) does not occur. In fact, high dilution of Fe3+ in the range 250-320 "C should be shown by an abrupt decrease in the dispersion of Fe3+ in this temperature range, in disagreement with our experimental results.Thus it seems more probable that an increase in the calcination temperature from 250 to 320 "C promotes the formation of a strongly associated Fe3+ species on the surface of the support at the expense of supported a-Fe,O,. Brown et aZ.,18 working with iron catalysts, have proposed the formation of such a species to explain the appearance of a temperature-programmed- reduction peak at 950 "C. We denote this species as Fe3+-S (see fig. 5). Inspection of fig, 3(b) shows that a further increase in the calcination temperature does not cause a change in the concentration of supported a-Fe,O,. Additional but weak evidence for the occurrence of process (a) can be drawn from the constancy of the values of the binding energies.It thus seems reasonable to assume that a high dilution of Fe3+ into the alumina lattice would alter the Eb values, contradicting our experimental results. In view of the above considerations the decrease in the dispersion of iron(II1) on increasing the calcination temperature may, in principle, be attributed to one of the following processes : (a) solid-state transformation of a-Fe,O, into Fe3+-S, (b) surface diffusion of the loosely bound Fe3+ resulting in augmentation of the a-Fe,03 crystals or (c) an increase in the size of the Fe3+-S aggregates via surface diffusion of the strongly associated Fe3+. Careful examination of the results demonstrates that process (a) does not contribute to the decrease in the dispersion of Fe3+.In fact, this decrease takes place at a constant rate in the region 200-480 "C whereas the transformation of a-Fe,O, into Fe3+-S is complete at 320 "C. Moreover, since the rate of surface diffusion of the loosely bound Fe3+ must be higher than the rate of surface diffusion of the strongly associated Fe3+, the contribution of process (c) to the decrease in the dispersion is negligible. Based on the above considerations it seems reasonable to assume that process (b) is responsible for the decrease in the dispersion. Fig. 3(a) shows that the rate of this process increases at temperatures > 480 "C. The above does not exclude low dilution of the iron(n1) into the alumina lattice. INFLUENCE OF IRON(III) CONTENT ON THE PHYSICOCHEMICAL CHARACTERISTICS OF THE Fig.2(b) shows that the amount of supported a-Fe,O, remains almost constant in the region 0.173 < X < 0.615 and then decreases. This suggests that the formation of Fe3+-S at expense of a-Fe,O, is accelerated for X > 0.615. The absence of a discontinuity point on the smooth curve illustrating the drop in the dispersion with iron(m) content [fig. 2 (a)] implies that Fe3+ has a similar dispersion in both the a-Fe,O, and Fe3+-S phases. This is in excellent agreement with the conclusion stated above that the solid-state transformation of a-Fe,O, into Fe3+-S is not accompanied by a decrease in the dispersion of Fe3+. CATALYSTS2050 DECOMPOSITION OF N 2 0 ON Fe203/A1203 In line with the considerations mentioned above the decrease in the dispersion of Fe3+ is attributed to the increase in the size of the a-Fe203 crystals with increasing iron(II1) content.This is corroborated by the variation in the size of the a-Fe203 crystals with iron(rI1) content deduced from diffuse reflectance spectroscopy, magnetic measurements and X-ray spectroscopy. The activation energy of conduction [fig. 2(c)] shows a continuous decrease from pure alumina to X = 0.615, stabilising at ca. 1 eV. Bearing in mind the variation of the active-phase properties with iron(Ir1) content discussed above, the dependence of the Eg values on Fe3+ content can be interpreted in a simple way. The almost linear decrease of Eg up to X = 0.615 is justified because in this region the ratio of the Fe3+ ions distributed in the two supported iron(II1) species, i.e.a-Fe203 and Fe3+-S, is constant. Moreover, since the decrease in the dispersion of iron(II1) has no effect on the activation energy of conduction, an almost constant Eg value per mmol of Fe3+ is expected. The deviation from linearity is expected for X > 0.615 because in this range the above-mentioned ratio of Fe3+ ions is altered in favour of Fe3+-S, which is related to higher Eg values than a-Fe203. In conclusion, our electrical-conductivity results offer additional support for our view of the nature of the active phase of samples calcined above 320 "C. RELATIONSHIP BETWEEN PHYSICOCHEMICAL PROPERTIES AND CATALYTIC ACTIVITY Inspection of table 1 shows that the supported iron(rI1) ions are more active than the pure carrier.Moreover, from the smooth curve illustrating the drop in activity on increasing iron(rI1) content (fig. I), we obtain the following conclusions: (a) the semiconducting properties of supported iron(rr1) catalysts estimated by the activation energy of conduction do not govern the catalytic activity and (b) supported a-Fe203 and Fe3+-S have similar activity. If one of the above conclusions is not valid, then a discontinuity would appear in the curve at X = 0.61 5. The linear increase in activity with increasing dispersion of Fe3+ (fig. 4) suggests that the dispersion of the active phases is a key physicochemical parameter determining the catalytic activity. CONCLUSIONS In conclusion, we point out that the present study is a part of an attempt to obtain a better understanding of Fe203/A1203 catalytic ~ystems.l~-~~ The main results of this paper are as follows: (a) two species of iron(m), i.e.a-Fe203 and Fe3+-S, are formed on the carrier surface above 280 "C, and the ratio of the iron (111) species is constant from X = 0.172 to X = 0.61 5 mmol of Fe3+ per g of alumina, (b) Fe3+-S has a higher activation energy of conduction compared with a-Fe203, (c) an increase in the calcination temperature and iron(rI1) content causes a decrease in the dispersion of Fe3+ attributable to surface diffusion of the loosely bound Fe3+, resulting in an increase in the size of cc-Fe203 crystals, and (iv) the catalytic activity is determined by the dispersion of Fe3+ and not by the nature of the supported iron(II1) species nor by the semiconducting properties of the specimens.We thank the Services de Programmation de la Politique Scientific, Belgium for support (Ch. K.). A. Lycourghiotis and D. Vattis, React. Kinet. Catal. Lett., 1981, 18, 377. A. Lycourghiotis, D. Vattis, G. Karaiskakis and N. A. Katsanos, Rev. Chim. Miner., 1982, 19, 139. A. G. Keeman and R. D. Iyengar, J . Catal., 1966,5, 301. A. Cimino, R. Bosco, V. Indovina and M. Schiavello, J . Catal., 1966, 5, 271.P. POMONIS, D. VATTIS, A. LYCOURGHIOTIS AND CH. KORDULIS 205 1 M. L. Volpe and J. F. Ready, J. Catal., 1967, 7, 76. A. Cimino, V. Indovina, F. Pepe and M. Schiavello, Proc. 4th Int. Congr. Catal., Moscow, 1968, paper 12, p. 187. ' A. Cimino and V. Indovina, J. Catal., 1970, 17, 54. A. Cimino and F. Pepe, J. Catal., 1972, 25, 362. T. A. Egerton, F. S. Stone and J. C. Vickerman, J. Catal., 1974, 33, 299. lo T. A. Egerton, F. S. Stone and J. C. Vickerman, J. Catal., 1974, 33, 307. l 1 P. Pomonis and J. C. Vickerman, J. Catal., 1978, 55, 88. l 2 C. Defossk, P. Ganesson, P. G. Rouxhet and B. Delmon, J. Catal., 1978, 51, 296. l3 P. J. Angevine, J. C. Vartuli and W. I. Delgass, in Proc. 6th Int. Congr. Catal., ed. P . B. Wells and F. C. Tompkins (The Chemical Society, London, 1976), pp. 61 1-618. l4 S. C. Fung, J. Catal., 1979, 58, 454. F. P. J. M. Kerkhof and J. A. Moulijn, J. Phys. Chem., 1979,83, 1612. l6 M. Houalla, ScientiJic Basesfor the Preparation of Heterogeneous Catalysts, ed. G. Poncelet, P. Grange and P. A. Jacobs (Elsevier, Amsterdam, 1983), p. 273. l7 Ch. Kordulis, S. Voliotis, A. Lycourghiotis, D. Vattis and B. Delmon, Appl. Catal., 1984, 11, 179. R. Brown, M. E. Cooper and D. A. Whan, Appl. Catal., 1982,3, 177. l9 A. Andreev, E. Proinov, N. Neshev and D. Shopov, J. Catal., 1982, 74, 1. 2o F. F. Volkenshtein, Kinet. Katal., 1981, 21, 776. 21 M. Niwa, K. Yagi and Y. Murakami, Bull. Chem. Soc. Jpn., 1981,54,975. 22 B. Lover, J. Juhasz, K. Mihalyi and Z. G. Srabo, J. Res. Inst. Catal. Hokkaido Univ., 1980, 28, 223. 23 V. Perrichon, P. Turlier, J. Barrault, G. Forguy and J. C. Menezo, Appl. Catal., 1981, 1, 169. 24 D. Ying-Ru, Y. Qi-zie, H. Yuan-Fu, J. Yong-Shu and Q. Jin-Heng, Adv. Chem. Ser., 1981, 194, 609. 25 T. Tomov, D. Klissurski and I. Mitov, Phys. Status Solidi A, 1982, 73, 249. (PAPER 4/ 1630)