J. Chem. Soc., Faraday Trans. I , 1984, 80, 1447-1456 Kinetics of the Decomposition of Hydrogen Peroxide Catalysed by Copper and Nickel Ferrites BY ANTHONY I. ONUCHUKWU Physical and Industrial Chemistry Research Laboratory, Department of Chemistry, Bayero University, P.M.B. 301 1 , Kano, Nigeria Received 13th July, 1983 The kinetic activities of the heterogeneous decomposition of hydrogen peroxide by metal-iron spinel oxides, M,Fe,-,O, (M = Cu or Ni), have been investigated with a view to defining the effects of composition and microstructure on their catalytic activity. The entire composition range, 0 < x < 3, was prepared and characterized. Although the series of nickel catalysts was found to possess a lower aggregate diameter than the copper series, the decomposition activity of the former was surprisingly much less than the latter.This poor performance of the nickel series is explained in terms not of a consideration of microstructural defects but rather of the restricted redox couple represented by Mn/Mn-l in the electronic composition of the catalysts and, possibly, the absence of a more active ion (Mn) at high compositions on the octahedral lattice sites which may initiate the cyclic electron-transfer process on the catalyst surface, as proposed by Abel and Cota et al. The value of x corresponding to maximum activity was experimentally evaluated for each catalyst series by determining the highest specific rate constant having minimum values of the Arrhenius pre-exponential factor, rather than minimum values of the activation energy.The production of oxygen gas has for many years relied on the electrolysis of water. Compression in steel cylinders is possibly the only means of storage of the hydrolysis products. The choice of a suitable peroxide (H,O,) decomposition catalyst as an alternative for oxygen-gas production has proved difficult because the alternatives tried, silver oxide, platinum and palladium blacks, are expensive and therefore unattractive.lP3 Similarly, cheap corrosion-resistant catalysts such as MnO,, C0,03 and Fe,O, are not suitable because of the poor activity of these catalysts towards peroxide decomposition. 4-6 Cobalt-iron spinel oxides of the general formula Co,Fe3_,O4, where x, the composition variable, can take values between 0 and 3, were reported by Cota et al. to possess activity for peroxide decomposition in alkaline media.5 Goldstein et al.prepared the entire composition range of the cobalt-iron spinel oxides by hydroxide and oxalate routes and concluded that the cobalt catalysts produced by these techniques possessed similar structures and ~ ~ m p ~ ~ i t i o n . ~ - ~ Goldstein and Sat0 et al. lo reported high surface areas for catalysts produced via the hydroxide route. Thus the order of activity of the series of catalysts from the two routes was explained in terms of intrinsic factors, e.g. electronic composition and surface morphology of the catalyst p o ~ d e r s . ~ ~ The high activity of the cobalt-iron spinel oxide system towards peroxide decomposition was explained by a redox-couple mechanism in which the presence of Corl ions at the octahedral lattice sites of the cobalt spinel oxide structure initiated a cyclic electron-transfer proces~.~~ ' 7 Prior to the postulation of a redox mechanism, Erdey et al.stated that in peroxide reactions with oxidants containing oxygen the peroxide bond in the peroxide molecule is never dissociated, as was proved by isotopic 14471448 HETEROGENEOUS DECOMPOSITION OF HYDROGEN PEROXIDE measurements.12 These authors proposed a mechanism with a scheme involving an interaction between the peroxide molecule and the perhydroxide ion, HO;. l2 Recently13 other authors have suggested that peroxide decomposition in an alka- line medium proceeds via two mechanisms : the radical-chain mechanism, initiated by the ions of heavy metals (e.g.Cu, Fe or Mn), and the non-radical route, oia unspecified intermediates. However, in the preparation of hydrogen peroxide by the reduction of oxygen in 5 mol dm-3 KOH (to provide sufficient conductance) it was found that traces of any of the heavy-metal ions decomposed the perhydroxide ion in situ, through the redox-couple Interest in the kinetics of peroxide decomposition in dilute solutions of alkali hydroxides stems from its use in oxygen production, bleaching cellulose and textile materials and the electrolytic reduction of ~ x y g e n . ~ ~ lo* 1 5 9 l6 According to Burki et aL2 and Weiss4 peroxide decomposition in alkali-metal hydroxide solutions is a first-order reaction with respect to the peroxide, the decomposition rate increasing with hydroxide concentration.Cota et al. and Parravano also reported first-order kinetics for a series of cobalt-iron spinel The authorlg? 2o has recently reproduced the first-order reaction for cobalt spinel oxides and established that maximum activity in the series occurs at x = 1.5. The present paper describes an investigation of the peroxide decomposition reaction with copper-iron and nickel-iron spinel oxide catalysts of general formulae Cu,Fe,-,O, and Ni, Fe3-, O,, respectively. This study takes cognizance of the redox-couple mechanism in both catalyst systems represented by CuI1/Cu1 and Ni1I1/NiI1. The structural and electronic composition, as well as the surface morphology, of the entire catalyst series have been considered. A comparison of the performance of these spinel oxides with the previously reported series of cobalt catalysts revealed that the copper series is of comparable activity to the cobalt series.However, copper-iron spinel oxide is cheaper than cobalt-iron oxide. The decomposition activity of nickel-iron oxide series is found to be much lower than copper, despite its lower aggregate diameter. The maximum value of x was evaluated in each case by comparing the specific rate constant, activation energy and pre- exponential factor of the two catalyst systems. Thus the activity order in each series was established. The difference in the efficacy of these catalyst series towards peroxide decomposition was found to be influenced by their electronic composition and diffusion effects arising from surface morphology.13+ l4 EXPERIMENTAL PREPARATION OF M, Fe,-, 0, CATALYSTS Two different metal-iron spinel oxides, namely copper-iron and nickel-iron, were investigated. The entire composition range between x = 0 and 3, differing in microstructure and composition, was synthesized by hydroxide coprecipitation following the method of Sat0 er uE.,Io this technique being preferred for preparing high-surface-area catalyst series. Details are given in ref. (2), (6) and (21H23). Oxides with values of x of 0, 0.5, 1.0, 2.0, 2.5 and 3.0 were prepared. CHARACTERIZATION OF THE SPINEL OXIDE SERIES To ensure that the spinel or any other phase had been formed, the oxide samples were subjected to the Debye-Scherrer X-ray analysis technique. The work was carried out on a PlOlO X-ray generator using appropriate filters, Lindemann glass tubes of 0.5 mm internal diameter and an analytical camera of 1 1.46 cm length.The values obtained for the d-plane spacings and their relative intensities of reflection were compared with those from ASTM literature in order to determine the major phases of spinel present. Scanning electron microscopy (SEM) and EDAX analyses were conducted on samples in theA. I. ONUCHUKWU 1449 Cu and Ni spinel oxide series in order to determine particle-size and shape distributions as well as the relative proportions of cations in the spinel series. An Electron Optics model JEMIOOB microscope with an EDAX attachment was used in this investigation. The microstructures of the catalyst series powders were characterized by visual studies using a high-resolution ( x 100) optical microscope and the Coulter-counter particle-sizing technique.An electronic Coulter counter was used by ultrasonically dispersing the powder particle in an isotonic aqueous electrolyte and evaluating the mean aggregate size of the whole oxide powder series. The Brunauer, Emmett and Teller (B.E.T.),* specific surface areas of the powder series were measured by nitrogen adsorption at 80 K, after degassing of the catalyst at 298 K for 1 h.21,24 DETERMINATION OF KINETIC ACTIVITY OF THE PEROXIDE DECOMPOSITION REACTION The kinetic activity of the spinel oxide catalysts for H,O, decomposition was evaluated by the rate of production of oxygen gas in the liquid phase. A constant catalyst weight (50 mg) was injected into a thermostatted reaction vessel containing 5 cm3 of 0.4 mol dm-3 H,O, in 5 mol dm-3 KOH as a diluent for each oxide specimen.The mixture was stirred vigorously using a magnetic stirrer. The alkali and peroxide were AnalaR and Hopkins & Williams reagents, respectively. Both solutions were standardized immediately before use, KOH using standard HC1 ampoules and H,O, using KMnO, solution. Oxygen gas evolved from the reaction vessel according to the equation ks 2HOi + 0, + -OH (9 where the peroxide was present in alkaline solution as the perhydroxide ion, HO;.ls The rate of oxygen evolution was monitored using the gasometric assembly of Cota et aL5 This technique has been used extensively in previous work and described in detail e l ~ e w h e r e . l ~ * ~ ~ $ ~ ~ The timedependent volume, &, of evolved oxygen was monitored at 30 s intervals in all cases studied.The maximum recorded, V,,,, was also indicative that the reaction was complete. Values of the composition variable, x, and the Arrhenius parameter^^^ were studied within the temperature range 300-3 12 K. In a stability study the catalyst retrieved from the reaction vessel was stored in 5moldm-3 KOH in air, and the time taken to attain V,,, was monitored periodically for 72 days. RESULTS X-ray characterization of the oxide series revealed that using deionized distilled water to prepare the solutions produced impurity-free spinel oxides. The effect of x on the particle-size and cation distribution of each system determined by SEM is displayed in tables 1 and 2 for the two spinal systems.Fig. 1 shows the effect of varying the composition on the specific surface area and mean aggregate diameter (d/pm) determined by the B.E.T. and Coulter-counter techniques, respectively. In fig. 2 the volume ( 5 ) of 0, is plotted against the time ( t / s ) of evolution for 50 mg of each of the two catalysts for x = 1.5. These plots were corrected for self- decomposition of the peroxide, i.e. the 0,-evolution rate at 298 K is equal to 1.3 x lop4 cm3 s-1.26-28 The Cu and Ni spinel systems investigated were found to follow a rate law which was first order with respect to peroxide in 5 mol dmP3. The first-order rate constant defined for reaction (i) is (ii) where V, is the initial volume of 0, gas produced by self-decomposition, [HO;], is the initial concentration of the peroxide (0.4 H,O,)mol dm-3 and k, is the rate constant.In order to enhance reproducibility, some authors5* '9 lo, 2o have expressed the peroxide decomposition rate constants per unit mass of catalyst, i.e. ks/s-l g-l. A Commodor microcomputer model PET2001-16NB5 was used to obtain the rate 48 F A R 11450 HETEROGENEOUS DECOMPOSITION OF HYDROGEN PEROXIDE Table 1. Properties of the copper-iron oxide series B.E.T. specific spinel cation valency distributions surface X composition in the spinel structurea area/m2 g-' 0 Cu,Fe,O, (Fe3+)tet[Fe2+Fe3+]oct 0;- 135 Oe5 Cu0.5Fe2.504 (Fe3+)tet[C~~5Fe$=Fe2+loct 0;- 92 1.0 Cu,FE,O, (Fe;+)tet[Cu+Fe3+]oct 0;- 86 2.0 Cu,Fe,O, ( C U + ) ~ ~ ~ [ C U ~ + F ~ ~ + ] ~ ~ ~ 0;- 64 2*5 CU2.5Fe0.504 (Cu+)tet[Cu~;Fei3oct 0;- 66 3.0 Cu,Fe,O, (CU+)tet[CU2+CU+]oct 0:- 78 (inverse) (inverse) Cu1.5Fe1.504 (Cu~5Fe~~)tet[Cu~5Cu$~Fe3+loct 0;- 74 (normal) (normal) ~ ~~ a tet = Tetrahedral site, oct = octahedral site.Table 2. Properties of nickel-iron oxide series B.E.T. specific spinel cation and valency distributions surface X composition in the spinel structurea area/m2 g-l 0 Ni,Fe,O, (Fe3+)tet[Fe2+Fe3+]oct 0;- 135 (inverse) (normal) (inverse) 0.5 Ni,,,Fe,.,O, (Fe3+),,,[Nif:',Fe$;Fe3+],,, 0;- 120 1.0 Ni,Fe,O, (Fe3+)tet[Ni2+Fe3+]oct 0;- 86 1.5 Nil.5Fel.,0, (Ni~?Jtet[Ni~~Nii~Fe3+],,, 0;- 84 2.0 Ni,Fe,O, (Ni2+)tet[Ni3+Fe3+]oct 0;- 81 2.5 Ni,.,Fe,.,O, (Ni2+)tet[Ni3+Fe3+]oct 0;- 83 3.0 Ni,Fe,O, (Ni2+)tet[Ni3+Ni2+]oct 0;- 68 (normal) a tet = Tetrahedral site, oct = octahedral site.constant (k,) for various values of x at temperatures between 300 and 312 K. The reaction rate constants were independent of initial peroxide concentration and a first-order rate plot was linear for over 95% of the total reaction period. Also, the first-order rate constants were directly proportional to the steady mass in the reaction mixture for the weight range 20-100 mg considered. The effect of x on k, is shown for two temperatures (300 and 308 K) in fig. 3. The maximum error check in the programmed solutions of k, values is _+ 1.5% and reproducibility was good. Fig. 4 depicts the influence of the diffusion effect on the Arrhenius activation energy (E,) for each oxide in the two catalyst systems. The activation entropy of decompositionA. 1. ONUCHUKWU 1451 120 3 I w N E .cd cd h I I 0 1 .o 2.0 3.0 X Fig. 1. B.E.T. surface area (triangles) and mean aggregate diameter, d/pm (circles), for Cu, (closed symbols) and Ni, (open symbols) plotted as functions of x. 0 200 LOO 600 t Is Fig. 2. Volume, 5, of oxygen evolution plotted against t / s for Cu, (A, A) and Ni, (0, 0) systems at 300 K. (---) Unstirred and (---) stirred systems. in 5 mol dm-3 for the two series was evaluated by a programmed solution ot'the Eyring equation : (iii) where AH$ and AS$ are the activation enthalpy and entropy changes, respectively, and the other symbols have their usual meanings. For identical values of x (1.5) and under vigorous stirring the Cu and Ni catalysts gave values of 56.3 and 29.7 kJ mol-l, respectively, for the enthalpy of activation.A plot of the effect of diffusion on the variation of entropy of activation with x is shown in fig. 5. The changes in activation enthalpy and entropy evaluated in this study were subject to an experimental error 48-21452 HETEROGENEOUS DECOMPOSITION OF HYDROGEN PEROXIDE 3f 2E 20 " I M I --. - Atm 12 . 4 0 i 1 .o 2.0 3 .O X Fig. 3. Variations of k,/s-l g-l with x for Cu, (m, 0) and Ni, (a) at 300 K and Cu, (0, 0) and Ni, (A, (>) at 308 K. (-) Unstirred and (---) stirred systems. of & 1.5 kJ mol-l. It is also important to note that a stability test on the catalyst for a period of 72 days showed no deterioration in the peroxide decomposition reaction. DISCUSSION Of the various methods reported for the preparation of spinel oxides, the copre- cipitation method of Cota et al.was adopted in this study because the technique produces higher-surface-area oxide powders. Characterization of these high-surface- area powders by X-ray diffraction and EDAX revealed that the spinel oxides prepared were impurity free. Unlike several studies5* 9 v 28 two different catalyst systems (Cu and Ni spinel oxides) have been produced, with differences in composition and microstructure. Thus the significant objective of this work vis u vis the peroxide decomposition reaction was to arrive at an intrinsic order of catalyst activities that is far more dependent on catalyst composition than microstructural differences. In tables 1 and 2 we show the composition (cation distribution) and surface morphology, and specific surface area (in m2 g-l) of the two oxide systems.Again, the variation of mean aggregate diameter and specific surface area with catalyst composition (fig. 1) indicates that the nickel oxide catalysts have much smaller aggregate diameters, seen by the high specific surface area recorded. One possible approach to ascertainingA. I. ONUCHUKWU 1453 \ \ I I I 0 1 .o 2 .o 3.0 Fig. 4. Activation energy as a function of x for Cu, (A, A) and Ni, (m, 0). (-) Unstirred and (---) stirred systems. X the intrinsic activity is to make use of a kinetic parameter which is effectively independent of the catalyst microstructure for each of the two oxide series. A suitable choice of such a parameter would be the activation energy or entropy for the peroxide decomposition reaction.These kinetic parameters are dependent on catalyst composition rather than surface 28 A plot of 0, volume ( q ) against the time of evolution (fig. 2) revealed that the activity of the copper catalyst superseded that of the nickel catalyst under similar experimental conditions. In contrast to the findings of Spalek et aZ.,lS the decomposition of peroxide in 5 mol dm-a KOH obeys the first-order rate equation (ii) with respect to the total peroxide content. This finding is consistent with first-order behaviour for peroxide decomposition in alkaline media reported by several author~.~-~$ ' 9 2o It is, however, interesting to note that despite the observed differences in performance (fig. 2), the variation of the rate constant, k,, with composition (fig. 3) determined at two different temperatures (300 and 308 K) is more interesting. First, an increase in temperature enhanced the activity of the two oxide systems.This increase is more marked for the Cu series than the Ni series (fig. 3). Secondly, although an increase in composition enhanced the activity of Cu catalysts to a maximum at x = 2.0 the nickel catalyst series deteriorated in performance for an identical increase in catalyst1454 HETEROGENEOUS DECOMPOSITION OF HYDROGEN PEROXIDE '*r 0 1 .o 2.0 3.0 A Fig. 5. Effect of stirring (---) on the variation of activation entropy with x in the Cu, (A, A) and Ni, (0, 0) systems. composition. These differences in performance, particularly for the Ni series despite its microstructural advantage (low aggregate particle size), suggest that compositional effects operate within the catalyst series. Hence the poor activity of the Ni catalyst series suggests that an increase in composition inhibits the decomposition reaction ; this is because it is possible that the redox-couple mechanism postulated in earlier studies is hindered, as the formation of nickel as Nil* is more favourable than in a higher oxidation state on the iron oxide support.lV 2 t 4 9 l1 Although the Cu series is favoured by an increase in x, the fall in activity for x > 2 is surprising. However, at x > 2 the Cu series catalysts have a lower mean aggregate diameter, reflected in the high surface area as shown in fig.1. An explanation for this observed anomally may be found by considering diffusion-control effects in the heterogeneous peroxide decomposition reaction.It was observed during the course of the experiment that although 0, evolution produced pronounced turbulence, the high reaction rate with the Cu series was dependent on vigorous mechanical stirring, (fig. 2 and 3). According to Goldstein et al.,9928 who studied the decomposition of peroxide by cobalt spinel oxide, when reaction turbulence occurs near the surface of a solid dispersed in a liquid, a thin liquid layer adheres to the solid surface. Furthermore, they reported28 that for peroxide decomposition characterized by a three-phase [liquid- solid-gas (oxygen)] heterogeneous reaction, the ingress of fresh reactant (H20,) through such a layer to the solid (catalyst) surface to sustain the reaction becomes restricted. This phenomenon would be more serious for solids with cracks and micro- pores where gas bubbles could be occluded.Hence, partial mitigation by vigorous stirring of the reaction indicates that diffusion effects may possibly control the overall decomposition kinetics, especially for high-surface-area solids observed for Cu(x > 2)A. I. ONUCHUKWU 1455 and the entire Ni series. Although an increase in temperature enhanced 0, evolution (fig. 3), the values of the activation energy (fig. 4 and tables 1 and 2) were not consistent with this diffusion-control phenomenon. Despite the vigorous mechanical stirring adopted during the reaction, effects of diffusion control could not be mitigated completely. It may be that the diffusion effect is significant in this study because an increase in the mass of the catalyst provides more particles in which diffusion layers could be set lo, 2o This explains the dependence of the rate constant on the catalyst mass (fig.3), suggesting that diffusion kinetics would be dependent on composition. In order to verify this reasoning, the activation energies (fig. 4) and the entropy (fig. 5), which depend more on composition, were evaluated. The maximum error is of the order of 1.5 kJ mol-l. There is in each case a pronounced minimum at x = 2 and x =2.5 for the Cu and Ni series, respectively. The increase in E: beyond this valley in each system is consistent with a lower mean aggregate diameter, which promotes diffusion control of the catalyst activity, leading to an increase in activation energy (fig. 4).However, vigorous stirring of the reaction has minimized the defusion effect. The restricted redox-couple mechanism (NiI1/Ni1I1) and the increase in composition with minimum surface area could account for the poor activity of the Ni spinel system. Hence, the low activation energy recorded for the Ni oxide series is suspect, because steric hinderance accompanying the attainment of Ei is not reflected in the activation entropy, as shown in fig. 5. The values of the AS$ in the peroxide-catalysed decomposition could be related to the more easily available redox couple CuI1/Cul or the entropy of adsorption of the peroxide on the catalyst. This postulation is reflected in the lower ASS values recorded for the Cu series than for the Ni system.This finding is consistent with the increase in A S (at x > 2.0 and x > 2.5 for the Cu and Ni catalysts), for which a decrease in surface morphology could be responsible for the increase in steric hinderance. Thus the variation of entropy with composition (fig. 5 ) is considered to be significant in the overall assessment of the catalyst systems. Nevertheless, the activities of the two catalyst series may be compared on the basis that the more effective catalyst for peroxide decomposition possessed a lower entropy without a corresponding low activation energy. In this work an attempt has been made to establish the activity order of the two catalyst systems considered using the composition factor. Thus for the two temperatures studied, 300 and 308 K, the order of activity for the copper spinel oxide series is 3.0 < 0.5 < 1 .O < 1.5 < 2.0 < 2.5, while the nickel-containing series maintain the order 0.5 > 1.0 > 1.5 w 3.0 > 2.0 w 2.5.Also, catalysts stored in 5 mol dmP3 KOH for a test period of 72 days showed no deterioration in stability. In conclusion, this work has established the significance of catalyst composition towards the enhancement of catalyst activity in the peroxide decomposition reaction. However, it is equally important to match the microstructural factors with composition in order to achieve optimum efficiency for metal-iron spinel oxides. I thank the technical staff of the Physics Department of The City University, London, for the use of X-ray and electron micrographic facilities.I also thank Mr P. B. Mshelia for preliminary work, and the Research and Higher Degrees Committee, Bayero University, Kano for financial support. G. Tammann, Z . Phys. Chem., 1889,4,441. F. Burki and F. Schaaf, Helu. Chim. Acta, 1921, 4, 418. W. Vielstich, Z. Phys. Chem., 1958, 15, 409. J. Weiss, Trans. Faraday SOC., 1935, 31, 1547.1456 HETEROGENEOUS DECOMPOSITION OF HYDROGEN PEROXIDE H. M. Cota, J. Katan, M. Chim and F. J. Schoenweis, Nature (London), 1964, 203, 1281. W. C. Schnumb, C. N. Satter-Field and R. N. Wentworth, H202 (Reinhold, New York, 1955). J. R. Goldstein and A. C. C. Tseung, J. Muter. Sci., 1972, 7, 1383. A. C. C. Tseung and J. R. Goldstein, J. Phys. Chem., 1972, 76, 3646. J. R. Goldstein, Ph.D. Thesis (The City University, London, 1970). lo T. Sato, M. Sugihara and W. Saito, Rev. Electron Commun. Lab., 11, 1963, 26. l1 G. Blasse, Phillips Res. Rep., 1963, 18, 383. l2 L. Erdey, Acta Chim. Acad. Sci. Hung., 1953, 3, 95. l3 Kh. Raskina, F. I. Sadov and G. A. Bogdanov, Zh. Prisk. Khim., 1970,43,447; 1449; 1977,50,724. l4 K. Balogh, J. Balej and 0. Spalek, Chem. Zvesti, 1976, 30, 385; 61 1. l5 0. Spalek, J. Balej and K. Balogh, Collect. Czech. Chem. Commun., 1970, 42; 312; 394. l6 0. Spalek, J. Balej and I. Paseka, J. Chem. SOC., Faraday Trans. I , 1982, 78, 2349. l7 E. Abel, Monatsh. Chem., 1952,83,422. G. Parravano, Proc. Int. Congr. Catal. (1970), vol. 1, p. 157. A. I. Onuchukwu and A. C. C. Tseung, U S . Patent Application 4,335,754, June, 1982; U.K. Patent Application 2,064,587, November, 1980. 2o A. I. Onuchukwu, J. Catal., submitted for publication. 21 A. I. Onuchukwu, Electrochim. Acta, 1982, 27, 529. 22 A. I. Onuchukwu, J. Electrochem. SOC., 1983, 130, 1077. 23 A. I. Onuchukwu and P. B. Mshelia, J. Chem. Educ., in press. 24 P. H. Emmett, Measurement of the Surface Area of Solid Powders (Reinhold, New York, 1954), 25 M. J. Pilling, Reaction Kinetics (Oxford University Press, Oxford, 1975), vol. 1, p. 10. 26 K. B. Keating and A. G. Romer, J. Phys. Chem., 1965,69, 3658. 27 M. Ardon, Oxygen, Elementary Forms and Hydrogen Peroxide (Benjamin, New York, 1965), p. 84. 28 J. R. Goldstein and A. C. C. Tseung, J. Catal., 1974, 32, 459. V O ~ . 2, pp. 31-74. (PAPER 3/1210)