The rate constants for the heterogeneous decomposition of hydrogen peroxide ... The measurements of the reaction rate of hydrogen peroxide decomposition.
69
J. Electnmnal Chem., 339 (1992) 6943 Elsevier Sequoia S.A., Lausanne
JEC 02132
Study of the heterogeneous decomposition of hydrogen peroxide: its application to the development of catalysts for carbon-based oxygen cathodes * H. Falu5n and R.E. Carbonio * l
de Investigaciones en Fisicoqdmica de Gkioba (INFIQC), Dpto. de F&ico Quiinica, Facultad de Ciencias Quiinicas, Universidad National dr Ghbba, Summa1 16, CC. 61, 5016 C&Ma (Argentina)
htimt~
(Received 9 March 1992)
The heterogeneous decomposition of hydrogen peroxide was studied in alkaline electrolytes on LaFe,Ni,_,03 perovskite-type oxides. The reaction is first order in HO, concentration. Maximum catalytic activity was found for x = 0.25. This increase in activity was attributed to the appearance of mired oxidation states of iron and nickel. The optimum temperature of synthesis was between 600°C and XMPC, this being a compromise between a sufficientiy high temperature to synthesize the perovskite structure and a suf&iently low temperature to producea high surface area catalyst with a low number of oxygen vacancies.. The catalytic activity for 0, reduction on carbon-based electrodes correlates very well with the catalytic activity for HO; decomposition. A linear relationship was derived between the electrode potential at constant current and the logarithm of the peroxide decomposition rate constant. This relationship is followed quite well at relatively low current densities.
INTRODUCTION
During the last decade much effort has been directed toward the development of electrode materials with high catalytic activity for the electrochemical reduction of oxygen, because of the importance of this reaction in fuel cells, industrial water electrolysers and secondary metal-air batteries [l-12]. For commercial applications noble metals are generally used as catalysts; however, for low temperature fuel cells there are cheaper alternatives, such as transition metal oxides [4,6,11-U]
l Dedicated to Professor Ernest Yeager on the occasion of his retirement and in recognition of his contriiution to electrochemistry. ** To whom correspondence should be addressed.
0022-072S/92/$05.00
8 1992 - Elsevier Sequoia Sh
Ah rights reserved
and transition metal macrocycles [2,16-221 incorporated into gas-diffusion type PTFE-bonded carbon electrodes. On carbon, oxygen is reduced mainly through the two-electron reaction, to form hydrogen peroxide [2,5]; consequently, the catalyst incorporated into the carbon-based electrode has to be effective at decreasing the hydrogen peroxide concentration in order to reach sufficiently high operation potentials. For this reason, the discovery of good catalysts to be incorporated into carbon-based air electrodes requires the development of new materials for hydrogen peroxide decomposition in aqueous solutions: The aim of this work is to examine the hydrogen peroxide decomposition reaction on LaFe,Ni,_,O, perovskite-type oxides and to correlate the catalytic activity for this reaction with the activity for oxygen reduction on Teflon-bonded carbon electrodes loaded with LaFe,Ni, _xO, catalyst. EXPERIMENTAL
Catalyst preparation and characterization The LaFe,Ni,_,O, perovskite catalysts were prepared by citrate amorphous precursor decomposition [7,23,24]. A concentrated solution of citric acid was added to a concentrated solution of the metal nitrates until the molar ratio of citric acid to total metal was unity. The solution was evaporated in a. rotary evaporator at 90°C until the precipitate acquired the consistency of a viscous syrup. The residual water was evaporated in a vacuum oven at 1loOC for 24 h. The precursor thus obtained, is ‘a fine mixture of the nitrates and citric acid. After being burned at 200°C in air, the precursor was heat treated in air at temperatures T’,, between 500 and 1100°C for 12 h, and then slowly cooled to room temperature for a period of 8 h; The powders thus obtained were analysed by X-ray diffraction with a Philips PW 1140 diffractometer using Cu I&Y radiation, at a sweep rate of 2 deg min-‘. Catalyst surface areas were measured by the B.E.T. method with STROLEHIN Area-Metter II equipment. Hydrogen peroxide decomposition reaction The rate constants for the heterogeneous decomposition of hydrogen peroxide were measured by the gasometric method [13,25,26]. A small amount of catalyst G-20 mg) was dispersed in 45 ml of O,-saturated KOH (Merck p.a.1 solution at room temperature. A 5 ml aliquot of 0.70 M hydrogen peroxide standard solution presaturated with O2 was injected into the reaction vessel containing the stirred suspension of the catalyst. The final concentration of KOH was 5 M. The hydrogen peroxide solution was prepared from 30% solution (Carlo Erba RSE grade without stabilizer). A 0.17 M KMno, solution titrated with sodium oxalate was used to titrate the hydrogen peroxide standard solution,
71
In order to show that homogeneous decomposition is negligible, the following experiment was performed: 20 mg of catalyst were dispersed in 50 cm3 of 5 M KOH and allowed to be in contact with the solution for 2 h. The solution was then separated from the catalyst by filtration and the catalytic activity of the supernatant solution was measured. The measurements of the reaction rate of hydrogen peroxide decomposition were carried out by measuring the volume of evolved 0, as a function of time. The experimental assembly is similar to those described in the literature [13,26]. The concentration of HO;(c,ol) at time r is calculated as: C HOT =
co- [2Po*Vo,/RT KS,*]
(1)
where co is the initial concentration of HO;, Vo, is the volume of evolved 0, at time f, V,, is the volume of solution and po, is the 0, partial pressure (atmospheric pressure corrected for the water vapour pressure). Electrochemical reduction of 0, Floating gas-fed electrodes [25,27,28] were used as working electrodes for the electrochemical measurements. Their preparation was as follows. Teflon T30B (DuPont) suspension was diluted to approximately 2 mg cmV3 in water and slowly added to an aqueous suspension of 34 mg of perovskite and 34 mg of Shawinigan black (Gulf Oil Corporation, Shawinigan Products Dept.) under ultrasonic agitation until 20% w/w of Teflon based on the total weight was reached. The suspension was filtered using a 1 pm pore size filter membrane. The resulting paste was kneaded with a spatula until slightly rubbery, applied to a 0.5 mm thick disk of Teflon-carbon black hydrophobic backing material containing a nickel mesh (Electromedia Corp., Englewood, NJ), put into a 1.30 cm diameter die and then pressed at a pressure of 380 kg cme2. The electrode was finally heat treated at 300°C for 2 h in flowing N,. The electrode (0.5 cm2 geometric area) was positioned at the electrolyte surface with a stream of 0, passing over it to maintain a constant pressure of 1 atm at its back. Steady-state polarization curves were performed galvanostatically and were then corrected for the solution-phase IR drop external to the gas-fed electrode. All the measurements were performed at 25°C. RESULTS AND DISCUSSION
X-ray difiaction
and sugace area measurements
The room temperature powder X-ray diffractograms for the compounds prepared at different temperatures are shown in Fig. 1 for LaNiO, and LaEe,,Ni,,sO,. At 500°C the diffractograms show, for both compounds, the presence of the unreacted precursor; however, at T 2 600°C the peaks corresponding to the perovskite structure start to develop. These results show that the
72
L L
1
l-L-_L StUC
m
L.J__-L L_
2
3
220
m
2m-c
%ouc4
222
A-42
L I
40
I
m-
-
’
Fig. 1. Powder X-ray diffractograms for hvo selected catalysts prepared at different temperatures, the T,,, values are indicated in the figures: (a) LaNiO,, (b) LaFeO,,Ni,O,.
minimum temperature necessary to obtain a single perovskite phase is 600°C. This is lower than the minimum temperatures reported by T&n et al. 1231for LaNiO, and LaFeO,, probably owing to the fact that in the present. work we used longer times of calcination. Minimum temperatures of synthesis in the range SO-650°C
73 TABLE 1 Surface areas A determined by the B.E.T. method A/m2 g-’
L&t /“c
~eo.d%7503 600 700 800 1100
10.0
LaNi03 3.6 3.0 3.3 1.0
4.1 1.0
were also found by Zhang et al. [29] for perovskites, using the amorphous citrate precursor. We could index the diffractograms of the compounds prepared at 8CMYC with the rhombohedral system (Miller indexes shown in the figure). This is in agreement with the fact that in the LaFe,Ni,_,O, solid solutions, the compounds have the rhombohedral structure of LaNiO, at x < 0.5 [30]. The refined paramet,ers obtained with the computer program LATTICE [31[ were a = 5.399 f 0.007 A, a = 60.83” f 0.03” for LaNiO, and a = 5.399 f 0.006 A, a = 60.51” f 0.03” for LaFe,,Ni,,O,. The parameters obtained for LaNiO, are in good agreement with those in the literature [32]. The increase observed in the number of reflections in the diffractogram of LaNiO, as the temperature of synthesis increases is probably due to the fact that at high temperatures the number of oxygen deficiencies increases and the structure of the compound transforms into a triclinic structure [32]. More detailed structural studies will be presented in another publication. The surface areas of LaNiO, and LaFe,,Ni,,O, prepared at different temperatures are shown in Table 1. For LaFe0.uNi,,50, there is a clear decrease in the surface area as Tsint rises; however, for L&O, it remains approximately constant between 600 and 800°C and then decreases at 1100°C. The influence of these changes on the catalytic activity will be discussed later. Heterogeneous decomposition of hydrogen peromik For a heterogeneous v,+A+v~B+
reaction:
. . . -,+P+v~Q+
...
(2)
that takes place in a volume I$,, of solution that contains a catalyst of mass meat and surface area A, the catalytic rate per unit area of catalyst in mol cm-* s-l is defined by Spiro [33] as: 1 Ucat =--Av,
dn, dt
where nA represents the number of moles of A and t is time. The quantity u,~ is
74
sometimes termed the area1 rate of reaction [34]. Division by meat instead of A gives a related quantity often called the specific rate of reaction [34]: 1
dn*
(4)
unl= --m&VA
dt
Since the reaction is generally followed by measuring the concentration of either the reactants or products, then eqns. (3) and (4) can be expressed more conveniently as: &, u cat =
-
dcA -
-
Av,
(5)
dt
and Vsol unl=
dcA
(6)
--mcatVA
dt
If the reaction is xA order in A as a reactant, then: khetciA
v., -de,
=- -
k&%
- --
(7)
dt
vA
Vsol
dcA
mcatvA
(8)
dt
where khet is a heterogeneous rate constant, with units of cm s-l for the case of xA = 1 and k, is a specific rate constant with units of cm3 s-l g-’ for xA = 1. k, is useful from the practical point of view, when a series of catalysts are compared in experiments using a constant weight of catalyst; k,, is correlated with the true catalytic activity per unit surface area. For the hydrogen peroxide decomposition reaction: HO;(aq)
--) &(g)
+ OH-(aq)
(9)
the rate constant which is calculated directly from the experimental data is the pseudo-homogeneous first-order rate constant k,,,,,,, (sml), which is defined as: k horn
=
d(ln c,?)/dt
The heterogeneous rate constant khet and the specific rate constant calculated from k h,,mrespectively as follows: lthet = [V~ikhornl/[mcat(A/m)cat]
(10) k,
are (11)
and k, =
(12)
[v~lkhoml/[m~~~
where (A/m),,
is the specific area for the catalyst.
75
- 1 .oo
i_ -i; E \ b” B 3
-1.25
- 1.50
- 1.75
I
I
0
40
20
60
Time/min Fig. 2. log cHoi
vs. time, meat =0.005
g, ~=0.074
M, T,i”,=lloo”C:
*, LaFeo,,Nio,O,;
*,
~Feo,~Nidh.
Plots of log cue, vs. time are shown in Fig. 2 for two different catalysts. The good linearity indicates first order in HO; concentration. This is in agreement with results reported in the literature for spine&type oxides 135-381 and perovskite-type oxides [25]. The observed decomposition of HO, can be shown to involve only heterogeneous catalysis by running an experiment as mentioned in the value obtained in this experiment was less than experimental section. The k,, 1% of the value obtained in the same solution containing the same amount of catalyst. The influence of the temperature of synthesis on the catalytic activity was analysed by measuring the k, values for constant weights of LaNiO, and LaFe,,Ni,,O, prepared at different temperatures. The results are shown in Fig. 3. Very low activities are observed at TSt,,,= 500°C. These results together with those in Fig. 1 show that in order to obtain reasonably good activity for HO; decomposition, formation of the perovskite structure is necessary. A maximum in the catalytic activity is obtained at 700°C for LaNiO, and at 600°C for LaFe0.uNi,,03, then the activity decreases to reach very low values at llOO°C. The decrease in activity as Tsint rises is partly due to sintering of the catalyst particles and as a consequence to the decrease in surface area (Table 1). A decrease in surface area with increasing temperature was reported for La, _xSrxMnO, prepared using the same method [24]. There is another factor which must be taken into account. It is very well known that these oxides are oxygen deficient [30,32] and the number of oxygen vacancies increases as T&t increases. An increase in the number of oxygen vacancies will produce a decrease
76
im
5
-J
,400
600
800
1000
1200
Trmt ‘“C
Fig. 3. k,
vs. Tint, m,,
= O.O05g, co = 0.074 M: (a) LaFe,,Ni,,O,,
(b) LaNiO,.
in the average oxidation state of iron and nickel in LaFe,Ni,_,O, and, as will be shown later, this also decreases the activity. The maximum value in activity is thus a compromise between a temperature sufficiently high to obtain the perovskite structure and a temperature sufficiently low to produce a high surface area catalyst with a low content of oxygen vacancies. The influence of catalyst composition on the catalytic activity was analysed by measuring the’ k, values for constant weights of LaFe,Ni,_,O, obtained at Tsint= 1100°C as a function of x. The results are shown in Fig. 4. A maximum is obtained at x = 0.25 and negligible activity for LaFeO, is observed. This increase in catalytic activity for intermediate compositions of the solid solution, compared with the two end members L.aNiO, and LaFeO,, a phenomenon which is often called synergism, might be explained in terms of changes in the solid state properties. Yeager and co-workers [7,39] reported that in LaFe,Ni,_,O,, part of the Fe(IIf) is oxidized to Fe(W): and part of the Nit1111 is reduced to Ni(I1).
77
* 0.0
0.2
0.4
0.6
0.8
1.0
X Fig. 4. k, vs. catalyst composition (x in LaFe,Ni,_,O,),
meat = O.O2Og,co = 0.066 M, T& = 1100°C.
Moreover, Rao et al. [30] reported that based on the Goodenough-Kanamori superexchange rules [40], the electron transfer Fe(III)(t&e:)
+ Ni(III)(t!&e:)
--$Fe(IV)(t&ei)
+ Ni(II)(t!&ei)
(13)
is likely to occur. The appearance of Fe(IV) when Ni(I1) is oxidized to NXIII) in battery-type nickel hydroxide electrodes doped with iron hydroxide has also been shown using Miissbauer spectroscopy [41]. LaFe,Ni,_,O, perovskite-type oxides can thus be considered to be mixed valence catalysts with the formula La[Fe(III)Fe(IV)],[Ni(III)Ni(II)]l _XO, __&,where 6 represents oxygen vacancies. Based on the experimental evidence, it could be reasonable to assume that the increase in the mixed valence states produced by the formation of the solid solution enhances the catalytic activity for the heterogeneous decomposition of HO;. This is in agreement with the traditional explanation proposed by Latimer [42] for the catalytic decomposition of. hydrogen peroxide, that any .redox couple which falls in the potential range between the O,/HO; and HOT/OHredox couples can, in principle, catalyse the disproportionation reaction; then, if a catalyst simultaneously contains both strongly oxidizing and strongly reducing species, the coupled HO; oxidation and reduction might be fast. Since the first explanation for the decomposition of hydrogen peroxide on metallic surfaces given by Haber and Weiss 143,441,several mechanisms have been proposed for this reaction [26,35-38,45&l. AI1 of these mechanisms are based on a cyclic electron-transfer process, which is inititated by either the transfer of an electron from a reduced site on the surface of the catalyst to the peroxide to yield an OH radical or the transfer of an electron from the peroxide to an oxidized site
on the surface of the catalyst, to produce an HO; radical [26,45]. In this particular case, these reactions might be + HO&,,, + H,O * 20H- + OH;,,,, + and Ni( III) [ Fe( IV)
1
+HO,-+HO;+
Ni( II) [ Fe( III)
Ni( III) [ Fe( IV)
1
(14)
1
The relative rates of both reactions should depend on the redox potentials of the different redox couples. Several authors [37,47,48] observed that the dual valence oxides with standard reduction potentials considerably above the standard reduction potential for the redox couple O,/H,O, were effective catalysts for the peroxide decomposition, whereas those oxides with low standard reduction potentials were much less active. Moreover, high catalytic activity was also correlated with the presence of surface oxygen on manganese oxides [26,49] and the surface concentration of Cr6+ ions on chromium oxides [50]. Consequently, it seems reasonable to assume that in a transition metal oxide with two or more stable oxidation states, the most oxidized states (in this case Ni(II1) and Fe(W)) will be the active sites for the peroxide decomposition. Electrochemical reduction of oxygen Polarization curves obtained for 0, reduction on floating-gas fed electrodes in 5 M KOH are shown in Figs. 5 and 6. For the catalysts prepared at different temperatures (Fig. 5) and for LaFe,Ni, _XO, with different x values (Fig. 61, there is a clear correlation between the activity for 0, reduction (at j < 50 mA cm-*) and the activity for HO; decomposition (Figs. 3 and 4). This suggests that the activity for 0, reduction of these electrodes is due mainly to the ability of the catalyst to decompose hydrogen peroxide produced by the reduction of 0, on the behaviour is carbon support. At j > 50 mA cm-* a similar current-potential obtained for all the catalysts, probably owing to the fact that at these high current densities, diffusional processes start to be rate limiting. A correlation between the activity of the carbon-based oxygen cathode and the activity for peroxide.decomposition on perovskite-type oxides was also reported by Shimizu et al. [4]. In order to find some quantitative correlation between the activity for 0, reduction of the carbon-based cathodes and the catalytic activity for HO; decomposition on the catalyst, we will make the following assumptions. (i) The reduction of 0, on the high surface area carbon is fast. (ii) The heterogeneous decomposition of HO; on the catalyst is rate determining. (iii) The four-electron reduction