Catalytic Decomposition Kinetics of Aqueous Hydrogen Peroxideand Solid Magnesium Peroxide By Birnessite

A. M. Elprince* and W. H. Mohamed

ABSTRACTPeroxides are used as Degenerating agents. The decomposition of

H2O2 (aq) or MgO2(s) in the presence of synthetic birnessite [8-MnO2(s>]as a catalyst was observed by measuring the volume of O2(g) givenoff. The experimental data fit a first-order kinetic law and the mech-anism proposed by Habes and Weiss can explain the experimentalresults obtained for the decomposition of H2O2(aq). The applicabilityof the proposed mechanism involving H2O2(aq) was based on the fol-lowing: (i) the activation energy was relatively high (82 ± 3 kj mol~');(ii) the rate constant (k) was pH dependent, indicating that H+ andOH' ions were formed in the process; (iii) a linear relationship (andnot logarithmic) between k and the ionic strength indicated a reactionbetween an ion and a neutral molecule; and (iv) using birnessite withdifferent fractional coverages of Co indicated that the reaction washeterogeneously catalyzed by Mn"'/Mn* active centers. RegardingMgO2(s) decomposition, the experimental data obtained at pH 7.6followed a rate law derived from a shrinking-core model for fixed-size particles. The rate of the reaction was controlled by diffusionthrough the Mg(OH)2(s) product layer rather than by chemical re-action at the core surface. Compared with ionic peroxides, MgO2(s)could be a potential O2-generating agent for generating O2(g) duringa relatively long period of time.

THE O2 CONCENTRATION in soil solutions drops tonear zero within 24 h of submergence (Patrick

and Sturgies, 1955). This rapid depletion of O2(aq)inhibits plant growth (Rowell, 1981, p. 120) and ac-tivates anaerobes to reduce oxidized C, N, Mn, Fe,S, As, Cr, Cu, and Pd (Sposito, 1981). The traditionalway of increasing soil aeration is by lowering the watertable, using open or tile drains. A less common methodis by introducing an O2-generating agent into the soil.According to Westcott and Mikkelsen (1983), the ad-dition of CaO2(s) as a seed coating greatly enhanced

A.M. Elprince, Kuwait Institute for Scientific Research, P.O. Box24885, Safat 13109, Kuwait; and W.H. Mohamed, Dep. of Soilsand Agricultural Chemistry, Alexandria Univ., Alexandria, Egypt.Received 5 Aug. 1991. 'Corresponding author.

Published in Soil Sci. Soc. Am. J. 56:1784-1788 (1992).

the emergence of rice (Oryza sativa L.) seedlings fromflooded soils. The H2O2(aq) produced on hydrolysisof CaO2(s) decomposes, releasing O2(aq). Umeda etal. (1986, 1987) reported the development of differentcoating materials containing CaO2(s) for seeds to pro-mote germination and root growth.

A method proposed to eliminate unwanted odor insewage systems is the oxidation of sulfide in solutionof H2O2(aq) (Hoffman, 1977). According to Akira andMitsuzawa (1987), the addition of a mixture of CaO2(s)and CaCO3(s) was found more effective than the aer-ation method for the decomposition of organic sub-stances in wastewater by activated sludge.

The dismutation of H2O2(aq) to O2 and H2O is knownto be accelerated by some soil constituents. A largeportion of the H2O2(aq)-catalytic capacity in soil isnonezymatic (Johnson and Temple, 1964; El-Wakil,(1986); some researchers have suggested that most ofthe catalytic capacity is due to Mn compounds in soil;others ascribe it to Fe compounds and colloids (Sku-jins, 1967, p. 185). According to El-Wakil (1986),pretreatment of a soil (Typic Torrifluvent) by NH2OH-HC1, designed to selectively dissolve MnO2, com-pletely deactivated its catalytic capacity for H2O2(aq)decomposition. Birnessite [8-Mno2(s)] is the mostprobable inorganic catalyst in Torrifluvents (El-Wakil,1986). Birnessite is one of the most common Mn min-eral in terrestrial and aquatic environments; it occursin a disperse form and is nonstoichiometric (Oscarsonet al., 1983). When Co2+ is adsorbed on syntheticbirnessite, it undergoes a redox reaction with Mn3+

or Mn4+ (Traina and Doner, 1985). Murray and Dil-lard (1979), using XPS, presented the first directchemical evidence for the formation of Co3+ whenCo2+ is adsorbed by 5-Mno2(s). The XPS data indi-cated that only Mn4+ was present, although it is notpossible to rule out the presence of trace amounts ofMn2+ and Mn3+ (Murray and Dillard, 1979).Abbreviations: XPS, x-ray photoelectron spectroscopy; /, ionicstrength.

ELPRINCE & MOHAMED: CATALYTIC DECOMPOSITION KINETICS 1785

For inhibiting a vigorous reaction, CaO2(s) is usu-ally coated with a moisture-resistant material (Umedaet al., 1986, 1987; Tsukisaka et al., 1987). Never-theless, the reported rates of O2 generation are stillhigher than required in practice (Wesseling, 1974;Rowell, 1981, p. 120). According to Cotton andWilkinson (1972), MgO2(s) is intermediate in char-acter between the ionic [e.g., H2O2(aq) and CaO2(s)]and the covalent peroxides [e.g., Zno2(s)]. Therefore,MgO2(s) may release O2(g) less vigorously than theionic peroxides, and the reaction may continue for alonger period of time as desired in practice.

The decomposition of H2O2(aq) by metal ions ofvariable valency (Me2) was studied by Habes and Weissin 1934 (cited by Ahuja et al., 1988), who proposedthe following mechanism:

Reduction of H202H2O2(aq) + Me* -* OH- + Me2*1 + OH [la]

H202(aq)Oxidation of H20Z

+ Me2 + HOO [Ib]

The radicals then enter into a chain reaction withH2O2(aq) to form O2(g) and H2O(1) (Panchenkov andLebedev, 1976). On the other hand, no data are avail-able on the birnessite-catalyzed decomposition ofMg02(s).

Our objectives were to: (i) determine the rate ofH2O2(aq) decomposition on synthetic birnessite as in-fluenced by temperature, pH, ionic strength, and poi-soning by Co to establish the rate law and test theapplicability of the Habes and Weiss mechanism; and(ii) measure the rate of birnessite-catalyzed MgO2(s)decomposition as a function of temperature to estab-lish the rate law based on a shrinking-core model.

MATERIALS AND METHODSBirnessite Preparation

Birnessite [8-MnO2(s)] was prepared by adding 12 MHC1 to a boiling solution of KMnO4, as outlined by McKenzie(1971). The birnessite was washed repeatedly with distilledwater, treated with concentrated H2O2 to reduce trace amountsof MnO4~, and diluted to make a 10 gL~' suspension.

Cobalt-Birnessite Sorption IsothermBirnessite suspensions (10 gL -') were equilibrated at 26 °C

with CoCl2 with initial concentrations varying from 0.15 to2.25 mol kg-1 of birnessite at pH 7.6 (tris buffer of 10mmol/L-1) for 24 h. Cobalt in solution was measured usingatomic-absorption spectroscopy. The amount of sorbed Cowas determined in duplicate by difference between the totalinitial amount and the amount in solution at equilibrium.The equilibrated Co-birnessite suspensions were used insome of the kinetic runs.

Aqueous Hydrogen Peroxide Decomposition ExperimentsAnalytical-grade 8.8 M H2O2 was used to prepare a stan-

dard solution of 0.1 mol L-1. The solution was standard-ized by iodometric titration (Kolthoff and Sandell, 1948).An initial concentration of H2O2(aq) of 71 mmol L-1 wasselected for the kinetic runs at a desired pH (tris buffer of10 mmol L-1) and a desired ionic strength (NaCl) with 2.5mg birnessite (or Co-birnessite) in a total volume of 25cm3.

Solid Magnesium Peroxide Decomposition ExperimentsAnalytical-grade MgO2(s) was used. The particle-size

distribution of its powder was: 3.5, 4.3, 21.6, and 70.6%for the diameters 315 to 400, 160 to 315, 125 to 160, and< 125 |o,m, respectively. The kinetic runs were conductedwith 0.1000 g of MgO2(s) powder and 2.5 mg of birnessitein tris buffer (10 mmol L-1) at pH 7.6 with a total volumeof 25 cm3.

The volume of O2(g) generated from H2O2(aq) or MgO2(s)decomposition under controlled agitation and constant tem-perature, was determined using a gas burette. The temper-ature was controlled using a Julabo circulator (JubaboLabortechnik GMBH, Schwartzwald, Germany) mountedto a closed water bath.

RESULTS AND DISCUSSIONRate Law for Aqueous Hydrogen Peroxide Decomposition

A first-order kinetic law given as

In [I/a] = kt, [2]

fits well the birnessite-catalyzed H2O2(aq) decompo-sition data in this study. In Eq. [2], k is the rate con-stant and a is the unreacted fraction, i.e., a =[H2O2(aq)]/[H2O2(aq)]0, where [H2O2(aq)]0 is the in-itial concentration and t is reaction time. If K» is thetotal volume of O2(g) evolved during the entire reac-tion and V the volume evolved up to time t, then[H2O2(aq)]0 is proportional to K. and [H2O2] is pro-portional to (V* - V). Therefore [I/a] equals [VJV*- v\.

100

O 21.1 C24.6

t> 28.2» 31-9

35 6-••39.5

360 72O 1080t ,S

144O 1800

Fig. 1. First-order kinetic plot of the effect of temperature onH2O2(aq) decomposition on birnessite at pH 7.6; a is theunreacted fraction of H2O2.

1786 SOIL SCI. SOC. AM. J., VOL. 56, NOVEMBER-DECEMBER 1992

Figure 1 shows log [I/a] vs. t for the birnessite-catalyzed H2O2(aq) decomposition of pH 7.6 and tem-perature between 21.1 and 3.95 °C. The goodness offit is evident and confirmed by the high values of thesquared correlation coefficeint (r2), which varied from0.9963 to 0.99998. All the other kinetic runs have r2

values within this range.

Aqueous Hydrogen Peroxide Decomposition as Influencedby Temperature, pH, and Ionic Strength

Figure 2 shows the values of k as a function of (i)temperature at pH 7.6, (ii) pH at 30 °C, and (iii) ionicstrength at 30 °C and pH 7.6. As the temperature in-creased from 21.1 to 39.5 °C, the value of A: increasedfrom 1.235 x 10~3 to 9.13 x 1Q-3 s-1. As seenfrom Fig. 2, the experimental data are well describedby a linear relationship between In k and 1/T based onthe Arrhenius equation: k = A exp(-E/RT), where Ais the frequency factor, T is temperature, R is the gasconstant, and E the activation energy. The values ofE and A are 82 ± 3 kJ mol L-1 and 4.40 x 10" s-1,respectively. Since, diffusion-controlled reactions havelow activation energies (Lasaga and Kirkpatrick, 1983),the H2O2(aq) decomposition reaction is not controlledby diffusion but by a chemical reaction process.

As the pH increased from 6.6 to 8.0, the value ofk increased from 2.09 x 1Q-3 to 4.62 x 1Q-3 s-1 at30 °C. The experimental data are well described by alinear relationship between In k and pH based on theequation: k = k0 aH", where aH is the activity of H+

in the aqueous solution, n is the order of the reactionwith respect to H+, and k0 is the rate constant at aH= 1 (Fig. 2). A linear least-squares analysis yieldsvalues forn and k0 of -0.23 ± 0.04 and 5.4 x 10~5

s~*, respectively. The dependence of k on pH indi-

Table 1. Decomposition of H2O2 (aq) at 26 °C and pH 7.6 onCo-birnessite.

i ( m o l / O0-25 0.5

3.1 3.3 3.410007 T(K)

3-5

6.5 7.5PH

8 8.5

Fig. 2. The rate constant, k (Ms-1), for the birnessite-catalyzedH2O2(aq) decomposition reaction as a function of temperature(T), pH, and ionic strength (/); Ms = 106 s and k° is therate constant at / equals zero.

Ofcokx 103(s-')t

0.002.08

0.091.04

0.191.08

0.480.833

0.880.623

0.990.730

t & is the rate constant.

cates that H+ or OH~ ions are formed in the process,which lends further support to the Habes and Weissmechanism (Eq. [1]). Since the catalyst dissolves inacid medium, the low activity in acid media may bedue to a homogeneous rather than a heterogeneousreaction. In basic medium the rates are high and theentire reaction takes place heterogeneously. This is inagreement with Ahuja et al. (1988), who made a sim-ilar suggestion for H2O2(aq) decomposition on MnSsurfaces, and with Bartlett (1981), who stated that, inacid medium, oxidized Mn will oxidize H2O2(aq) and,in alkaline medium, Mn dioxides will catalyze its de-composition.

Figure 2 also shows the increase in the relative rateconstant (k/k°) for H2O2(aq) decomposition at pH 7.6and 30 °C as a function of the ionic strength (/), whereA:0 is the rate constant at / = 0. It can be shown, byemploying the transition state and Debye-Huckel the-ories for dilute solutions, that in the case of a reactionbetween charged particles A and B: In k/k° = 1.022ZAZB V?; and in case of a reaction between a neutralmolecule and a charged particle: k/k° = 1 + p I,where p is a constant (Panchenkov and Lebedev, 1976).As seen from Fig. 2, the experimental data are welldescribed by a linear relationship (and not logarith-mic) between k/k° and / for the solutions with valuesof / < 0.1 mol L"1. This finding indicates a rate-determining reaction between an ion and a neutralmolecule, which lends support to the Habes and Weissmechanism (Eq. [1]).

Fig. 3. (a)Sorption isotherm for Co on birnessite at 26 °C andpH 7.6, g and c are sorbed and solution concentrations atequilibrium, respectively; (b) effect of Co on the birnessite-catalyzed H2O2(aq) decomposition reaction at 26 °C and pH7.6, ftp and ku are rate constants for poisoned and unpoisonedbirnessite. The dashed curve is a step function approximationfor the solid curve.

ELPRINCE & MOHAMED: CATALYTIC DECOMPOSITION KINETICS 1787

Aqueous Hydrogen Peroxide Decompositionon Cobalt-Birnessites

Table 1 presents the values of k for the decompo-sition of H2O2(aq) at 26 °C and pH 7.6 on Co-bir-nessite with fractional coverage, cxo, ranging from 0to 0.99. The «Co value is computed from the sorptionisotherm (26 °C and pH 7.6) shown in Fig. 3a. Asshown in this figure, the isotherm is of the Langmuirtype (Elprince and Sposito, 1981) with maximum ca-pacity of 1.7 mol kg-1, which is slightly less than thevalue of 1.84 mol kg-1 reported by Traina and Doner(1985). As aco increases from 0 to 0.99, the value ofk decreases from 2.08 x 10~3 to 7.30 x 10~4 s-1,indicating a poisoning effect by Co.

Figure 3b shows kp/ku vs. aCo, where kp and ka arethe rate constants for the poisoned and unpoisonedbirnessite, respectively. As seen from Fig. 3b, thereis an initial dramatic deactivation followed by lesspronounced deactivation, then the catalytic activitybecomes independent of the sorbed Co. Note that,although the birnessite is completely saturated withCo, there is still some catalytic activity (Fig. 3b). Thisresidual activity is probably due either to the forma-tion of Co3+/Co2+ or Mn3+/Mn2+ active centers re-sulting from the surface interaction of Co2+ with Mn4+

and Mn3+. Based on thermodynamic considerations,the first pathway is dismissed. Thermodynamics im-ply that any redox couple at a potential (£°) greaterthan that of the redox couple O2(g)/H2O2(l) (E° =0.68 V) and less than the redox couple H2O2(aq)/H2O(l)(E° = 1.77 V) is a catalyst for the H2O2(aq) decom-position reaction. Thus Mn4VMn3+ [E° = 1.1V] andMn3+/Mn2+ (E° = 1.54 V) are catalysts for this re-action, but not Co3+/Co2+ (E° = 1.82). The work ofBurns (1976) and Hem (1978) provides further sup-port for the second pathway, namely the formation ofreduced active centers, for explaining the residual cat-alytic activity on complete saturation of the birnessitesurface with Co. Burns (1976) proposed a model inwhich Co2+ is oxidized to Co3+ by 8-MnO2 and theCo3* ion is stabilized by filling the vacancies in theedge-shared [MnO6] octahedral layers. Hem (1978)proposed a similar model based on the presence ofMn3+.

On approximating the Co-poisoning curve (solidcurve in Fig. 3b) by a step function (dashed curve inFig. 3b), the amount of Co2+ consumed in reducingthe oxidized active centers is estimated to be 0.17 molkg-1 (i.e., 0.1 times the maximum adsorption capac-ity of 1.7 mol kg-1). Thus, the catalytic activity ofthe unpoisoned birnessite is caused by an amount ofthe Mn4+/Mn3+ active centers = 0.085 mol kg-1 (i.e.,0.17 mol kg-1 divided by 2 mol per active center).Such small amounts of Mn3+ may have been beyondthe detection limit of the XPS work reported by Mur-ray and Dillard (1979). Furthermore, as seen fromFig. 3b, the reduced active centers are 35% as effec-tive as a catalyst as the oxidized active centers. Thesefindings lend further support to the Habes and Weissmechanism (Eq. [1]) by showing that the H2O2(aq)decomposition reaction is heterogeneously catalyzedby Mnz+1/Mnz active centers.

Rate Equation According to a Shrinking-Core Modelfor Solid Magnesium Peroxide Decomposition

Solid MgO2 probably decomposes according to thehydrolysis reaction:

Mg02(s) + 2H20(1) = Mg(OH)2(s) H2O2(aq)[3]

and the birnessite-catalyzed reaction:

H202(aq) = 2 H2O(1) + O2(g) [4]

Adding Eq. [3] and [4] yields the overall reaction:

2 Mg02(s) + 2 H20(l) = 2 Mg(OH)2(s) + O2(g)[5]

As reaction [5] occurs, a layer of product [Mg(OH)2(s)]probably builds up round the unreacted core of MgO2(s).The surface reaction then proceeds successively in-ward, constantly reducing the size of the core of un-reacted material. Two different cases may be consideredfor this model. The first assumes the continuous for-mation of solid product and inert material (ash) with-out flaking off, thus the particle size remains unchanged.In the second, the particle size changes as the reactionproceeds due to the flaking off of the solid. Since ourkinetic runs were conducted under alkaline conditions(pH 7.6), we proceed by analyzing only the first case,characterized by fixed-size particles.

For fixed-size particles, one of three process stepsmay control the overall reaction rate: diffusion throughthe liquid film, diffusion through the product layer,or chemical reaction. Since our reaction cell has beenunder an adequate degree of agitation, diffusion con-trol through the liquid film is dismissed. An analysisfor the case in which diffusion through the productash controls the overall rate has been outlined by Smith(1970). Under this condition (diffusion through prod-uct controlling) the rate equation is:

[302/3 - 2a] = 1 - (l/fx)f, [6]

where a is the unreacted fraction of MgO2(s) at timet and tx is the time for complete conversion.

An analysis for the case in which chemical reactioncontrols the overall rate has also been outlined bySmith (1970). Under this condition (chemical reactioncontrolling) the rate equation is:

a1'3 = 1 - (lltjt. [7]

The kinetic experiments with MgO2(s) are recordedat pH 7.6 and temperatures 50 and 60 °C. The reasonfor not using lower temperatures is that the reactionis too slow for the accurate determination of the vol-ume as a function of time with the apparatus used inthis study. Equations [6] and [7] are then fitted to theexperimental data in order to determine whether theMgO2(s) decomposition is diffusion through productor chemical reaction controlling.

Figure 4 is a plot of (3a2/3 — 2a) vs. t (see Eq.

1788 SOIL SCI. SOC. AM. J., VOL. 56, NOVEMBER-DECEMBER 1992

1-0

• 96

_.938s

•88

•84

.80

O f (U) : 3 <X-32 <X

i /3K*): <*

IO 2O O 4OO 60O

t, S800 1OOO

Fig. 4. Birnessite-catalyzed decomposition of MgO2(s) at pH7.6 and two temperatures; a is the unreacted fraction ofMg02(s).

[6]) and a1/3 vs. t (see Eq. [7]) for the isotherms at50 and 60 °C. As seen from Fig. 4, the diffusion-through-product Eq. [6] linearly fits the experimentaldata much better than the chemical-reaction-control-ling Eq. [7]. This conclusion is confirmed by calcu-lating r2. The values of r2 for linear-fitting Eq. [6] tothe data at 50 and 60 °C are 0.90 and 0.98, respec-tively. On the other hand, the values of r2 for linearfitting Eq. [7] to the data at 50 and 60 °C are signif-icantly lower and equal to 0.79 and 0.92, respectively.

The model parameter tx then is computed by least-squares fitting of Eq. [6] to the experimental data.The time for complete conversion (tx) is found to de-crease from 5.7 x 104 to 1.7 x 104 as the temper-ature increases from 50 to 60 °C. Employing theArrhenius equation: l/f* = A exp(-E/RT), the valuesof E and A are 108.2 kJ mol-1 and 5.5 x 1012 s-1,respectively. Subsequently, at 25 °C the time for com-plete conversion is 1.67 x 106 s (19.3 d) and theaverage rate of O2(g) generation is 0.13 cm3 kg"1 s~l.For soils that consume O2(g) at rates of 0.12 to 0.35cm3 m-3 s-1 (Wesseling, 1974; Rowell, 1981, p. 120)4615 to 13 461 kg of MgO2(s) would generate therequired O2(g) for the top 0.5-m layer of 1 ha of cropsoil and would last for 19.3 d. The decomposition ofMgO2(s) in soil is undoubtedly not as simple as im-plied by this computation. However, the results atleast suggest that MgO2(s) could be a potential 02-generating agent in soil for a long period of time rel-ative to ionic peroxides.