J . Phys. Chem. 1990, 94, 8339-8345
8339
Kinetics of Reductive Dissolution of Colloidal Manganese Dioxide Paul Mulvaney, Ron Cooper, Franz Grieser,* Department of Physical Chemistry, University of Melbourne. Parkville. 3052, Victoria, Australia
and Dan Meisel* Chemistry Division, Argonne National Laboratory, Argonne, Illinois 60439 (Received: April 2, 1990; In Final Form: June 12, 1990)
The temporal characteristics of electron transfer to and proton uptake by colloidal manganese oxides in acidic aqueous solutions have been examined in the time regime of milliseconds to tens of seconds. Time-resolved conductometric measurements coupled with electron-transfer-rate results indicate that electron deposition into the colloidal oxides and protonation occurs simultaneously; Le., there is no polarization of the colloid during the reduction process. Reduction of the manganese oxides leads to their dissolution with quantitative amounts of Mn(I1) produced in the bulk aqueous phase. A detailed examination of both thermal- and radiation-induced dissolution of the metal oxides indicates that Mn(II1) centers present in the colloid are continually depleted to produce Mn(l1). The driving force for this reaction is the chemical potential gradient between the colloid and the aqueous solution due to the homogeneous nature of mixed Mn(lV)/Mn(lll) oxides. Equilibrium between the solution and the colloid is established in both acidic and alkaline solution by either conproportionation or disproportionation of Mn(ll1) depending only on pH or pMn.
Introduction A number of mechanisms have been postulated to account for various aspects of manganese dioxide dissolution or reduction in aqueous solution. Coleman1 proposed the proton-electron insertion model in 1946, which has been very successful in explaining the lattice dilation of manganese dioxides during electrochemical reduction. Scott2 was able to extend this idea by considering the diffusion of protons through the lattice during reduction, and he quantitatively explained the recovery in electrode potential following discharge in alkaline solution. Later Kozawa and Powers,' Vetter and Jaeger," Vosburgh and co-workers,s Huber and Bell: and Tye and co-workers,' improved the original concept by considering heterogeneous and homogeneous lattice reduction, effects of manganese oxidation state, pH, and pMn (-log aMn2+)on the behavior of manganese dioxide electrodes. More recently, Ruetschi? Tye,%Vcand Pohl and Atlung9 have introduced statistical mechanical models for the Mn02-water interface. In contrast, marine chemistry studies have focused on modeling the reverse process-the precipitation of M n 0 2 from Mn(I1) solutions as a function of the solution conditions. The models invoked by Hemlo to explain the dependence of the final oxidation state of the precipitate on the solution conditions are similar in principle to the ideas of Vetter4 on the equilibrium of MnOz electrodes. Radiation-induced dissolution has been utilized to examine these same processes by microheterogeneous kinetics." Henglein and co-workers studied the mechanism of the reductive dissolution of ( I ) Coleman, J . J . Trans. Electrochem. Soc. 1946, 90, 545. (2) Scott, A. B. J . Electrochem. SOC.1960, 107, 94. (3) (a) Kozawa, A.; Powers, R. J. Electrochem. Soc. 1966,113,870. (b) Kozawa, A. J. Electrochem. Soc. 1959, 106, 79. (4) (a) Vetter, K. J.; Jaeger, N. Electrochim. Acta 1966, 1 1 , 419. (b) Vetter, K. J. J. Electrochem. Soc. 1963, 110, 597. (5) (a) Vosburgh, W. C.; Lou, P A . J. Electroehem. Soc. 1961, 108,485. (b) Vosburgh, W. C.; Mark, H. B. Ibid. 1961, 108, 615. (6) Huber,R.;Bell, G . S. J. Electrochem. SOC.1964, I l l , I . (7) (a) Caudle, J.; Summer, K. G.; Tye, F. C. J. Chem. Soc., Faraday Trans. I 1973,69,885. (b) Tye, F. C. Electrochim. Acta 1974, 21,415. (c) Tye, F. C. [bid 1985, 30, 17. (8) Ruetschi, P. J. Electrochem. SOC.1984, 131, 2737. (9) Pohl, J.; Atlung, S. Elecrrochim. Acta 1986, 31, 391. (IO) (a) Hem, J. D. In Advances in Chemistry Series; Kavanaugh, M. C., Leckie, J. O., Eds.; American Chemical Society: Washington, DC, 1980; No. 189, pp 45-72. (b) Hem, J. D. Chem. Ceol. 1978, 21, 199. ( 1 1 ) (a) Pick-Kaplan. M.; Rabani, J. J. Phys. Chem. 1976, 80, 17. (b) Lume-Pereira, C.; Baral, S.; Henglein, A.; Janata, E. J. Phys. Chem. 1985, 89, 5772. (c) Baral, S.; Lume-Pereira, C.; Janata, E.; Henglein, A. J . Phys. Chem. 89. 5779.
0022-3654/90/2094-8339%02.50/0
colloidal M n 0 2 using this approach. While the crystal structure or the oxidation state of their particles was undefined, the role of a Mn(ll1) species in the autocatalytic reduction of the oxide was emphasized.llb-c In a previous study,I2 the steady-state radiolytically induced dissolution of a number of manganese oxides was reported. Two important conclusions were drawn from those results. Firstly, it was shown that the stoichiometry of the dissolution process depends upon the composition of the oxide. Secondly, it was concluded that the catalytic decomposition of hydrogen peroxide by manganese oxides participates in the radiation-induced dissolution at longer times. In this report, pulse radiolysis is used to examine the kinetics of these processes at shorter times (milliseconds to tens of seconds).
Experimental Section y-MnOOH was synthesized by the method of Lux.', -pMn2O3 was synthesized by the thermal decomposition of P-MnO, at lo00 OC.I4 The BET areas of these preparations were 66 and 1 m2 g-', respectively. y-MnOOH readily formed stable suspensions following sonication while the y-Mn2O3 suspension settled within an hour. The former, however, rapidly disproportionated upon acidification. The preparation and characterization of the other oxide powders used in this study have been described previously.12 All other reagents were of the highest purity commercially available, and were used without further purification. Samples were sonicated by using a Sonifier B-30 Cell Disruptor ( 5 min, strength 2) immediately prior to irradiation. The pH of all suspensions was adjusted with perchloric acid to about pH 3.3. This is close to the point of zero charge (pzc) for all the oxides used. Samples were deaerated with either Ar or N 2 0 by using the syringe technique. Unless otherwise stated all solutions contained 0.2 M ethanol or propan-2-01 to convert the O H and H radicals produced by the radiolysis of water to the reducing CH,CHOH or (CH3)2COHradicals. Some settling of the suspensions was observed at long times after sonication (>1 h). This slow settling did not appear to affect the kinetic results outlined below, and it is believed that only the larger particles which have a negligible effect on the rates of reaction precipitated during this time. ~~~~~~~
~~~~
~~
~~~~~~~
~~
~
~
(12) Mulvaney, P.; Denison, L.; Grieser, F.; Cooper, R.; Sanders, J.; Meisel, D. J. Colloid Interface Sci. 1988, 121, 70. (13) Lux, H. In Handbook of Preparative Inorganic Chemistry; Brauer, G . , Ed.; Academic Press: New York, 1965; Vol. 2. ( 1 4) @-Mn02was obtained from the I.C. Sample Office, 1473 I Sprengel Ave., Cleveland, OH 44135.
D 1990 American Chemical Societv
8340 The Journal of Physical Chemistry, Vol. 94, No. 21, 1990
1.5
absence of any further added electrolyte. As can be seen in Figure 1 b, the pzc was 2.9 f 0.2. Average particle size of d = 150 A was determined by electron microscopy. The electron micrographs also showed considerable clustering of the primary particles into chains. While this clustering may have been due in part to the method of specimen preparation, light-scattering results consistently yielded considerably larger particle sizes Figure 1c, suggesting that some clustering of the primary particles also exists in solution. The aggregation number, Nag, of the manganese in a primary particle of the sol was calculated by using a molecular formula of M n 0 1 . 9 4 ~ . 1 2 0 0based . M , on the average oxidation state of the manganese and a density of p = 2.0 g cmd3. This yielded Na,s = 2.3 X lo4. Likewise the surface area of the sol was estimated to be S = 6/pd = 200 m2 g-I. The ANL Linac facility and the spectrophotometric and conductivity detection systems used in this study have been described elsewhere.'* The radicals produced by pulse irradiation of aqueous solution were all converted to the viologen radical (V-) by the following reaction sequence:
i\,
0.0 200
h
300
400 500 600 Wavelength (nm)
700
800
0
E
-0
-?
Mulvaney et al.
-1
v)
H20
E
3 -2 9
Ll
-4 2
6
4
8
1
0
1
ea(, H, OH, H 2 0 2 ,H30+, H2
(CH3)2C0
2
PH
30
2 20
3 €9
- -
+ (CH3)2CHOH H20(H2) + (CH3)ZCOH eaq- + N 2 0 + H 2 0 N2 + OH- + OH (CH3),COH + MV2+(ZV)
OH(H)
0
10
0 122
164
193
244
307
Diameter (nm)
Figure 1. (a) Absorption spectrum of 2 X IO4 M MnO, sol at pH 10.5. (b) Electrophoretic mobility of 2 X lo4 M MnO, sol measured by laser Doppler electrophoresis at 22 f 1 OC in the absence of added electrolyte. (c) Size distribution of 2 X lo4 M MnOz sol measured by light scattering using a Malvern Instruments Autosizer.
Transparent colloidal M n 0 2 was prepared by the reduction of permanganate with manganous ionI5 according to 2Mn04- + 3Mn2+ + 2 H 2 0
-
5MnO2 + 4H+
(1)
This reaction is very fast in alkaline solutions and produces negatively charged sols which are stable for up to 6 months. For all the experiments described in this report, the sol was prepared at 2 X lo-" M Mn02 as follows: KMn04 standardized by oxalate titration was added to NaOH to produce a 250-mL solution of 8.0 X 1 0-5 M Mn04- at pH 10.5 f 0.1. To the vigorously stirred permanganate solution, 400 pL of 0.075 M Mn2+,as either the perchlorate or sulfate, was then rapidly added by micropipet. The golden brown sol formed within several seconds. The spectrum is shown in Figure la. The sol was found to be poorly crystalline. Electron diffraction revealed lines at d (A)= 3.09, 2.30, and 1.83, which index best to K2MnsOl6. The oxidation state of the sol, defined as x in MnO,, was measured by oxalate titration according to Chapman et a1.I6 A value of x = 1.94 f 0.03 was obtained after triplicate analysis, which is slightly higher than expected from the structural formula. The electrophoretic mobility of the M n 0 2 sol was measured by laser Doppler electrophoresis17in the (15) Stumm, W.; Morgan, J. J. J . Colloid Sci. 1964, 19, 347. (16) Freeman, D. S.; Chapman, W. G. Analyst 1971, 96, 865.
(2) (3) (4)
+ MV+(ZV-) + H+ (5)
Ethanol and propan-2-01 were used interchangeably as the OH radical and H atom scavengers in these experiments. Both methylviologen ( MV2+) and the zwitterionic viologen (ZV) 1,l'bis(sulfonatopropyl)-4,4'-bipyridinium were used as electron donors to the particle. Irradiation of a dilute suspension containing alcohol and viologen generates equal numbers of protons and viologen radicals, and this provides a convenient method for determining the ratio of reduction equivalents to protons consumed at any time following the electron transfer from V- to the M n 0 2 colloid. This fact was used to help interpret the conductivity experiments. The actual rate of electron transfer from V- to the particle was also measured directly, by monitoring the absorption band19 of V- at 600 nm (em = 1.28 X lo4 M-' cm-'). Changes in the conductivity of the suspension were followed for up to 100 s after the electron pulse. At longer times, temperature fluctuations caused conductivity drifts which made the results less reliable. However, as will be seen below all the transfer processes were complete within 100 S.
Blank solutions were also irradiated to check for possible side reactions. These were prepared in an identical manner to the colloidal solutions, including sonication, except that the colloid was not present. In the absence of manganese oxide, there was no decay of the conductivity or absorption signals on the time scale of the experiments. This was verified for all the systems studied. In N20-saturated solution, it was found both spectrophotometrically and conductometrically that the yield of viologen radicals was G(V-) = 6.8 radicals per 100 eV. It was also confirmed that the rate of formation of V- was not altered by the presence of the sol particles or the colloids in the suspensions. Occasionally a small, fast decay was observed in the conductivity signal due to submicromolar quantities of oxygen which reacted according to eqs 6-8. A second pulse delivered within few a milliseconds after the previous one to the same solution showed that it was completely removed. Trace amounts of oxygen are rapidly depleted in a cyclic fashion (reactions 6-8), forming H202and consuming two radicals
-v + + +
v- + 0 2 02-
HO2
H+
H02
+
0;
(6)
HO2
(7)
H202
+0 2
(8)
(17) Hayes, D. Ph.D. Thesis, University of Melbourne, 1987. (18) Meisel, D.; Mulac, W.; Matheson, M. J . Phys. Chem. 1981,85, 179. (19) Willner, I.; Yang, J.-M.; Laane, C.; Otvos, J. W.; Calvin, M. J. Phys. Chem. 1981, 85, 3277.
The Journal of Physical Chemistry, Vol. 94, No. 21, 1990 8341
Reductive Dissolution of Colloidal M n 0 2
4.00
1 L
,
c-c,
..--CI :
.
t -1.00
0.0
t,
s
Figure 2. Time profile of the conductivity changes following the pulse irradiation of a solution containing 0.27 g L-’y-MnO, at pH 3.3 in the presence of 0.2 M ethanol and IO-3 M ZV and saturated with N 2 0 .
per oxygen molecule. In the acidic pH range a reaction of HO, radicals with the particles is much slower than its disproportionation. However, the effect of this sequence of reactions is very different if enough time elapses between the two pulses to allow a reaction between H 2 0 2and the oxide, as discussed below.
Results ( i ) Stoichiometry. A typical conductivity decay curve is shown in Figure 2 for a y-Mn0, suspension. Three, temporally well separated processes could be observed, occurring on a time scale ranging from milliseconds to tens of seconds. Similar features were observed for all the oxides examined regardless of their crystal structure and composition. The amplitude of the conductivity signal immediately after the pulse corresponded to the yield of 6.8 H+/100 eV expected from reactions 2-5. The conductivity signal decayed to a level lower than the prepulse level, indicating that more than one proton was consumed per electron transferred to the M n 0 2 particle. The ratio of the total change in conductivity to the initial signal at the end of the conductivity decay (C& amplitude (Ci) was used to determine the stoichiometry of protons consumed per electron transferred, as shown in Figure 2. For all the samples with compositions close to the stoichiometric value, i.e., the a-,/3-, 6-, and y-Mn02 samples, which range in oxidation state from 1.86 x < 2.00, this ratio was found to be 2.0 f 0.2. This agrees with the value expected from the overall dissolution, reaction 9. MnO,
+ 2xH+ + 2(x - 1)e-
-
Mn2+ + xHzO
H+/e- = (Ci - Cf)/Ci = x / ( x - 1)
(9) (10)
For the MnS08suspension, the oxidation state is x = I .6 and the proton to electron ratio was found to be 2.7, as compared to the theoretical value of 2.67. This value is sufficiently high to distinguish it from the other manganese oxides. The agreement between the theoretical and observed stoichiometries, in particular the high value of H+/e- for MnSOa,verifies the authenticity of the oxidation state, x . If the formula MnSOa merely reflected a mixture of Mn(1V) and Mn(I1) oxidation states, the H+/e- ratio would be at 2.0 for this oxide as well. This conclusion is in agreement with our earlier results.I2 The contribution from the catalytic decomposition of H 2 0 2by the oxide (reaction 11) to the H+/e- ratio is expected to be minimal due to the relatively low yield
+
(x - 1 ) H 2 0 2 MnO,
+ 2H+
-
Mn2+ + 2xH,O
+ (x - 1)02(1 1)
of hydrogen peroxide (G(H202)= 0.65 molecules per 100 eV). For x = 2 the H+/e- ratio will increase to 2.1 and for x = 3/2 it will increase to 3.2. However, if viologen radicals are still present
0.40
sec Figure 3. Effect of repetitive pulsing on the decay of the conductivity signal under identical conditions to Figure 2. (a), (b), and (c) are the first, second, and fifth pulses, respectively. Dose in each pulse was 0.465 t,
krad. in the solution when reaction 1 1 occurs, they will efficiently react with the oxygen generated in reaction 1 1 and a chain reaction involving reactions 6-8 and 11 will set in. This chain reaction will maintain the same stoichiometry discussed above but will of course change the reaction mechanism. It is therefore important to establish the time scale for reaction 11 at low pH’s. We address this point in the following section. (ii) Fast Electron- and Proton-Transfer Processes. For all the oxides studied, the electron transfer, as monitored by the decay of the viologen radical absorption, proceeded by two temporally well separated processes. For convenience, these are labeled the “fast” and “slow” steps. Both followed simple, first-order kinetics. Initially there was no fast decay in well-deaerated systems. However, viologen radicals produced on consecutive pulsing of MnOz suspensions decayed more rapidly than those in fresh suspensions. While the percentage of V- that disappeared in the fast step increased with increasing preirradiation of the sample, the rate of the slow decay did not change. It was therefore concluded that a product was formed during the reaction that reacted faster with the viologen radical than the original colloid did. The behavior described above was common to all the samples studied regardless of crystal structure and oxidation state. A similar autocatalytic effect has been reported by Henglein and co-workers.I I b v c The rate of proton uptake in the fast step as determined conductometrically, was the same as the rate of electron transfer determined spectrophotometrically. In Figure 3, the effect of repetitive pulsing on the conductivity decay is shown. Due to experimental difficulties, the percentage of electrons and protons being transferred could not be measured at identical doses. The fact that the conductivity signal never dropped below the prepulse level on this time scale suggests that no more than one proton per electron was transferred in the fast step. Even after repetitive prepulsing of the sample to ensure that the radical decayed completely by the fast process, the conductivity leveled off at the prepulse level. If more than one proton had been consumed, the signal would have attained negative values at this time. Nevertheless, regardless of the fraction of the viologen which decayed by the fast process, the limiting conductivity at long times always reached the value predicted by the stoichiometry of reaction 9. By varying the time delay between pulses, it was possible to estimate the time required to produce the intermediate species at the oxide surface using the spectrophotometric detection method. Thus, when the solution was given two 4-11s pulses separated by only 1/60 s (after some prepulsing), the absorption signal from the first pulse decayed rapidly to zero. The signal from the second pulse, however, did not decay at all in the fast time regime. Conversely, when the two pulses were delivered 15 s apart, the signal in both pulses decayed completely by the fast process to the prepulse level. The results in Figure 4 were obtained by adjusting the time interval between the two pulses for the MnO,
8342
Mulvaney et al.
The Journal of Physical Chemistry, Vol. 94, No. 21, 1990
0
10
5
(V-1~10M ~
Figure 6. Dependence of the observed rate of fast proton uptake by 1.6 X M M n 0 2 on the dose, at pH 3.76. Solution composition as per Figure 4. 1
It
10
0
30
20 Time (s)
40
Figure 4. Percent decay of the signal after the second pulse by the fast process as a function of the time interval between pulses, monitored spectrophotometrically at 600 nm. Solutions contained 1.6 X IO4 M MnO,, 0. I M propan-2-01,2 X IO4 M ZV and were N 2 0 saturated. The pH was adjusted with HCIO4 and N a O H . 1'0E'05
I
1.OE+04
T
I
I
0
I
I
4
1.OEtO2 1.OE-07
I
1 .OE-05
15
20
25
Figure 7. Changes in conductivity following the pulsed irradiation of an MnO, sol ( I .6 X IO4 M) at pH 3.6 in the presence of 0.1 M tert-butyl alcohol and saturated with N 2 0 . Dose 1950 rad. Solid line is fit to first-order kinetics yielding 0.27 s-I for the observed rate constant of peroxide oxidation by the sol.
p"
1.OE-06
10
Time (s)
/
/ .
i
5
1.OE-04
[vrl, M
Figure 5. Dependence of the observed first-order decay of the viologen radical in the fast step on the initial viologen radical Concentration. Solid circles: 0.27 g L-l r-MnO2 monitored at 600 nm; solid squares, a-MnO, monitored conductometrically; open circles, y - M n 0 2 monitored conductometrically. Solution composition as given in Figure 2.
sol at pH 3.6 f 0.1 and 10.0 f 0.1. Assuming that V- reacts immediately with the intermediate, the half-life for its formation may be roughly estimated as I O and 1 s at pH 3.6 and 10, respectively. The half-life for the recovery of the fast process in a - M n 0 2suspensions was close to 15 s at 0.26 g L-' and pH 3.68. This is much slower than the complete decay of viologen radicals, i.e., the generation of the intermediate occurs on a slower time scale than the decay of the radical. When the initial viologen radical concentration exceeds the concentration of the intermediate, the observed rate constant for the decay by the fast process should depend linearly on the viologen radical concentration, Le., on the dose in the pulse. From the dependence of the observed rate constant (Figure 5) on the dose, the second-order rate constant for the fast process was found to be (2.2 f 0.3) X 1 O8 M-l s-!. A similar rate constant was obtained for all the different suspensions investigated. On the other hand, when the decay of the viologen was entirely by the fast process, the observed rate was (2.9 f 0.5) X IO3 s-' at pH 3.5 and 1.6 X IOy4 M MnO,. From the aggregation number of 2.3 X IO4, a second-order rate constant greater than the diffusion-controlled limit is inferred for a reaction with the particles. This strongly supports the contention that the fast process is due to a solution species rather than a surface intermediate. Changing the sol concentration did not alter the rate of viologen radical decay when all the decay occurred in the fast step, indicating t h a t the fast
process is controlled by a product of the preirradiation. The rate of the proton uptake during the fast electron transfer mirrored the rate of viologen radical decay in all of these observations, as can be seen in Figure 6 for a 1.6 X 1 0-4 M MnO, sol at pH 3.6. The derived second-order rate constant was equal to the rate of electron transfer within experimental error. The major ion produced by radiolytic reduction of M n 0 2 in acidic media is Mn2+. However, the addition of Mn2+ up to lo4 M to y-Mn0, (0.26 g L-I) or up to M (higher concentrations induce coagulation) to the transparent sol at pH 3.3 did not affect the viologen radical decay. Hence it was concluded that the product affecting the radical decay was not due to the Mn(I1) produced or to a conproportionation reaction between Mn( 11) in solution and Mn(1V) on the particle at this acidic pH. To explain the nature of the fast process we return to the effects of radiolytic H202. The catalytic effect of prepulsing on the fast viologen decay can be explained in terms of the oxygen produced by oxidation of radiolytically produced hydrogen peroxide. This mechanism was previously inferredi2on the basis of the observed C(Mn*+) values in acid solution. To confirm the participation of H 2 0 , we measured the rate of oxide dissolution by H 2 0 2in the presence of N 2 0 and 0.1 M tert-butyl alcohol (in the absence of viologen). Under these conditions all radicals disappear by recombination since the @ hydroxy radicals from rerr-butyl alcohol (reactions 3, and I2 and 13) cannot reduce the oxide. The linac (CH3)jCOH + O H (H) (CH,)2C(OH)CH, + H2O (H2) +
2(CH,),C(OH)CH2
-
(12)
[(CH,),C(OH)CH,I,
( 1 3)
pulse, then, serves as a fast hydrogen peroxide source. The kinetics of peroxide oxidation were monitored by the change in conductivity due to reaction 1 I . A first-order fit to a typical decay curve for the MnO, sol is shown in Figure 7, and yielded a first-order rate
The Journal of Physical Chemistry, Vol. 94, No. 21, 1990 8343
Reductive Dissolution of Colloidal MnO, constant of 0.27 s-I at 1.6 X IO4 M MnO, at pH 3.6. From the specific area of the sol, the surface rate constant is calculated to cm s-l, which is much higher than the values obbe 9.2 X tained by Kanungo et al.” by volumetric analysis of the rate of 0, production, for various manganese dioxide suspensions. The greater values obtained by conductometric detection may be due to the greater catalytic properties of the sol and to the use of the geometric area rather than the BET area for the calculation of the specific rate. Indeed, Kanungo et aI.,O found no correlation between the BET area and the catalytic activity of their samples. The time scale for peroxide oxidation measured conductometrically is similar to that required for the appearance of the intermediate determined by the double-pulse experiments (Figure 4). Each peroxide molecule that is oxidized by the MnO, generates one oxygen molecule which reacts with the viologen molecule during the fast process of the next pulse according to reaction 6. When the geometric series of reactions 6-8 is taken into account (without further reaction of H 2 0 2with the oxide since this reaction occurs on a much slower time scale) at most a C(V-) = 2G(H202) = 1.3 may be expected in the fast time regime of the second pulse. However, H 2 0 2is accumulated by repetitive pulsing and eventually all of the viologen radicals will react through the chain of reactions 6-8 and 1 I . The reduction of two Mn(ll1) oxides suspensions was also briefly examined to determine whether the kinetics of Mn(II1) reduction could be observed directly in acid solution. y-MnOOH readily formed stable suspensions following sonication. However, it was subsequently found that it rapidly disproportionated prior to irradiation, generating MnO,. A y-Mn203suspension was then synthesized as described in the Experimental Section. However, the viologen radicals in fresh y-MnzO3 suspensions never decayed completely on the first time scale, and qualitatively these results support the earlier finding that Mn(lI1) is not stable on the surface of acid MnO, suspensions. If small amounts of Mn(II1) at the surface were responsible for the fast process, then a pure Mn(II1) oxide would be expected to show much greater reactivity toward the viologen radicals. To summarize this section, all our experimental observations on the “fast” time scale (a few milliseconds) may be explained by the release of oxygen to the solution via the oxidation of hydrogen peroxide. None of the kinetic data need be explained by invoking an intermediate oxidation state in the solid phase. (iii) The Slow Electron- and Proton- Transfer Steps. Following the fast electron transfer of one proton and electron to oxygen, the remaining viologen radicals reacted much more slowly. The slow electron transfer was found to be first order for all the systems studied. Furthermore, the decay remained first order, even after repetitive pulsing, and the rate was independent of dose. This step consumed all the radicals in the absence of the fast process. The observed rate constants for this slow electron transfer were 9 f 3 s-I at pH 3.5, and 9 f 2 s-l at high pH using large 20-11s a second-order rate constant of 1.3 X IO9 M-’ pulses. Using Nagg, s-l is obtained, which is close to the diffusion-controlled limit. Likewise, the conductivity signal decayed at the same rate as the electron transfer. The independence of the observed rate constant on dose for the sol, followed conductometrically (Figure 8), and the linear dependence on oxide concentration indicate that the slow step is the reduction of surface Mn(IV) sites, whose concentration exceeds the concentration of the radicals. Since both the electron transfer and proton consumption proceed at the same rate, reaction 14 may describe this process.
H+ + V-
-
+ [Mn02]surf
[MnOOH],,rf. + V
(14)
The question as to whether or not the Mn(II1) center, produced at the surface via reaction 14, rapidly reacts with another Vradical (reaction IS) may now be raised. Since the Mn(lI1)-rich
H+ + V-
+ [MnOOH],,,f
-
[Mn(OH),lsurf+ V ( I S )
colloids, mentioned in the previous section, showed no accelerated (20) Kanungo, S.;Sant, B.; Parida, K . Electrochim. Acta 1981.26, 1157. (21) Sawyer, D. T.; Valentine, J. S.Acc. Chem. Res. 1981, 14, 393.
80
5
0
10
[V-1 x 10
M
Figure 8. Dependence of the slow proton transfer rate to 1.6 X lo-” M MnO, on dose at pH 3.76. Solution composition as in Figure 4.
reaction rate with V-, we believe that reaction 14 dominates. (iu) The Final Proton- Transfer Srep. The electron-transfer reactions described above involve the simultaneous uptake of protons, and consequently when the viologen radicals had been consumed, the conductivity achieved the prepulse level. Typically this was reached about 1 s after the pulse at pH 3.3 and 0.26 g L-’of oxide. The final process was a slow proton uptake which was in the 1- 100-s time regime as can be seen in Figure 2. This process obeyed neither first nor second-order kinetics. The conductivity signal reached the value predicted by eq IO at the end of this step. We assign these conductivity changes to the disproportionation of Mn(II1) centers 2[Mn00H]s,,f,
+ 2H+
-
+
Mn2+aq 2 H 2 0
+ MnO,
(16)
where the reaction is considered to be confined to two surface Mn(II1) centers. The first half-life for the y-MnO, (0.27 g L-I) suspensions was 17 s at pH 3.3. This corresponds to a surface cm s-I for reaction 16, based on the rate constant of 1.7 X BET area. For a 1.6 X IO4 M MnO, sol at pH 3.76, the slowest proton consumption step occurred with a half-life of 3.5 s, leading cm s-l, based to a disproportionation rate constant of 7.2 X on the geometric surface area. It is interesting to note that the observed rate constant of H 2 0 2oxidation, determined in section (ii) above, is very similar to that of the final proton consumption step (0.27 s-l for the former and 0.2 s-I for the latter in the case of the MnO, sol). This suggests that the rate-determining step in the case of H 2 0 2oxidation is the Mn(II1) disproportionation rate, Le., replenishing the particle surface with fresh Mn(IV) sites. (V) Thermal Dissolution of MnO,. In our previous report,I2 it was noted that Mn(I1) was found in acidic solution prior to radiolysis. This has been observed in many previous studies. Thermal dissolution of all the manganese dioxides occurred when the dry powder was added to an acidic solution, even when no reduction was apparently possible (Le., when the acid was perchloric or sulfuric). Stumm and MorganIs and later Murray2, postulated that, since the MnO, becomes unstable with respect to water oxidation below about pH 3, the Mn2+is associated with slow water oxidation by reaction 17 2MnO2
+ 4H+
-
2Mn2+ + 2 H 2 0
+ 0,
(17)
In Figure 9, the thermal dissolution of @-Mn02at pH 3.0 is shown. The release of Mn(I1) into the solution appears to be linear with time over the short time scales relevant for the radiolysis experiments. The oxide was added to an alkaline solution under N2 to prevent dissolution. The reaction was then initiated by addition of an aliquot of N2-saturated H2S04. Complexation of both Mn(I1) and Mn(II1) with sulfate is known to be low. The rates of both 0, formation and Mn(I1) productioc were monitored. Almost negligible amounts of 0, were produced (3% of the amount demanded by reaction 17). The small amount of O2may be due to slow air leakage. The origin of the dissolved Mn(I1) is therefore associated with surface exchange reactions with solution acid, or to disproportionation of Mn(I1I) at the surface.
Discussion It is clear from the steady-state results in our previous study and from the kinetic analysis presented so far that hydrogen (22) Murray, J. W . J . Colloid Interface Sci 1974, 46, 357
8344
The Journal of Physical Chemistry, Vol. 94, No. 21, I990 3.0
Mulvaney et al.
I
*-
4
0
2000
6000
4000
Time (s)
Figure 9. Thermal dissolution of 1.0 g L-' P-Mn02 at pH 3.0 in N2saturated solution. Mn(I1) measured by atomic absorption, and O2 with an oxygen probe.
peroxide interferes with the electron-transfer chain during the radiation-induced dissolution of manganese oxides. However, our interest is focused on the slow proton-transfer steps, since these include the genuine manganese dioxide reductive dissolution processes. It was postulated in section (iii) that the slow electron transfer was the reduction of Mn(1V) sites. The subsequent steps therefore involve the conversion of Mn(II1) centers into Mn2+(,p). Since this conversion process also occurs during the thermal dissolution of the oxide, we first discuss the mechanism of the thermal dissolution prior to addressing the question of the fate of viologengenerated Mn(ll1) centers. The thermal dissolution, shown in section (v), indicates that the oxide has not achieved thermodynamic equilibrium during the radiolysis experiments. To understand the thermal dissolution, it is useful to relate x , the oxidation state of the oxide, to the solution pH and pMn at equilibrium. Vetter4 derived the following expression for the dependence of x on the solution composition RT In [H+] F
y,
2F
- 1 dAGfo(x)
In [Mn2+] = 2F
AGfo(x) 2F (18)
--
dx
where AGfo(x) is the free energy of formation of the oxide with oxidation state x . In dilute suspensions, where equilibration does not alter the solution composition significantly, the left-hand side of eq 18 is constant, so the equilibrium oxide composition is fixed by the solution conditions only. Once AGfo(x) is known analytically as a function of x , it is possible to determine the equilibrium situation for any oxide in a solution of known pMn and pH. Gorichev and c o - ~ o r k e r have s ~ ~ derived an empirical equation from which the values of AGfo(x) necessary to obtain numerical oxide potentials can be interpolated. Using their values, the electrode potential of an anhydrous oxide of formula MnO, on the hydrogen scale is then23b-c
pMn
1.0
0
2
4
6
8
2
4
10
6
12
X
i
14
PH Figure 10. Equilibrium oxidation state of manganese oxides as a function of solution pH at (a) pMn 2, (b) pMn 4, (c) pMn 6, and (d) pMn 8. Curves calculated from eq 21.
conproportionate with Mn(1V) in the oxide to create Mn(II1) centers. In Figure 10 a number of isotherms are presented, calculated from eq 2 1. As the pH decreases and pMn increases, the equilibrium oxidation state approaches 2.0. It is clear that, in acidic solutions containing little manganous ion, the bulk oxide tends to a state that includes insignificant Mn(II1). Certainly the surface layers that are most likely to be in equilibrium with the solution will contain very little lattice Mn(II1). However, the time scale for equilibration is determined by the slow solid-state diffusion; the rate of Mn(lI1) diffusion out of the oxide24 have revealed very slow diffusion coefficients (D -IO-" cm2 s-]). Thus, as experimentally observed and as expected from the thermodynamic considerations, the reductive dissolution of manganese oxides occurs while the much slower thermal dissolution is taking place. We now turn to the mechanism of the reductive dissolution. We have already concluded that the first step of electron-transfer results in the formation of a Mn(II1) state at the surface (reaction 14) followed by disproportionation and release of Mn2+ to the solution (reaction 16). The rate-determining step in reaction 16 may be either the disproportionation step (reaction 22) or the desorption of surface Mn( 11) hydroxides (reaction 23). Since 2[MnOOHIsurr
-
[MnOzlsurf + [Mn(OH)2lsu,f
(22)
+
(23)
[Mn(OH)2]surf 2H+
-
Mn2+
+ 2H20
Equation 21 determines whether Mn(II1) sites are stable with respect to disproportionation, or whether Mn(I1) in solution will
the results discussed in section (iv) demonstrate that Mn(II1) disproportionation at the surface is a facile process, it seems likely that the slow, rate-determining step is reaction 23. Whether the disproportionation step occurs between two newly formed adjacent Mn(l1I) centers or involves a preexisting center cannot be decided upon from the kinetic results alone. However, the stoichiometry discussed in section (i) clearly indicates that the preexisting Mn(II1) sites do participate in this process. Finally, we may indicate some implications of our discussion to the reduction process in alkaline pH's. As seen in Figure 10, the value of the equilibrium manganese oxidation state is much lower in alkaline solutions, even at high pMn. The conproportionation of Mn(I1) and Mn(IV) sites to form Mn(II1) becomes increasingly favorable at higher pH's. Adsorption of Mn(I1) leads to the accumulation of Mn(II1) at the surface, which may slowly diffuse into the solid, leading to further adsorption. The high exchange capacities of M n 0 2 for Mn(I1) compared with other divalent transition metal ions25,26 can thus be rationalized. These Mn(ll1) sites may be expected to live longer at the particle surface and thus lead to the autocatalytic effect observed by Lume-Pereira et al." b*c However, the Mn02-catalyzed decomposition of peroxide is also more efficient in alkaline solution, as reported by Kanungo et a1.20and as seen in Figure 4a,b.
(23) (a) Gorichev, I. G.; Ashkharua, F. G. Russ. J . Phys. Chem. 1977,5/, 524. (b) Gorichev, 1. G.; Shevelev, N. P.; Ashkharua, F. G. Russ. J . Phys. Chem. 1978,52, 2408. (c) Gorichev, 1. G.; Kipriyanov, N . A. Russ. Chem. Rev. 1984, 53. 1790.
(24) Mulvaney, P. Ph.D. Thesis, University of Melbourne, 1988. ( 2 5 ) McKenzie, R. M. Geodernn 1971, 8, 29. (26) Posselt, H . S.;Anderson, F. J.; Weber, W. J. Enuiron. Sci. Technol. 1968, 2, 1087.
E = EoMn2+/Mn
+
0.687(x2 - 0.21 Ix) - 0.0592pHx x-l
+ 0.0296pMn (19)
The equilibrium composition of the colloidal oxide is defined by
[E] ax
=o
pH.pMn
Differentiation and rearrangement yields 0.0296pMn - 0.0592pH = 0.0687(x2 - 2x
+ 0.21 I )
(21)
J . Phys. Chem. 1990,94, 8345-8350 Conclusions The thermal- and radiation-induced dissolution processes can now be unified. In acid solution which is initially devoid of Mn(II), there is a fast dissociation of surface Mn(1I) and a slower, solid-state diffusion-controlled disproportionation. Mn(lI1) is unstable under these conditions. Mn(II1) sites are continually removed by disproportionation, regardless of their source; Le., both radiation-induced Mn(II1) centers and those present within the lattice undergo continual destruction to release Mn(I1). The driving force for this reaction is the chemical potential (concentration) gradient, due to the homogeneous nature of mixed Mn(IV)/Mn(II) oxides. Equilibrium is established in both acidic and alkaline solution by either conproportionation or disproportionation depending only upon pH and pMn. The steady-state yields of Mn(l1) from our earlier study, and the limiting conductivities from the pulse radiolysis experiments demonstrate that
8345
electrons transferred to particulate manganese oxides are localized quantitatively at the surface, even in the absence of complexing agents. As observed for the case of iron(II1) oxides,*’ fixation of the charge at the surface is facilitated by protonation. Acknowledgment. Work at Argonne National Laboratory is performed under the auspices of the Office of Basic Energy Sciences, Division of Chemical Science, US-DOE, under contract no. W-31-109-ENG-38. Support for this work by grants from the Australian Institute of Nuclear Science and Engineering is gratefully acknowledged. P.M. acknowledges the receipt of a Commonwealth Postgraduate Research Award. Registry No. ZV-, 77951-50-9;MV’, 25239-55-8;MnO,, 1313-13-9. (27) Mulvaney, P.; Cooper, R.; Grieser, F.; Meisel, D. Langmuir 1988, 4, 1206.
A Heterobinuclear Cation Containing Two Eiectroactive Centers That “Diffuse”through Nafion Coatings by Different Mechanisms Ching-Fong Shu and Fred C. Anson* Arthur Amos Noyes Laboratories, Division of Chemistry and Chemical Engineering.t California Institute of Technology, Pasadena, California 91 125 (Received: April 9, 1990)
A heterobinuclear complex, [4-(3-ferrocenylpropyl)-4’-methyl-2,2’-bipyridine] bis(2,2’-bipyridine)osmium dichloride, was synthesized, and the diffusion coefficients of the two metal centers were evaluated from electrochemical measurements. In homogeneous solutions the two centers exhibited identical diffusion coefficients, as expected. However, when incorporated in Nafion coatings, the two diffusion coefficients were far from equal. That measured by oxidation of the Os(I1) center was larger and increased nonlinearly with the concentration of the cation in the Nafion, while that measured by reduction of the ferricenium center was much smaller and exhibited a concentration dependence of the opposite sign. Possible origins of the disparate diffusion coefficients and their concentration dependences are suggested. The relative importance of electron self-exchange, molecular diffusion, and strong ion pairing of the cation with the fixed sulfonate groups in the Nafion are discussed.
The C ~ ( b p y ) , ~(bpy + = 2,2’-bipyridine) cation was the first example of a simple, electroactive species that, upon incorporation into a polyelectrolyte coating, exhibited an apparent diffusion coefficient that was much smaller when measured by electrochemical oxidation than when it was measured by electrochemical oxidation than when it was measured by electrochemical reduction of the complex.’ The difference provided clear evidence of the ability of electron self-exchange to enhance apparent diffusional rates of the components of redox couples that undergo high rates of self-exchange.’ Related experiments have also been conducted with a few molecules containing two electroactive centers that exhibit disparate diffusion coefficients as measured by electrochemical techniques when the molecules are incorporated in polymer coatings on ele~trodes.~JThis report deals with another such molecule prepared by linking together 0s1Ii/I1and ferricenium/ferrocene electroactive centers in the same molecule. The apparent diffusion coefficients of these two centers are significantly different when they are present as mononuclear complexes in Nafion coatings4 Our intent was to determine if this difference would persist when the two centers were linked together or if the heterobinuclear cation would diffuse like the single molecule it is. What we observed was a persistent difference in the apparent diffusional rates of the linked centers and a pattern of concentration-dependent apparent diffusion coefficients that shed light on the mechanisms of charge transport. ‘Contribution No. 8123
Experimental Section Materials. Tetrahydrofuran (THF) was dried by refluxing with and distilling from Na/K alloy and benzophenone. (Ferrocenylmethy1)trimethylammonium hexafluorophosphate, [CpFeCpCH2N(CH3)3]PF6,was prepared by metathesis of the corresponding iodide salt5 with NH4PF6. The crude product was recrystallized from water. 2-Ferrocenylethano16and Os(bpy),CIJ were synthesized according to the cited references. Nafion was obtained as a 5 wt % solution of the polyelectrolyte having an equivalent weight of 1100. Other commercially available chemicals were reagent grade and were used as received. . 2-Ferrocenylethyl bromide was prepared by adapting the procedure described in ref 8: 1.5 g of PBr, was added dropwise ~
~
~~~~
653. (7) (a) Abruna, H. D. J . Elecrrochem. SOC.1985, 132, 842. (b) Kober, E. M.; Caspar, J . V.; Sullivan, B. P.; Meyer, T. J . Inorg. Chem. 1988, 27, 4587.
0022-3654/90/2094-8345$02.50/0
~~~~~
( 1 ) Buttry, D. A.; Anson, F. C. J . Am. Chem. SOC.1983, 105, 685. (2) Tsou, Y.-M.; Anson, F. C. J . Phys. Chem. 1985, 89, 3818. (3) Ohsaka, T.;Oyama, N.; Takahira, Y.; Nakamura, S . J . Elecrroanal. Chem. 1988, 247, 339. ( 4 ) (a) Martin, C. R.; Dollard, K. A. J . Elecrroanal. Chem. 1983, 159, 127. (b) White H. S.;Leddy, J.; Bard, A. J. J . Am. Chem. SOC.1982, 104,481 I . (c) Oyama, N.; Ohsaka, T.; Ushirgouchi, T.; Sanpei, S.;Nakamura, S. Bull Chem. SOC.Jpn. 1988, 61, 3103. (5) Hauser, C. R.;Lindasy, J . K. J . Org. Chem. 1957, 22, 355. (6) Lednicer, D.; Lindsay, J. K.; Hauser, C. R. J . Org. Chem. 1958, 23,
0 1990 American Chemical Society