Interfacial electron transfer in colloidal spinel iron oxide silver ion reduction in aqueous medium

Interfacial Electron Transfer in Colloidal Spinel Iron Oxide Silver Ion Reduction in Aqueous Medium JEAN PIERRE JOLIVET, ELISABETH TRONC, C H R I S T O P H E BARBE, AND JACQUES LIVAGE Chimie de la MatiOre Condensde, CNRS UA 302, Universitd Pierre et Marie Curie, T.54 E.5, 4 Place Jussieu, 75252 Paris Cedex 05, France

Received November l, 1989; accepted January 22, 1990 The redox behavior of spinel iron oxide colloids has been evidenced by studying silver reduction by magnetite using potentiometry, X-ray diffraction and M6ssbauer spectroscopy. At pH >/6, silver reduction involves an interfacial electron transfer without iron desorption, and the Fe304 --~ 3'-Fe203 transformation. In reverse, the 3'-Fe203 colloid can be refilled with electrons by Fe H adsorption, which results in the growth ofa Fe304 surface layer. This reduced colloid is much more reactive than the initial Fe304 particle. © 1990 Academic Press, Inc.

INTRODUCTION

Electron transfer at the solid/liquid interface plays a central role in various areas such as photoelectrochemistry or heterogeneous catalysis. It generally involves chemisorption on the surface of a metal oxide particle. A wide range of phenomena can occur. For instance, species adsorbed as an active complex can take part in a redox reaction with a reagent in solution (surface catalysis (1)). The surface of the particle then acts as a support; the core does not get involved. Electron transfers between the solid and the solution can occur when reducing ligands bond themselves to surface sites; this usually leads to the particle's dissolution (reductive dissolution (2, 3 )). Electron transfers can also be induced by irradiating the oxide when it is semiconducting (4, 5 ). Electron-hole pairs are created in the solid, and electrons transferred into the conduction band can be trapped by the adsorbate. The opposite may occur: by means of a photosensitive adsorbate, electrons can be injected into the solid (6). Also free radicals may ap-

465 Journal of Colloid and Interface Science, Vol. 138, No. 2, September 1990

pear in solution, and react with the colloid (photoreductive dissolution (7)). Because of their electronic properties, mixed-valent oxides may exhibit a specific behavior. Thus, in magnetite Fe304, fast intervalence transfers between octahedrally coordinated iron ions give rise to exceptional catalytic activity (8). They also lead to enhanced redox response against surface acid-base phenomena (9): at pH 2, in the absence of any oxidizing agent, colloidal Fe304 is converted into 3'-Fe203. The conversion is driven by desorption of surface Fe ~I ions. Electron delocalization and small structural changes renew the ferrous sites at the surface and feed the reaction up to completion. In reverse, Fe Hions adsorbed onto 3'-Fe203 colloids grow a superficial layer of Fe304 (10). Electrons are injected from the adsorbed layer into the bulk and delocalized there. These two examples show that electron transfers can proceed in both ways through the interface, and that the spinel iron oxide colloid may behave as an electron tank (microelectrode or microbattery). Here, we report

0021-9797/90 $3.00 Copyright © 1990 by Academic Press, Inc. All rights of reproduction in any form reserved.

466

JOLIVET ET AL.

a study of the interaction of Ag + ions with colloidal Fe304 and with 3,-Fe2Os reduced by Ve Il adsorption. These results further illustrate the electrochemical behavior of these materials. We recall that Fe304 has the inverse spinel structure, with F e IIl ions evenly distributed among tetrahedral (A) and octahedral (B) sites, whereas Fe n ions occupy B sites only. The chemical unit may be written as (FenI)A(FemFen)ao4. Fast electron hopping between B ions at room temperature makes them appear with the average valence 2.5 in Mrssbauer spectroscopy. Nonstoichiometric magnetite has a deficient B sublattice. The ion distribution may be written as (FelII)A[(FemFen)l_3x FeIsI~3x]BO4where [] represents a vacancy. For x vacancies, there are ( 1 - 3x) paired and 5x unpaired FenI(B) ions. The limit (Fem)A(Fems/3[31/3)BO4 corresponds to "y-Fe203.

MATERIALS A N D M E T H O D S

Synthesis Colloidal magnetite was prepared according to the procedure already described (9) by coprecipitating the stoichiometric mixture 2 Fe(NO3 )3 + Fe(C104)2 into an NH3 solution. Great care was taken to exclude oxygen during all the stages of the preparation. The solid, washed with water and separated by magnetic settling, was dispersed in deaerated water and kept under argon atmosphere. The suspension at pH ca. 8.5 had a total concentration in Fe of ca. 0.5 mol dm -3. Colloidal 3,-Fe203 was prepared by a similar way, but the alkalinization of the mixture 2 Fe(NO3)3 + Fe(C104)2 was carried out in air. In order to promote oxidation, the alkaline suspension-was vigorously stirred in air for 15 h. After separation by magnetic settling, the solid was treated with HC104 3 mol dm -3, centrifuged, and dispersed in distilled water, giving a homogeneous cationic sol ( 11 ) with pH ca. 2. After being aged for a few days, the Journal of Colloid and Interface Science, Vol. 138,No. 2, September 1990

sol was ultrafiltered to separate desorbed residual Fe n ions (9), and redispersed in distilled water. This was repeated until the Fe n content (FeII/Fem) in the colloid was less than 10%. For the sake of simplicity this oxidized magnetite will be called "y-Fe203. Adsorption of Fe n on ff.e 3'-Fe203 colloid was carried out as previously reported (10). A solution of Fe( C I O 4 )2 was added to a given amount of sol carefully deaerated. The overall composition of the mixture was fixed at Fen/ Fe In slightly below 1/2. The adsorption was regulated under N2 by adding N(CH3)4OH using a potentiograph. Base addition was stopped at the end of the process (pH ca. 9.5 ).

Techniques Reactions between the colloids and Ag + were studied by potentiometry. Known volumes of colloidal suspensions, homogenized by stirring, were added to deaerated water. N2 was continuously bubbled. A variable amount of AgNO3 0.1 mol dm-3 was introduced and after a few minutes the mixture was titrated with N(CH3)4OH using a potentiograph Metrohm E536. Back titrations were performed by adding HC104. Kinetic studies were carried out with N (CH3)4OH addition adjusted automatically to keep the pH constant, using a titrimeter Metrohm Combi-Titreur 3D. Sols and suspensions were ultrafiltered under N2 pressure using a cell and membranes P30 Amicon. The suspension composition was determined after dissolution of the sample in concentrated HC1. Fe n was titrated potentiometrically with K z C r 2 0 7 . The total Fe content was determined by the same way, after reduction of iron by SnC12. Transmission electron micrographs were obtained using a Jeol 100 CXII apparatus. Samples were prepared by evaporating very dilute sols or suspensions onto a carbon coated grid. Particle sizes were estimated by measuring ca. 200 particles.

SILVER

REDUCTION

X-ray diffraction ( X R D ) and M6ssbauer effect investigations were carried out on solids. X R D measurements were performed using a powder diffractometer Philips PW 1310 operating in the reflexion mode with the radiation C o K a . Mrssbauer spectra were recorded at room temperature using a conventional spectrometer Elscint Inel and a 57Co/Rh source. Velocities were calibrated with an iron foil. Because the M6ssbauer spectrum is sensitive to the aggregation state of the particles (12), and materials are oxidized by drying, absorbers were made up in two different ways. Absorbers of rich F e Ix particles, always strongly agglomerated in the suspensions, were made up as rigid homogeneous films by dispersing the suspension into a solution of polyvinylic alcohol and drying under N2 circulation. Oxidized particles, with limited aggregation in the sols ( 11 ), were powdered by ultrafiltering the sols, drying the solid under vacuum, and embedding it in an inert resin. RESULTS

Freshly prepared magnetite was practically stoichiometric, with a ratio F e l I / F e III in the range 0.47-0.5. The particles, roughly spherical, were ca. 100 A in size (~ ~ 30 A). 3'Fe203 particles were ca. 85 A in size (a ~ 25 A), with a ratio FeII/Fe nI ca. 0.05. 1. Reaction o f A g + on Fe304

Addition ofAgNO3 (Ag+/Fe n = 1.5 ) to the suspension of colloidal magnetite (pH 8.5 ) induced instantaneous acidification of the medium, with pH dropping down to 2.5-3. After a few minutes, the solid and the solution were separated by ultrafiltration. The solid washed with water and dried at room temperature gave the X R D pattern shown in Fig. 1. It exhibits two systems of lines, typical of oxidized spinel iron oxide (a = 8.35 A) and of Ag metal (a = 4.08 A). Particle sizes deduced from the line widths using the Scherrer formula are equal to 100 and 150 A, respectively. Electron micrographs of the sus-

BY IRON

467

OXIDE

111 Ag

311 sp

2{10

s~ I

I

I

sp

] 50

I

I

I

I

[ 40

I

i

t o 2e

I

i 30

lqG. 1. X-ray diffraction pattern of colloidal Fe304 after reaction with Ag + ions.

pension (Fig. 2) also show two types of particles that were identified by their diffraction patterns: spinel iron oxide (I) ~ 100 A) and silver (I) ~ 200 A) particles. Mrssbauer spectra of the material before and after reaction with Ag + are shown in Fig. 3a and 3b. The aspectrum is typical of magnetite particles about 100 A in size ( 13 ). It is mainly based on the two sextets of bulk Fe304, one for Fe In ions in tetrahedral sites (A), and the other for FeII and Fem ions in octahedral sites (B), with the average valence of 2.5. Relative spectral areas are theoretically in the ratio A:B = 1:2. The b-spectrum gives no evidence of any oxide or oxyhydroxide other than the spinel oxide. It differs from the a-spectrum only in the relative areas of the two basic sextets. The Fe 2"5+ ion contribution is significantly reduced. This is characteristic of nonstoichiometric magnetite (14), where unpaired Fe In (B) ions have practically the same hyperfine parameters as Fe nI ions of type A, complete oxidation leading to a nearly symmetrical six-line pattern. The composition of the b-sample was actually Fe"/ Fe III= 0.25. Both spectra exhibit asymmetric line broadening related to particle size effects (12, 13, 15): The analysis of the filtrate showed that it contained only Ag + ions and small quantities Journal of Colloid and Interjace Science, Vot. I38; No. 2, September 1990

468

J O L I V E T E T AL.

FIG. 2. Electron micrographof suspension of colloidal Fe304 after reaction with Ag + ions. (Silverparticles are an'owed).

of Fe 2+ ions (ca. 5% of the total FeII amount)• The presence of Fe 2+ ions in solution was due to superficial hydrolysis of iron oxide particles

....•~

~.~ .

~.

e...,.:~.

~,

. :• • • . . . t

•

- •'~

g

•.." ' ! '5:. ?" '

g

< >'274.

r.'.X:i:

b • :" •

.

~" •

.-

4,

. ' -: . ,:

•

.'.. , J •

-12

•

.

• .

.

)

".~ ~.

I

, ,I

-6

0

,"

I

6

12

Velocity mm,/s FIG. 3. M 0 s s b a u e r spectra ( 300 K ) of colloidal Fe304 ( a ) freshly p r e p a r e d a n d ( b ) after reaction w i t h Ag + ions.

Journal o f Colloid a n d Interface Science,

Vol. 138, No. 2, September 1990

in acidic medium, before the filtration of the suspension (9). F r o m these results it appears that Ag + ions in solution can be reduced to Ag ° metal by colloidal Fe304 which oxidizes without noticeable size variation. Since the reduction of Ag + by the colloid acidifies the medium, the reaction can be studied quantitatively by neutralizing the suspension. N ( C H 3 ) 4 O H titration curves of mixtures with different Ag + amounts are shown in Fig. 4. They exhibit three successive steps at p H 4, p H 6, and p H 10, respectively. The latter corresponds to complete neutralization of the overall system. The curve analysis (Fig. 5) shows that whatever the amount ofAg + added, the total a m o u n t of base needed corresponds to a ratio O H - / A g + = 1. Back titrations also gave evidence of the same equivalent point at p H 6. The suspension at p H 6 was ultrafiltered and the filtrate was

469

SILVER R E D U C T I O N BY I R O N OXIDE

pH

a

b

c,

d/

e /

f /

lO

same as the amount of base needed from pH 6 to pH 10 in the forward titration. It was concluded that this amount of base just served to precipitate excess Ag + ions as AgOH. The amount of reduced Ag + ions, and therefore that of oxidized Fe ~I ions, was calculated (Fig. 5 ). The stage between pH 4 and pH 6 in the forward titration (Fig. 4) corresponds to the secondary reaction due to reduction of Ag + in excess in solution by desorbed Fe z+ ions, according to the redox reaction (16) Fe 2+ + Ag + + 3 O H - ~ Fe(OH )3 nt- Ag °.

.1

,2

.3

.4

,5

mmoI.HO"

FIG. 4. Titration of Fe304 in the presence of AgNO3. ]Feltot,l = 2.1 × 10 -2 mol d m -3. Fen/Fe Ix1 = 0.45.2.86 × 10 -4 mole FdI. x = IAg+laa/lFenl: x = 0 (curve a); 0.215 (b); 0.587 (c); 0.978 (d); 1.566 (e); 1.96 (f). Titrations were performed 2 m i n after AgNO3 addition to Fe304 suspension, using N ( CH3 )4OH (0.123 mol d m -3 ) added at the rate of 1 cm 3 m i n -1 .

analyzed. It contained no iron, and only silver titratable by chloride. The amount of acid needed from pH 10 to pH 6 was exactly the

In the present conditions, Fe 2+ ion desorption in acidic medium (9) involves less of 5% of the FeII content of the colloid. It actually depends on the delay between the additions of Ag + and of base (usually a few minutes). The equivalent point at pH 4 corresponds to neutralization of free protons which result from Ag + reduction by the colloid. Analysis of the shift of the titration curves with the Ag+/Fe ratio is consistent with these different steps. The reaction kinetics was studied by the automatic measurement of base addition as a function of time at pH 5.7, after introduction of Ag + into the Fe304 suspension. This pH value was chosen in order to avoid Fe 2+ desorption and AgOH precipitation. Two distinct stages appeared (Fig. 6a). The first step

.5

r

a

;5o f

E

o

i

i

-- • 0

•

~n

a ,L

A~ .,5 m m o l . A g + ad.

FIG. 5. Exploitation of titration curves of Fe304 in the presence of various a m o u n t s of AgNO3. A m o u n t of base added at pH 10 (O), of reduced Ag ( I ) , and of desorbed Fe 11( • ) vs. a m o u n t of AgNO3 initially added.

0

/

10

I

i

20

i

hours

FIG. 6. Kinetics o f F e n oxidation in mixtures (a) Fe304 + AgNO3 and (b) [3'-Fe203 + Fe(C104)/, F e n / F e m = 0.46] + AgNO3. Suspensions at 25°C. (a) Ag+~a/ 1t Feinitial = 1.92, IFe1Ilinitial = 1.65 × 10 -2 mol d m - 3 ; (b) + I1 A g a d / F e i n i t i a l = 1, [ F e n I initial = 1 . 8 4 × 1 0 - 2 m o l d m -3.

Journal of Colloid and Interface Science, Vol. 138, No. 2, September 1990

470

JOLIVET ET AL.

corresponded to oxidation o f 40 to 50% o f the FCII content. The second step, very slow, was followed for about 20 h. At this level, 70% of the Fe n content was oxidized.

a

.

~.

m~

•.; .:

.,~:

.-".;

•

.: ..

,..

2. Interaction o f A g + with ~/-Fe:03 Reduced by Fe I1 Adsorption <

Adsorption o f Fe n on the y-Fe203 colloid was followed potentiometrically during the addition o f base up to p H 9.5 (A to B, Fig. 7). The overall FeH/Fe m composition was fixed at 0.4. We have previously shown (10) that FcII uptake goes on as long as the overall composition o f the colloid corresponds to FeI~/ Fem ~< 0.5, and that adsorbed Fe I~ grows a superficial spinel (Fe304) layer. T h e M6ssbauer spectra confirm that no structural phase other than spinel is p r o d u c e d by Fe ~ adsorption (Fig. 8a, b) and that the Fe n content has clearly increased. Assuming identical recoilless fractions for all the atoms,

,A

•

~

:

t,,

..

•.;", ' 2 :g

. .

~ "

",'.-

,!: ..: c • ,:.

='..

-.,

-......:.. •~

::...;: I

-12

-6

:...

".:

..

....

-.

i

I

0

6 Velocity

12 mm/s

FIG. 8. Mrssbauer spectra (300 K) of (a) colloidal ~/-Fe203; (b) after reaction with FeI~;and (c) after further reaction with Ag+ ions.

pH

least-squares fits led to a ratio F e n / F e m ~ 0.3. It is less than the expected value, probably as a result o f the high reactivity o f the adsorbed layer which oxidized during the sampling• After this reaction, Ag + addition (Ag + / Fe n = 1.4) led to very rapid acidification o f the suspension (B to C, Fig. 7) as with magnetite• Electron micrographs and X R D patterns of 6 the material after reaction with Ag + are very similar to those obtained in the case o f magnetite (Fig. 1 and 2). The M r s s b a u e r spectrum o f the material after reaction with Ag + is practically identical to that o f the starting T-Fe203 colloid (Fig. 8a, I i I I | I c). Oxidation was therefore nearly completed, .1 .2 .3 .4 .5 m m o l . HO" without any m a j o r structural change. Particle FIG. 7. Ag+ reduction after FeIIadsorption on "y-Fe203• size distribution seems not to be altered noA to B: Neutralization of mixture [~'-Fe203 + Fe(CIO4)2, ticeably by the overall treatment. FeH/Fem = 0.40] using N(CH3)4OH (0•142 tool dm -3) The kinetic study at p H 5.7 (Fig. 6b) shows up to pH 9.5. IFell Iaa = 7•2 × 10-3 mol din-3 (0.168 mole Fell). B to C: Addition of AgNO3 (Ag+/Fen = 1.41). C that 80% o f previously adsorbed Fen were quasi-instantaneously oxidized. The effect o f to D: Titration of the global mixture by N(CH3)4OH. 10

Journal

of Colloid and Interface Science,

Vol. 138, No. 2, September 1990

SILVER REDUCTION BY IRON OXIDE

the reaction kinetics was therefore much weakened in comparison with the case of magnetite. DISCUSSION

The results unambiguously show that extensive redox reactions can take place at the surface of spinel iron oxide colloids. Such reactions involve the whole Fe304 particle; they start up a solid state reaction which makes the particle composition span the existence range of the nonstoichiometric spinel phase. In suspension at 4 ~< pH ~< 9 and under inert atmosphere (N2, Ar) colloidal magnetite is chemically stable. In more acidic medium, the Fe H ions are desorbed because of acidic hydrolysis of surface sites (9). It is really the Ag ÷ ions in the medium which induce the redox phenomena. The fact that no iron is desorbed from the particle at pH ~ 6 indicates that iron is oxidized in situ by adsorbed silver. The stoichiometry of the reaction may be written as F '~surfaoe ~II +

Ag + + O H

"-~ FenIOHsurface + Ag °.

In operating conditions (pH ~ 6), the reaction proceeds with two kinetic stages. The first one, very fast, corresponds to oxidation o f s u p e r f i c i a l FeII ions. The second stage is slow because it necessitates the renewal of Fe n ions at the surface of the colloid (electron and Fe ~ ion diffusion through the oxide network). Oxidation of surface iron ions starts up oxidation in depth. Let us consider the octahedral site sublattice in Fe304, with its Fe 25+ ions due to fast electron hopping (Scheme 1 ). Ado~ H A+

r,"no / ~ g

o-. H

O~,'n

nn ,,.

6

e~.O

,}

0

•

0

0\•

®

0

•

®

0 FeT M • Fe3+

*

/ Ag

•

0

0

H O ~,'n7

H ".,7~O

•

H 0?/77

•

•

0

•

0

0

n

0

®

®

0

•

0 Oxygen

SCHEME 1

471

sorbed Ag + pumps out a mobile electron, leaving 2 Fe 3+ ions with a positive charge in excess in the lattice. Charge imbalance is locally compensated by creation of a vacancy and outward migration of the corresponding cation. New surface Fe 3+ ions complete their coordination by means of hydroxylation. Electrons supplied from the interior keep the reaction going, which implies correlative vacancy creation and migration of excess cations that progressively renew the surface. Electroneutrality in vicinity to a vacancy requires five out of its six near neighbors to be Fe 3+ ions. At first, it may speed up the reaction by facilitating electron localization in surface sites. In the first very fast kinetic step of the reaction, 40 to 50% of the Fe H content is oxidized. It corresponds to oxidation of an outer layer, roughly a unit cell thick (8 A). The second step is likely rate-determined by cation diffusion. Then the reaction slows down. Such a process is the same as that proposed for aerial oxidation of magnetite ( 17 ), or its behavior in acidic medium (9). The reduction ofAg + by the 3,-Fe203 colloid after reductive Fe n adsorption likely proceeds in a similar way. The main difference with magnetite lies in the extent of the first, rapid, kinetic step of the reaction: 80 to 90% of the Fe n content is now quasi-immediately oxidized. The difference in reactivity probably comes from structural variations between the two colloids, even though they are characterized by the same Fe~I/Fe m ratio. Fe H uptake by the 3'-Fe203 colloid requires the growth of a magnetite layer and the reduction of inner octahedral Fem ions. In order to keep electroneutrality, electron injection must be coupled with inward migration of cations. These may be iron cations or protons which diffuse more easily. Both processes probably occur, iron diffusing only in relatively small depth so as to weaken concentration gradients. Enhanced reactivity ofAg ÷ on reduced 3,-Fe203 by comparison with Fe304 supports such a .view. Then, since diffusion involves only iron ions contained in an outer layer with limited thickJournal of Colloid and Interface Science, Vol. 138,No. 2, September1990

472

JOLIVET ET AL.

ness, r e o x i d a t i o n is accelerated. M o s t Fe II ions are o x i d i z e d d u r i n g A g + a d d i t i o n , a n d FeII des o r p t i o n b e c o m e s t o o slow a process to be detected. Silver is n o t d e p o s i t e d as a thin c o a t i n g o n i r o n oxide particles b u t a p p e a r s in the f o r m of large particles. S t r u c t u r a l m i s m a t c h b e t w e e n the two phases a n d roughness o f the oxide surface p r e c l u d e the p a r t i c l e ' s silvering. T h e strong A g - A g i n t e r a c t i o n s in the m e t a l very p r o b a b l y i n d u c e a n i s o t r o p i c clustering o f silver a t o m s after the r e d u c t i o n o f a d s o r b e d Ag + ions. Large particles likely result f r o m the lowering o f surface t e n s i o n o f the m e t a l l i c system. Such a growth process is k n o w n to o c c u r in the e v o l u t i o n o f silver colloids ( 18). CONCLUSION W e have s h o w n t h a t the r e d u c t i o n o f A g + ions b y colloidal m a g n e t i t e involves a n interfacial electron transfer, w i t h o u t i r o n desorption. A c i d - b a s e c o n d i t i o n s let excess c a t i o n s e m e r g i n g f r o m the core stay at the surface, c o o r d i n a t e t h e m s e l v e s to O H ions f r o m solution, a n d extend the spinel lattice. It is w o r t h n o t i n g t h a t after the t r a n s f o r m a t i o n Fe304 3'-Fe203 has occurred, o n e c a n refill the colloid with electrons a n d m u c h increase the reactivity as well. Such b e h a v i o r , a n a l o g o u s to t h a t o f a m i c r o b a t t e r y , suggests interesting catalytic potentialities. O t h e r e l e m e n t s are r e d u c i b l e b y colloidal Fe304 t h r o u g h an a n a l o g o u s process. F o r exa m p l e , C u 2+ i o n s give rise to the r e a c t i o n C u 2+ ~ C u O H . ACKNOWLEDGMENT We are grateful to M. Lavergne (CRMP, Universit6 P. et M. Curie) for electron microscopy measurements.

JournalofColloidandInterfaceScience,Vol.138,No. 2, September1990

REFERENCES 1. Stumm, W., Wehrli, B., and Wieland, E., Croat. Chem. Acta 60, 429 (1987). 2. Blesa, M. A., and Maroto, A. J. G., ,L Chim. Phys. Phys.-Chim. Biol. 83, 757 (1986). Baumgartner, E., Blesa, M. A., Marinovitch, H. A., and Maroto, A. J. G., Inorg. Chem. 22, 2224 (1983). 3. Segal, M. G., and Sellers, R. M., Z Chem. Soc. Chem. Commun., 991 (1980). 4. Henglein, A., in "Mechanism of Reactions on Colloidal Microelectrodes and Sizes Quantization Effects," Topics in Current Chemistry, Vol. 143, p. 113, Springer-Verlag, Berlin/New York, 1988. 5. Gratzel, M., J. Chim. Phys. Phys.-Chim. Biol. 84, 1049 (1987). 6. Waite, T. D., and Morel, F. M. M., Z Colloid[nterface Sci. 102, 121 (1984). 7. Buxton, G. V., Rhodes, T., and Sellers, R. M., d. Chem. Soc. Faraday Trans. 1 79, 2961 (1983). 8. Rethwisch, D. G , and Dumesic, J. A., Appl. Catal. 21, 97 (1986). 9. Jolivet, J. P., and Tronc, E., Z Colloid Interface Sci. 125, 688 (1988). 10. Tronc, E., Jolivet, J. P., Lefebvre, J., and Massart, R., J. Chem. Soc. Faraday Trans. 1 80, 2619 (1984). 11. Jolivet, J. P., Massart, R., and Fruchart, J. M., Nouv. J. Chim. 7, 325 ( 1983 ). 12. Tronc, E., and Jolivet, J. P., ,/. Phys. (Les Ulis, Fr.) 49, C8, 1823 (1988). 13. Morup, S., Dumesic, J. A., and Topsoe, H., in "Application of M6ssbauer Spectroscopy" (R. L. Cohen, Ed.), Vol. 2, p. 1, Academic Press, New York, 1981. 14. Topsoe, H., Dumesic, J. A., and Boudard, M., J. Phys. (Les Ulis, Fr.) 35, C6, 411 (1974); Haneda, K., and Morrish, A. H., J. Phys. (Les Ulis, Fr.) 38, CI, 321 (1977). 15. Tronc, E., and Bonnin, D., or. Phys, Lett. 46, 437 (1985); Tronc, E., and Jolivet, J. P., Hype~ne Interact. 28, 525 (1986). 16. Jolivet, J. P., and Tronc, E., unpublished results. 17. Sidhu, P. S., Gilkes, R. J., and Posner, A. M., J. Inorg. Nucl. Chem. 39, 1953 (1977). 18. Jolivet, J. P., Gzara, M., Mazi~res, J., and Lefebvre, J., J. Colloid Interface Sci. 107, 429 (1985).