J . Phys. Chem. 1988, 92, 2075-2076
greater charge. This should be even more exaggerated for metal halides or metal fluoroanions. Therefore, ion bombardment of fluoroanions should induce preferential loss of F and a subsequent reduction of metal atom, which is observed. Both models predict the general behavior of the decomposition of the fluoroanions. However, due to the unreliability of surface binding energies for complex species, the thermal-spike model can be more easily applied in these complex systems.
Conclusions All of the transition-metal fluroanion compounds studied preferentially lost potassium and fluorine upon Ar+ bombardment,
2075
while the central metal atom was reduced to lower oxidation states. Reduction to the metal was observed for K2TaF7 and K2NbF7 but not for K2TiF6. These results are consistent with the thermal-spike model of energy dissipation. The resulting surface contains those products with high negative free energies of formation. Acknowledgment. This material is based on work supported by the Air Force Research Office under Contract No. F1962886-K-0013 and by the R. A. Welch Foundation under Grant E656. Registry No. K2TiF6, 16919-27-0; K,TaF,, 16924-00-8; K2NbF7, 16924-03-1; Ar',
14791-69-6.
COMMENTS Comments on "Behavior of Surface Peroxo Species in the Photoreactions at TIO," Sir: In a recent investigation by means of linear cyclic voltammetry, Ulmann et al.' detect the generation of surface species (surface states) at polycrystalline TiO, film electrodes in contact with aqueous 0.1 M NaOH. From the identical position of the cathodic peaks appearing in the voltammograms, the authors conclude that species of the same nature are generated either during bandgap illumination at open circuit or anodic bias or during electroreduction of dissolved oxygen in the dark. Photogeneration of surface states at TiO, electrodes was first reported by Wilson2 with single crystals, in both acidic and basic electrolytes, and more recently with polycrystalline electrodes by Salvador et who associate these surface states with chemisorbed H202species photogenerated from basic hydroxyl groups, (OH),, singly coordinated to surface Ti, ions by a covalent bond, according to the reactions3q4 Ti,-(OH)b
20H'
Ti,
-
+ h+ ---*
+ H,O,
Ti,-OH'
(1) (2)
H202
Ti,-H20,
(3)
Further experimental evidence about the chemisorption of hydrogen peroxide on a Ti02 electrode, leading to the generation of peroxo titanium complexes, as assumed in (3), has been given by us r e ~ e n t l y . ~ Following the thesis of Brown and Darwent,6 and against our model, Ulmann et a1.I postulate the peroxo titanium species detected by them are photogenerated from surface acidic groups, (OH),, doubly coordinated to Ti, ions, according to the following reactions in basic medium: Ti,-(OH),
-
+ NaOH Ti,-0- + h+
2(Ti,-O')
-
Ti,-O-Na+
-
+ H,O
(4)
Ti,-0'
(5)
Ti,-0-0-Ti,
(6)
Besides, Ulmann et al.' state that the conproportionation reaction (6) appears plausible for Ti,-O' radicals originating from hydroxyl (1) 6523. (2) (3) (4)
groups singly coordinated to Ti, ions, forgetting that these are basic and not acid hydroxyls as they previously assumed in reaction 4. An additional objection to Ulmann et al.'s model' concerns the photogeneration of T4-0' radicals, which, as far as we know, have never been detected. Hole capture by a surface acidic hydroxyl should produce a Ti,-0'- group rather than a Ti,-0' one, as Ulmann et a1.l claim. The error of Ulmann et al. probably consists in considering that 2p electrons of 02-ions are shared by Ti ions in a bonding orbital, so that in eq 4 these authors write Ti,-0-. It is well-known that the top of the valence band in TiO, is formed by 0 2p nonbonding orbitals' (electrons are localized in the 2p levels of the 02-sublattice). Therefore, a Ti,-0'- group should be generated when a hole is trapped at a 02-surface ion, and consequently eq 4 and 5 should be replaced by
-
+
Ti,-O'-Na+ 2 H 2 0 (7) Ti,-(02--H30+) + N a O H + h+ (OH), As suggested by King and Freund for the case of Mg0,8 recombination of 0'- surface radicals may produce 02,-peroxy anions, which could remain coordinated to a Ti, ion after restructuration of the surface 2Ti-0'Ti-OZ2- Ti-V (8)
-
+
where V represents an anionic vacancy. Further, OZ2-ions would probably desproportionate releasing an oxygen atom.8 Summing up, photooxidation of surface acidic hydroxyls could hardly lead to the generation of peroxo complexes, as proposed by Ulmann et a1.I Rather, hole trapped at Ti,-O'- surface groups would be transferred to basic surface hydroxyls to generate OH' radicals. Then, titanium peroxo species would be formed spontaneously, as already proposed by us (reactions 2 and 3), without restructuring of the TiO, surface. On the other hand, the model proposed by Ulmann et al. does not explain the generation of peroxo complexes in the dark during the electroreduction of dissolved 0,. This drawback is obviated in our model, since hydrogen peroxide is a main product of the electroreduction of oxygen at basic pH on Ti02 electrode^.^ A fraction of the so-generated hydrogen peroxide is chemisorbed,I0 producing surface peroxo complexes of the same type as those photogenerated. This mechanism would be specially efficient with colloidal particles because of their enormous specific area and would explain why generated hydrogen peroxide is hardly detected with T i 0 2 dispersions."
Ulmann, M.; Tacconi, N. R.; Augustynski, J. J . Phys. Chem. 1986, 90, (7) Tsutsmi, K.; Aita, 0.; Ichikawa, K. Phys. Reu. B: Solid State 1977,
Wilson, R. H. J . Electrochem. S o t . 1980, 127, 228. Salvador, P.; Gutibrrez, C. J . Phys. Chem. 1984, 88, 3696. Salvador, P.; Gutibrrez, C. Surf. Sci. 1983, 124, 398. (5) Ferrer, I. J.; Muraki, H.; Salvador, P. J . Phys. Chem. 1986, 90, 2805. (6) Brown, G. T.; Darwent, J. R. J . Phys. Chem. 1984, 88, 4955.
0022-3654/88/2092-2075$01.50/0
15, 4628.
(8) King, B. V.; Freund, F. Phys. Reu. B: Condem. Mutter 1984, 29, 5814. (9) Tafalla, D.; Salvador, P. J . Electroanal. Chem. 1987, 237, 225. (10) Tafalla, D.; Salvador, P. Ber. Bunsen-Ges.Phys. Chem. 1987, 91, 475. ( 1 1 ) Harbour, J. R.; Tromp, J.; Hair, M. L. Can. J . Chem. 1985,63,204.
0 1988 American Chemical Society
2076
J . Phys. Chem. 1988, 92, 2076-2078
Moreover, Ulmann et al.’ observe that the voltammetric cathodic peak corresponding to the electroreduction of hydrogen peroxide, both dissolved and preadsorbed, shifts toward positive potentials by about 250 mV with respect to the peak observed during the electroreduction of photogenerated surface species. This observation is considered by the authors as evidence that chemisorbed, photogenerated hydrogen peroxide is not responsible for these species. However, these results are not conclusive. Rather, the peak attributed to the electroreduction of surface peroxo species by Ulmann et al. is due to the reduction of oxygen photogenerated inside the pores of an extremely rough TiO, surface. This oxygen trapped at surface cavities is hardly affected by electrolyte stirring’, and should diffuse very slowly toward the bulk of the electrolyte, as inferred from Figure 1 of ref 1. On the other hand, the cathodic shift of the peak of electroreduction of photogenerated oxygen with respect to the peak corresponding to dissolved 0, (see Figure 4 of ref 1) can be attributed to 0, supersaturation of the electrolyte inside the pores. Again, the peaks that Ulmann et al. show in their Figure 6 seem to be due to the electroreduction of oxygen photogenerated during electrode illumination at +O. 17 V (RHE), which is clearly a too anodic potential for complete electroreduction of photogenerated 0,. Experiments should be more carefully designed, and a TiO, single crystal with a very smooth surface should be employed before any conclusion can be reached from the type of experiments performed by Ulmann et al.’ Finally, we would like to comment on the results of Ulmann et al.’ about the increase of the anodic photocurrent by addition of H,O, to the basic electrolyte. We have recently studied the interaction of T i 0 2 with hydrogen peroxide, at both acidic and basic pH, by impedance and electrolyte electroreflectance techniquess We showed that, at pH 11.3, the distribution of potential at the TiO,-electrolyte interface was noticeably perturbed by addition of H z 0 2to the electrolyte ( V , is shifted negatively by about 200 mV), while it remained apparently unperturbed at pH 3. This was attributed to the fact that H02-, the stable form of hydrogen peroxide at basic pH, interacts more strongly with TiO, than H202,which is stable at acidic pH. The shift of Vb toward negative potentials at basic pH, and the simultaneous band-bending increase under constant polarization potential, account for the observation of Ulmann et al.’ that the addition of H 2 0 2to the NaOH solution causes a marked increase of anodic photocurrent in the initial part of their photocurrent-voltage curve (Figure 11). Their conclusion that the effectiveness of H202in increasing the magnitude of the photocurrent must be the result of a large value of the rate constant for charge transfer’ seems therefore incorrect. Acknowledgment. This work was partially financed by the USA-Spain Joint Committee for Scientific and Technological Cooperation under Contract CCA 83/038. Registry No. TiO,, 13463-67-7; H,O, 7732-1 8-5. (12) Salvador, P.; Decker, F. J . Phys. Chem. 1984, 88, 6116.
Instituto de C6talisis y Petroleoquimica (CSIC),Serrano 119, 28006 Madrid, Spain
P. Salvador* D. Tafalla
Received: February 17, 1987; I n Final Form: August 31, 1987
Reply to Comments on “Behavior of Surface Peroxo Species In the Photoreactions at 110,” Sir: The cathodically reducible surface species (surface states), photogenerated during anodic polarization of an n-type TiO, photoelectrode, have originally been considered by Wilson’ to be the intermediates of the photooxidation of water. Referring to (1) Wilson, R. H . J . Electrochem. SOC.1980, 127, 228.
0022-3654/88/2092-2076$01.50/0
Wilson’s paper, Salvador, has proposed the surface hydroxyl groups a t TiOz to be the surface states involved in both photoassisted (anodic) and dark (cathodic) electrode reactions. Subsequently, Salvador and GutiErrez’ have investigated by cyclic voltammetry the cathodic reduction of the species photogenerated at polycrystalline Ti02 electrodes (obtained by sintering a rutile powder), both in alkaline (1 M KOH) and in acidic (1 M Na2S04acidified to pH 3) solutions. For all measurements the solutions were saturated with oxygen, which rendered the resolution of possible multiple signals rather difficult. The increase of the cathodic voltammetric peak, following a few-seconds exposure of the TiO, electrode to UV irradiation under anodic bias, has been attributed to a local supersaturation of the solution by photogenerated molecular oxygen.’ At that time, the authors have underlined that the electrochemical reduction of the photogenerated 0, in alkaline as well as in acidic media takes place at the both being mediated by same potential as that of “dissolved 02”, basic OH groups at the TiO, ~ u r f a c e . ~ In a later paper,4 Salvador and Gutiirrez report again on analogous voltammetric measurements, carried out with sintered Ti02 electrodes in air-saturated 1 M aqueous N a 2 S 0 4at pH 3 (the solution used already in previous experiments). Due presumably to different experimental conditions (prepolarization of the illuminated T i 0 2 electrode at a more positive potential, different potential sweep range, higher sweep rate), the authors4 discover a new voltammetric peak, shifted ca. 0.2 V to more positive potentials with respect to the peak assigned previously’ to the reduction of photogenerated oxygen species. This new peak is explained in terms of the reduction of H202molecules chemisorbed at the TiO, surface, such an assignment being based on M H20, the fact that the TiO, electrode, first immersed in a solution and then dried in air, exhibited a similar reduction peak.4 The authors have suggested that hydrogen peroxide, photogenerated in an intermediate stage of the photooxidation of water, goes, in part, into the solution and, in part, becomes readsorbed (chemisorbed) at the TiO, surface, forming stable surface states! In order to check the latter hypothesis, Salvador and Decker5 have performed a series of experiments with a rotating Pt ringilluminated TiO, disk electrode (RRDE). In spite of the high sensitivity of the RRDE technique6 for the detection of even unstable reaction intermediates, the formation of H 2 0 2 at the illuminated TiO, disk photoanode could not be confirmed. Rather surprisingly, the authorsS conclude that hydrogen peroxide is actually photogenerated as an intermediate product of photoelectrolysis of water at n-type Ti02, but most of it undergoes further photooxidation before diffusing away from the electrode. Despite the clearly negative results of the above RRDE study, the validity of the previously proposed model, involving the photogeneration and the subsequent chemisorption of H202at titanium d i ~ x i d e , ~ have not been reconsidered. We were prompted to reexamine the electrochemical behavior of the surface species, formed in the course of photooxidation of water at Ti02, by the reports dealing with the photogeneration of surface-bonded peroxo complexes during photocleavage of water in aqueous suspensions of TiO, Under the latter conditions, the surface-bound peroxides are, in fact, the main final product of the photooxidation of water. This situation is clearly in contrast with the case of the anodically biased T i 0 2 photoelectrode at which molecular oxygen is photogenerated with high faradaic efficiency. Our recent cyclic voltammetric measurements,1° to which the (2) Salvador, P. J . Electrochem. SOC.1980, 127, 2650. (3) Salvador, P.; Gutitrrez, C. Chem. Phys. Left. 1982, 86, 131. (4) Salvador, P.; Gutitrrez, C. J . Phys. Chem. 1984, 88, 3696. (5) Salvador, P.; Decker, F. J . Phys. Chem. 1984, 88, 61 16. (6) Albery, W. J.; Hitchman, M. L. Ring-disc Electrodes; Clarendon: Oxford, 197 1 . (7) Yesodharan, E.; Gratzel, M. Helu. Chim. Acta 1983, 66, 2145. (8)Yesodharan, E.; Yesodharan, S.; Griitzel, M. Sol. Energy Mafer. 1984, 10, 287. (9) Gu, B.; Kiwi, J.; Gratzel, M. N o w . J . Chim. 1985, 9, 539. (10) Ulmann, M.; de Tacconi, N. R.; Augustynski, J. J . Phys. Chem. 1986, 90, 6523.
0 1988 American Chemical Society