Hydrogen-induced titanium oxide migration onto metallic rhodium in

G. Munuera, A. R. Gonzalez-Elipe, J. P. Espinos, J. C. Conesa, J. Soria, and J. Sanz. J. Phys. Chem. , 1987, 91 (27), pp 6625–6628. DOI: 10.1021/j10...
0 downloads 0 Views 557KB Size
J. Phys. Chem. 1987, 91, 6625-6628

6625

Hydrogen-Induced TiO, Migration onto Metallic Rh in Real Rh/TiO, Catalysts G. Munuera,* A. R. GonzPlez-Elipe, J. P. Espinh, J. C. Conesa,+J. Soria,+ and J. Sand Departamento de Quimica Inorginica, Instituto de Ciencias de Materiales, UNSE CSIC, P. 0. Box 1I 15, 41 071 Sevilla, Spain (Received: January 5, 1987)

-

When Rh/Ti02 polycrystalline catalysts deeply reduced by Ar’ bombardment (O/Ti 1.6) are heated at 773 K in vacuo (lo4 Torr),no significantchanges could be observed either in the intensity of the Rh(3d) XPS peak or in the O/Ti stoichiometry. However, a decrease in the Rh(3d) signal (up to 26%) is observed after heating the same sample in 10 Torr of H2 at the above temperature, while the O/Ti ratio rises up to 1.9. After this treatment the intensity of the Rh(3d) can be partially restored by a soft sputtering with Ar’. The above results can be explained by assuming, according to our previous results obtained by EPR and NMR, that hydrogen is incorporated as hydride-like species, Le., at the surface layers of the reduced Ti02 support. The new H-TiO, phase allows a strong metal-support interaction (SMSI state) and shows an enhanced ionic mobility, thus enabling the migration of TiO, moieties onto the Rh particles which may, finally, become encapsulated by the support.

Introduction The set of phenomena characteristics of the SMSI state (suppression of H2 and CO adsorption capacity,’ loss of hydrogenolysis activity,2 etc.) in Rh/Ti02 and other similar catalysts reduced at T 2 773 K is now widely accepted and explained in terms of the so-called “decorating m ~ d e l ” which ,~ assumes migration of TiO, moieties onto the metal particles where they block sites otherwise used for H2/C0 adsorption, hydrogenolysis, etc. This model has been formulated mainly based on evidence gained by working with “ideal systems” formed either by the metal (Pt, Rh, Ni, etc.) evaporated onto the surface of a TiOl single crystal4 (or on a preoxidized Ti foil5) or, alternatively, by using a noble-metal foil (or a single-crystal face) on which Ti was evaporated and then oxidized! Unfortunately, very little direct evidence exists from real catalysts in favor of the model. In previous papers on this subject’-’O we have examined in detail, using IR, NMR, and EPR techniques, the interaction of H2with Rh/Ti02 and other similar real catalysts, and we have concluded that during reduction in H2at T > 573 K hydrogen is taken up by the reduced T i 0 2 support, where it interacts with preexisting Ti3+ ions and becomes stabilized by incorporation into nearby anionic vacancies, thus giving diamagnetic species (Le., (Ti-H)3+) at the surface layers of the reduced support. The noble metal present in these catalysts is able to enhance the reduction and the incorporation of hydrogen atoms into the reduced support by spillover, leading to the hydride-like species according to a twestep mechanism:

-

+ T i 0 2 Rh + H 2 0 + 2Ti3+ + Vo Rh-H + Ti3+ + Vo Rh + (Ti-H)3+

2Rh-H

(1)

-.+

In a more recent work” we have examined the dependence of SMSI effects on this hydride-like species, showing that loss of hydrogen chemisorption capacity by the metal upon high-temperature reduction is parallel to the extensive incorporation of hydrogen into the support (both monitored directly by N M R of metal and support adsorbed hydrogen). Both effects were reversed (a condition essential to the formerly accepted SMSI definition) by high-temperature outgassing while the original intensity of the Ti3+ EPR signal was restored. In addition, we have observed that the reduction temperature at which the decrease in hydrogen adsorption on the metal occurs (SMSI state) does not coincide with the temperature giving the maximum intensity of Ti3+signal in EPR (ca. 573 K) but with a higher temperature at which the diamagnetic (Ti-H)3+ species have been generated; thus, we concluded that the presence of Ti3+ species (Le., reduction of the T i 0 2 support), even associated with anion vacancies, is not enough ‘Instituto de Catllisis y Petroleoquimica, CSIC, c/Serrano 119, 28006 Madrid, Spain. Instituto de Fisico-Quhica Mineral, CSIC, c/Serrano 115, 28006 Madrid, Spain.

*

0022-3654/87/2091-6625$01.50/0

to generate the observed SMSI state, contrary to previous assumptions12that relate the SMSI state to the generation of a new Ti407 phase. The aim of this work is to present some new results, using XPS combined with sputtering methods, that seem to confirm our previous conclusions and that suggest a new role for the hydride-like species incorporated into the TiOz support in connection with the TiO, migration onto the metallic particles.

Experimental Section The Rh/Ti02 polycrystalline sample (2.5% Rh by weight, SBET = 25 f 1 m2 g-l) was the same used by us in previous works7-’’ which has been reduced in H2 at 773 K for 2 h before storage in air ( H R sample). Hydrogen adsorption at 300 K on a H R sample re-reduced in H2 at 773 or 473 K and then outgassed at the same temperatures for 2 h gives H / R h = 0.13 f 0.01, while TEM indicate that rhodium exists in this sample as polyhedral particles with d = 2-10 nm, some of them showing epitaxial growth. Full characterization by IR, NMR, and XRD has been carried out, and complete details have been published el~ewhere.~ Photoelectron spectra were recorded with a Leybold-Heraeus LHS-10 spectrometer working with pass energy at 50 eV. A H-P lOOOE computer on line with the spectrometer allowed the recording and processing of spectra while a Q-200 quadrupole was used to monitor the gas-phase evolution during heating in vacuo of the samples in the main chamber. A straight base line was used for area calculation of Ti(2p), O(ls), and Rh(3d) levels. For all the spectra the binding energy reference (BE) was taken at the Ti(2p) level of Ti4+ at 458.5 eV. The sample was used in the form of pellets (-50 mg) and reduced “in situ” at the pretreatment chamber (-5 X lo3 cm3 volume) of the instrument where it was placed on a molybdenum (1) Tauster, S. J.; Fung, S. C.; Baker, R. T. K.; Horsley, J. A. Science 1981, 211, 1121 and references therein.

(2) Resasco, D. E.; Haller, G. L. In Metal-Support and Metal-Additive Effects in Catalysis; Elsevier: Amsterdam, 1982; p 14. (3) (a) Meriaudeau, P.; Dutel, J.; Dufaux, M.; Naccache, C. Stud. Surf. Sri. C a r d 1982, 11, 95. (b) Santos, J.; Phillips, J.; Dumesic, J. A. J . Catal. 1983, 81, 147. (c) Resasco, D. E.; Haller, G. L. J. Catal. 1983, 82, 279. (4) (a) Sadeghi, H. R.; Henrich, V. E. J . Catal. 1984, 87, 279. (b) Sadeghi, H. R.; Henrich, V. E. Appl. Surf. Sei. 1984, 19, 330. (5) (a) Belton, D. N.; Sun, Y. M.; White, J. M. J . Phys. Chem. 1984,88, 5172. (b) Takatani, S.;Chung, Y.-W. J . Catal. 1984, 90, 75. (6) KO,C. S.; Gorte, R. J. J . C a r d 1984, 90, 59. (7) Conesa, J. C.; Soria, J. J . Phys. Chem. 1982, 86, 1392. (8) Conesa, J. C.; Munuera, G.;Muiioz, A.; Rives, V.; Sanz, J.; Soria, J. In Spillover of Adsorbed Species; Pajouk, G., et al., Ed.; Elsevier: Amsterdam, 1983; p 149. (9) Conesa, J. C.; Malet, P.; Muiioz, A.; Munuera, G.; Sainz, M. T.; Sam, J.; Soria, J. Proc. Int. Congr. Catal., 8th 1984, 5 , 217. (10) Conesa, J. C.; Malet, P.; Munuera, G.; Sanz, J.; Soria, J. J . Phys. Chem. - .. . 1984. - ~ -,88. - -.,2986. -- - (11) Sanz, J.; Rojo, J. M.; Malet, P.; Munuera, G.; Blasco, M. T.; Conesa, J. C.; 6oria. J. J . Phys. Chem. 1985, 89, 5427. (12) (a) Baker, R. T.; Prestridge, E. B.; Garten, R. L. J . Catal. 1979, 56, 390; 1979, 59, 293. (b) Baker, R. T. J . Card. 1980, 63, 523.

0 1987 American Chemical Society

6626 The Journal of Physical Chemistry, Vol. 91, No. 27, 1987

Munuera et al. T i * L T1'3

TI+&

~

Rho

T'2P

470

466

Rh3d

.

462

L58

451

312

316

320

E. E n e r g y ( e V )

308

304

#

8. E n e r g y ( e V 1

m

I

0

5

10

15

-

20

-m / e

25

30

35

40

Figure 1. Mass spectra of a Rh/Ti02 sample heated in vacuo ( Torr) at 273 and 623 K after the standard treatment in H2 at 773 K. Inset: Rh(3d) and Ti(2p) photoelectron spectra of the sample recorded before Torr. (a) and after (b) the heating at 623 K for 1 h at

holder that could be heated resistively while controlling the temperature by a thermocouple placed at the rear of the holder. Otherwise stated, prior to any further manipulation, the sample was heated at 773 K in the preparation chamber, first under vacuum ( 573 K, though now the decay of the m / e 2 peak was much faster while the original shape of the Ti(2p) could not be restored by this treatment. A similar experiment was carried out using a low H2 pressure in order to assess carefully the effect of all the residual gases (e.g., 02,H 2 0 ) on the behavior of the Ti(2p) peak. Figure 2 shows the changes observed after H2introduction in the chamber at lo4 Torr on a fresh sputtered sample. Changes can be clearly observed from the difference of both spectra in this figure. In and 5 this experiment residual O2and H 2 0 pressures (3 X X lo-* Torr, respectively) were much lower (by a factor ca. 20) than the threshold values required to induce similar changes in the Ti(2p) signal when these two gases are in contact with a fresh sample under the same experimental conditions, so that the observed changes in Figure 2 can be unambiguously ascribed in this case to H2adsorption according to step 2 above. As for the sample exposed to 10 Torr of H2, again heating in vacuo of T > 573 K only gives hydrogen evolution, but now it readily decays to the background spectrum in a few minutes, indicating a strong pressure dependence of the incorporation of H, to the reduced T i 0 2 supports. In order to get more information on the origin of the small changes observed in the O/Ti ratios induced by the H2 adsorption, a second set of experiments were carried out on a sample which had been submitted to much deeper sputtering (10 min), which removes ca.40% of the intensity of the Rh(3d) signal and produces a strong reduction of the support (O/Ti ca. 1.6). Figure 3 shows both Ti(2p) and Rh(3d) XPS spectra recorded for this sample submitted to different treatments. It is worthy of note that heating under vacuum Torr) even at 773 K for 30 min does not induce any significant change either in the value of the O/Ti ratio or in the Rh(3d) intensity. However, if the sample is again heated at the same temperature but now under 10 Torr of H2 an important increase of the O/Ti ratio, up to a value of 1.8, is observed while the intensity of the Rh(3d) peak decreases by ca. 26% of its original value. A certain change in the Ti(2p) signal, though much smaller than that shown in Figure 3, could be recorded by heating a similarly deeply sputtered sample at 773 K under 0, or H 2 0at ca. lo6 Torr, so the apparent oxidation measured after the treatment in H2 at 773 K only can be ascribed partially in this case to hydrogen incorporation to the reduced TiO, support as before. A new sputtering for 10 s causes an increase in the

TiO, Migration onto Metallic Rh

,

.

,

.

I

.

,

.

,

.

470 466 462 458 454 --Binding Energy (eV1

I

450

The Journal of Physical Chemistry, Vol. 91, No. 27, 1987 6627

> . . . 316 312 +Binding

. . 308

.

. 304

. 300

E n e r g y (eV1

Figure 3. Ti(2p) and Rh(3d) photoelectron spectra of a Rh/Ti02 sample: (a) after sputtering for 10 min; (b) after heating at lo+' Torr for 1 h; (c) after heating in H2 (10 Torr) at 773 K for 30 min; (d) after a new heating at Torr in vacuo at 773 K; (e) after sputtering with Ar' for 10 s.

intensity of the Rh(3d) peak again up to ca. 82% of its original values as shown in the same figure, indicating some encapsulation of the Rh phase during the previous heating in H2 at 773 K.

Discussion The present results obtained by using a real Rh/Ti02 catalyst confirm the incorporation of hydrogen as such to the reduced Ti02 support and give some new clues to its role in the SMSI phenomenon. In fact, the mass spectra recorded during heating in vacuo a sample previously reduced in H2 at 773 K indicate that only hydrogen is evolved in these conditions while no significant amounts of water remain at the surface. This confirms our previous conclusions, based on N M R experiments,lOJ1on hydrogen uptake by the T i 0 2 support. In fact, the amount of hydrogen evolved and the temperature at which desorption starts agree quite well with our previous data from N M R and TPA.9v'o This behavior had been previously reported for the same type of systems by several authors from TPD TPRIel6 experiments and more recently by Marcelin et a1.,l1 who, using frequency response chemisorption techniques conclude that at least two new forms of hydrogen exist in Rh/Ti02 when compared with Rh/Si02, one of them being clearly related to the T i 0 2 support while the other might be related to the nature of the rhodium-titania interface or to specific rhodium ensembles induced during high-temperature reduction. Incorporation of hydrogen in our sputtered samples even at 300 K, in contrast with the T > 573 K required for the samples directly reduced with hydrogen, can be accounted for by the presence of Ti3+ and Vo at the surface of these samples already at room temperature, while hydrogen reduction of the support, according to step 1 above, must take place under thermal reduction conditions before any incorporation according to step 2 occurs. Therefore, the small changes observed in the Ti(2p) signal when H2was put (14) Kunimori, K.; Uchijima, T. In Spihuer of Adsorbed Species; Pajouk, G., et al., Ed.; Elsevier: Amsterdam, 1983; p 197. (15) Hongli, W.; Sheng, T.; Maosong, X.;Gnoxing, X.;Xiexian, G. i n Metal-Support and Metal-Additive Effects in Catalysis; Elsevier: Amsterdam, 1982; p 19. (16) (a) Beck, D. D.; White, J. M. J . Phys. Chem. 1984, 88, 2764. (b) Beck, D. D.; Bawagan, A. 0.;White, J. M. J . Phys. Chem. 1984.88, 2771. (17) Marcelin, G.; Lester, J. E.; Mitchell, S . F. J . Catal. 1986, 102,240.

in contact with a sputtered surface (at least in Figure 2) should be ascribed to generation of the hydride-like species (Ti-H)3+ according to step 2. Formation of such (Ti-H)3+ hydride-like species at 300 K had been previously assumed by Gopel et a1.18 to explain changes in conductivity and surface potential observed in T i 0 2 rutile single crystals with Ti3+ defects after exposure to low pressures of H2. Incorporation of hydrogen as (Ti-H)3+ into the polycrystalline T i 0 2 support of our samples, when they are heated in hydrogen a t T > 573 K, has been previously suggested by us on the basis of EPR results to explain the observed decrease in the Ti3+ EPR signal and its recovering upon outgassing at higher temperatures in terms of reversibility of step 2. In our view, this is the primary phenomenon previous to any other process (Le., decoration, alloying, encapsulation, etc.) that produces a loss of adsorption capacity on the rhodium particles as we have previously observed by N M R for hydrogen adsorptionlo*" or by IR for the case of the adsorption of C0.19 In both cases, the intensity of the ' H N M R line and the IR band due to hydrogen or CO adsorbed on the metallic particles is very low when hydrogen is present in the T i 0 2 support, after heating the Rh/Ti02 sample at 773 K in H2, while they grow when the adsorption is carried out on a sample which has been previously outgassed at 773 K to remove the incorporated hydrogen. In addition, the increase in the O/Ti ratio from 1.6 to 1.8 in Figure 3c points to a simultaneous enhancement of the ionic mobility at the titanium-oxygen lattice induced by the presence of hydrogen, which is detected even at room temperature provided that a direct reoxidation by O2or H 2 0 should be ruled out in the conditions used in the experiment shown in Figure 2. In fact, a complete rearrangement between the TiO, surface layers and the bulk would occur at the higher reduction temperatures probably up to equilibration of the chemical composition within the T i 0 2 particles. However, this high mobility of the reduced T i 0 2 phase only occurs in the presence of H2 since no significant modifications in the intensity and shape of the Rh(3d) and Ti(2p) peaks could be detected after heating the sputtered sample at 773 K in vacuo (Figure 2b); thus, we discard the possibility of Ti3+ and/or Vo sites as responsible for the generation of the SMSI state, in agreement with our previous conclusions on the basis of EPR studies.I0 Though enhancement of the metal-support relative mobilities has been previously assumed by Baker et a1.12 to account for the spreading of the metallic particles in Pt/Ti02 after heating at 773 K in hydrogen to form thin, flat "pillbox" structures (observed by TEM), this phenomenon was interpreted in terms of a tendency to increase the interaction at the metalsupport interface as a result of a higher capacity of wetting between the metal and the reduced Ti407phase. Our results in Figure 3 clearly suggest that this could only be possible when hydride-like species are incorporated into the reduced T i 0 2 phase. This enhanced mobility should be, in fact, responsible for the constant O/Ti ratio measured for the Rh/Ti02 samples thermally reduced in H2 at 773 K (Figure 1) as well as for the decoration of the rhodium particles with TiO, moieties (see below) that, as previously suggested by us,ll should be better described as H-TiO, (x < 2). Formation of a H-TiO, phase, which must occur extensively upon heating our Rh/Ti02 catalyst at 773 K in H2, is then the main factor contributing to the generation of the reversible SMSI state (H, and CO adsorption suppression which can be restored just by removal of the hydride-like species) previously reported by us,11,19though, in addition, it also seems to contribute to the migration of the TiO, moieties onto the metal particles, a process that should be then considered as an intermediate step in the irreversible encapsulation (or alloying) of the metal particles. In this context it is worth comparing the more recent EXAFS results for Rh/Ti02 catalysts. While Prins et al.,20using a sample (18) Gopel, W.; Rocker, G.; Feirabend, R. Phys. Rev. B Condens. Matter 1983, 28, 3421. (19) Mufioz, A,; Gonzilez-Elipe, A. R.; Munuera, G.; Espinbs, J. P.; Rives-Arnau, V. Spectrochim. Acta, Part A, in press. (20) Konisberger, D. C.; Martens, J. H.; Prins, R.; Short, R. D.; Sayers, D. E. J . Phys. Chem. 1986, 90, 3047.

6628

The Journal of Physical Chemistry, Vol, 91, No. 27, 1987

just reduced at 673 K, only observe new Ti-Rh interactions at 3.47 A, without any evidence of coverage of the metal particles by a suboxide of Ti02, Haller et aL2' claim, using samples deeply reduced in H2 at 773 K, the generation of an alloy with new short Ti-Rh bonds at 2.52 A, which they explain on the basis of TiO, migration. Both results can be conciliated, according to our results, if in the former case, where a rather mild reduction was carried out, only incorporation of hydride-like species to the support according to steps 1 and 2 has taken place, thus only increasing the interaction at the interface between the rhodium and the new H-TiO, phase at the surface of the support (with generation of new Ti-Rh bonds a t 3.47 A), while in Haller's experiment the deeper reduction in hydrogen at 773 K might allow the H-TiO, moieties to migrate onto the rhodium particles and even to be further reduced to give an alloy phase. Similarly, the strong decrease (ca. 26%) in the Rh(3d) signal intensity in Figure 3c can be explained by assuming that decoration of the metallic particles by TiO, species (actually H-TiO,) also has taken place in our severe conditions, thus attenuating the photoelectrons coming from the rhodium. A rough estimation of the average thickness of the decorating layer, taking the mean free path for the Rh(3d) electrons through the titanium dioxide gives a value of ca. 9 A, thus indicating a strong as 29.5 encapsulation of the metal by the H-TiO, phase in the deep conditions used in our experiment. The screening of the rhodium is unambiguously confirmed by the partial recovering of the original intensity of the peak after sputtering for a few seconds, which will remove the TiO, skin from the surface of the rhodium particles. In order to avoid the need of invoking long-range interactions between reduced centers at the T i 0 2 support and surface metal atoms of large metal particles (existing in our low dispersed catalysts), we had assumed" that under SMSI conditions Ti-H bond should appear not only on the titania support but also within the highly reduced TiO, moieties decorating the metal surface. In fact, the recent results by Marcelin et al.,I7 mentioned above, can be interpreted by assuming that the second type of hydrogen species observed by these authors on R h / T i 0 2 correspond to hydride-like species incorporated with the TiO, moieties decorating the metallic phase. It is worth noting that in a recent work Somorjai et al.23have reported that when T i 0 2 is evaporated onto a Rh foil at low coverages, CO adsorption becomes depleted to an extent which is even higher than that expected for a simple physical blocking of the surface sites, a conclusion which is e ~ ~ similar contrary to previous results by KO and G ~ r t using systems. However, the main point in Somorjai's work in relation to our resultsI9 lies in the fact that reduction of the T i 0 2 with 50 Torr of hydrogen enhances this loss of capacity for CO adsorption, (21) Sakellson, S.;McMillan, M.; Haller, G. L. J . Phys. Chem. 1986.90, 1733. (22) Seah, M. P.; Dench, W. A. Surf. Interface Anal. 1979, 1 , 2. (23) Levin, M.; Salmeron, M.; Bell, A. T.; Somorjai, G. A. SurJ Sei. 1986, 169, 123. (24) KO,C. S.; Gorte, R. J. Surf. Sci. 1985, 161, 597.

Munuera et al. as could be expected for a higher electronic interaction of H-TiO, moieties as compared with TiO,, a situation that cannot be obe ~ ~their samples had only served in the case of KOand G ~ r t since been heated under IO-' Torr of H2 at 650 K and then flashed to 1400 K, thus probably leading to the removal of any incorporated hydrogen. In fact, if the influence of the H-TiO, species on the metal properties is stronger than that of the non-hybrided TiO, species that would remain after outgassing,l' a partial reversal of the SMSI effects upon outgassing or oxidation would then have to be observed in these "ideal" systems because of the different nature of the electronic interactions involved. Unfortunately, neither in Somorjai's work nor in most of the experiments using these types of "ideal models" are reversibility tests normally carried out, a condition which is essential to the original SMSI definition.'

Conclusions Hydrogen incorporation into reduced T i 0 2 cannot be considered as a side phenomenon in the SMSI interaction but a fact of paramount importance in SMSI generation. In this sense it is worth noting that hydrogen is also incorporated into other supports such as V205,25Ce02,26or M o o t 7 on which some kind of SMSI state has been r e p ~ r t e d . ' > ~ * - ~ ~ From our results here and in previous it can be concluded that reduction of the T i 0 2 support to a Ti407phase cannot account, by itself, for the SMSI interaction in Rh/Ti02 catalysts. However, the incorporation of hydrogen, in the form of the (Ti-H)3+ species, gives a surface H-TiO, phase which allows a strong interaction with the metal leading to a SMSI state" reversible upon removal of this hydrogen. Meanwhile, the new H-TiO, phase shows a much greater mobility which favors the migration of H-TiO, moieties onto the rhodium particles (partial encapsulation). The decorating H-TiO, species should have a strong electronic interaction with the metal and may fully suppress H 2 / C 0 adsorption on the Rh particles, even before their total encapsulation occurs. In these conditions only a partial reversibility of the SMSI state should be expected upon outgassing depending on the relative interactions of the H-TiO, and TiO, moieties with the metal.

Acknowledgment. We thank the CAYCIT (Project 0234/84), the CSIC (Project 552), and the "FundaciBn RamBn Areces" for financial support. Registry No. Rh, 7440-16-6; Ti02, 13463-67-7; H,, 1333-74-0; Art, 14791-69-6. (25) Tinet, D.; Friplat, J. J. Reu. Chim. Miner. 1982, 19, 612. (26) Sanz, J.; Garcia-Fierro, J. L.; Soria, J.; Rojo, J. Z . Phys. Chem. (Munich) 1987, 152, 83. (27) Cirillo, A. C.; Ryan, L.; Gerstein, B. C.; Fripiat, J. J. Chem. Phys. 1980. 73. 3060. (28) Lin, Y.-J.; Resasco, D. E.; Haller, G. L. J . Chem. SOC.,Faraday Trans. 1 1987.83, 2091. (29) Meriaudeau, P.; Dutel, J. F.; Dufaux, M.; Naccache, C. Stud. Surf Sci. Catal. 1982, 1 1 , 95. (30) Meriadeau, P.; Albano, K.; Naccache, C. J. Chem. SOC.,Faraday Trans. 1 1987, 83, 21 13.