Oxygen-Induced Reconstruction and Surface Oxidation of Rhodium

Nanometric chemical clocks. J.-S. McEwen , P. Gaspard , T. V. de Bocarme , N. Kruse. Proceedings of the National Academy of Sciences 2009 106 (9), 300...
0 downloads 0 Views 190KB Size
Langmuir 1998, 14, 6151-6157

6151

Oxygen-Induced Reconstruction and Surface Oxidation of Rhodium V. K. Medvedev,† Yu. Suchorski,‡ C. Voss,† T. Visart de Bocarme´,† T. Ba¨r,† and N. Kruse*,† Chemical Physics at Surfaces and Heterogeneous Catalysis, Free University of Brussels, Campus Plaine, CP 243, B-1050 Brussels, Belgium, and Institute for Physical Chemistry, University of Hannover, Callinstrasse 3-3a, D-30167, Germany Received May 22, 1998. In Final Form: July 31, 1998 The reaction of oxygen with Rh single-crystal tips was investigated using field ion microscopy (FIM). Emphasis was laid on revealing the atomic structure of individual tip surface planes along with their influence on the early stages of the reaction, i.e., chemisorption and nucleation. Under field-free conditions, the interaction of oxygen with a (001) oriented Rh tip was found to lead to major reconstructions at temperatures above 400 K. While the original shape of the tip was nearly hemispherical before the reaction, it was polyhedral thereafter. In particular, the {011} and {113} planes were seen to adopt a “missing row” type reconstruction in the presence of oxygen adsorbate. On the Rh{113} plane, a (1 × 3) reconstruction prevailed for oxygen exposures larger than 60 L (1 L ) 1.3 × 10-4 Pa‚s) and temperatures T g 550 K. In this type of missing-row reconstruction two adjacent dense-packed chains of atoms are absent. Surface oxidation was found to be promoted by the presence of an external electric field of ∼15 V/nm. Studies as a function of the surface temperature were performed in real time by video-FIM leading to the observation of a strong local variation in the oxidation activity. While the {113} planes and the vicinals of the (111) pole turned out to undergo rapid oxidation, the flat planes with {001} and {111} symmetry remained rather inactive at temperatures between 350 and 483 K. The formation of surface granules with sizes of ∼5-10 nm in areas of high oxidation activity was interpreted as being due to a nucleation process forming RhxOy precipitates. Surface granules could be easily removed by reaction with CO gas, meaning that only the topmost layers of the Rh tip were involved in surface oxidation.

1. Introduction Studies of the early stages of metal oxidation have a long-standing history in surface science. Although there is general agreement that the mechanistic steps leading to oxide formation involve dissociative oxygen chemisorption on the metal surface, lattice penetration of atomic oxygen, nucleii formation of suboxides, and their growth to amorphous aggregates finally crystallizing in stoichiometric oxide phases, no standardized model could be worked out so far. On the basis of kinetic studies it soon became clear that the sequence of events may be complex. Rather than successively, these events may occur simultaneously, depending on temperature, pressures, and, of course, the chemical nature of the metal itself. A thorough review of the state-of-the-art along with a number of case studies was given some years ago.1 The chemisorption of oxygen on both Rh single crystals and polycrystalline samples was the subject of intensive research,2-24 but only a few studies have addressed the question for oxide nucleation and growth. A significant † ‡

Free University of Brussels. University of Hannover.

(1) Brundle, C. R.; Broughten, J. Q. In The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis; King, D. A., Woodruff, D. P., Eds.; 1990; Vol. 3, p 131. (2) Tucker, C. W. J. Appl. Phys 1966, 37, 4147; 1967, 38, 2696. (3) Tucker, C. W. Acta Met. 1967, 15, 1465. (4) Campbell, C. T.; White, J. M. J. Catal. 1978, 54, 289. (5) Campbell, C. T.; Shi, S. K.; White, J. M. Appl. Surf. Sci. 1979, 2, 382. (6) Castner, D. G., Sexton B. A.; Somorjai, G. A. Surf. Sci. 1978, 71, 519. (7) Castner, D. G.; Somorjai, G. A. Surf. Sci. 1979, 83, 60. (8) Castner, D. G., Sexton, B. A.; Somorjai, G. A. Appl. Surf. Sci. 1980, 6, 29. (9) Comelli, G.; Dhanak, V. R.; Kiskinova, M.; Pangher, N.; Paolucci, G.; Prince, K. C.; Rosei, R. Surf. Sci. 1992, 260, 7.

point in the development toward a better understanding of the oxide growth kinetics along with the compositional analysis marks the paper by Kellogg.17 Using small Rh crystal tips in a combined study by field ion microscopy (FIM) and imaging atom-probe mass spectrometry, Kellogg determined an activation energy as low as 13 kJ/mol for the early stages of oxide growth at temperatures between 400 and 650 K. He also reported on stoichiometric Rh2O3 formation in the millibar pressure range and at temperatures of 500 K. The motivation for the present study is partly given by our interest in the use of Rh as a catalyst for automotive pollution control.25 On one hand, Rh metal is active in (10) Dhanak, V. R.; Comelli, G.; Cautero, G.; Paolucci, G.; Prince; K. C.; Kiskinova, M.; Rosei, R. Chem. Phys. Lett. 1992, 188, 237. (11) Leibsle, F. M.; Murray, P. W.; Francis, S. M.; Thornton, G.; Bowker, M. Nature 1993, 363, 706. (12) Murray, P. W.; Leibsle, F. M.; Li, Y.; Guo, Q.; Bowker, M.; Thornton, G.; Dhanak, V. R.; Prince, K. C.; Rosei, R. Phys. Rev. 1993, B47, 12976. (13) Comicioli, C.; Dhanak, V. R.; Comelli, G.; Astaldi, C.; Prince, K. C.; Rosei, R.; Atrei, A.; Zanazzi, E. Chem. Phys. Lett. 1993, 214, 438. (14) Dhanak, V. R.; Prince, K. C.; Rosei, R.; Murray, P. W.; Leibsle, F. M.; Bowker, M.; Thornton, G. Phys. Rev. 1994, B49, 5585. (15) Kellogg, G. L. J. Catal. 1985, 92, 167. (16) Kellogg, G. L. Phys. Rev. Lett. 1985, 54, 82. (17) Kellogg, G. L. Surf. Sci. 1986, 171, 359. (18) Gierer, M.; Over, H.; Ertl, G.; Wolhgemuth, H.; Schwarz, E.; Chistmann, K. Surf. Sci. Lett. 1993, 297, L73. (19) Farias, D.; Tro¨ger, H.; Rieder, K. H. Surf. Sci. 1995, 331-333, 150. (20) Gorodetskii, V. V.; Nieuwenhuys, B. E.; Sachtler, W. M. H.; Boreskov, G. K. Appl. Surf. Sci. 1981, 7, 355. (21) Janssen, N. M. H.; van Tol, M. F. H.; Nieuwenhuys, B. E. Appl. Surf. Sci. 1994, 74, 1. (22) Rebholz, M.; Prins, R.; Kruse, N. Surf. Sci. 1992, 269/270, 293. (23) Voss, C.; Gaussman, A.; Kruse, N. Appl Surf. Sci. 1993, 67, 142. (24) Voss, C.; Kruse, N. Surf. Sci. 1998, 409, 252. (25) Taylor, K. C. In CatalysissScience and Technology; Anderson, J. R., Boudart, M., Eds.; Springer: Berlin, 1984; Vol. 5.

S0743-7463(98)00603-9 CCC: $15.00 © 1998 American Chemical Society Published on Web 09/24/1998

6152 Langmuir, Vol. 14, No. 21, 1998

both CO (or hydrocarbon) oxidation and NO reduction. In addition, it can act as a storage for oxygen under lean driving conditions. The process is reversible on returning to net reducing conditions so that the catalyst is maintained in a state of continual high activity for CO oxidation. On the other hand, considerable catalytic deactivation was observed once the formation of thick Rh-oxide layers followed simple oxygen storage.15 Crucial stages preceding the growth of an oxide layer are nucleation and precipitation, for which only limited information is available at the moment. The present study will address the kinetics of these processes and demonstrate correlations with the underlying surface structure. Similar to the approach by Kellogg,16 controlled reaction conditions were applied using Rh crystal tips as models for single catalyst grains. Both tips and grains simultaneously expose a large number of crystallographically different planes with sizes on the nanometer scale. Micrographs of a real time video study will be presented, revealing that surface oxidation of rhodium is promoted by the presence of an external electric field. 2. Experimental Section The experiments were performed in two different setups previously described in detail elsewhere.26,27 In brief, they both present bakeable all-metal UHV systems with base pressures of 10-8 Pa and below. Channel plates, either in a fixed position or on a movable rod, allow for image intensification. Both setups could be operated either in the traditional FIM mode using Ne and H2 as imaging gases or in oxygen gas serving as reactant and imaging gas at the same time. Individual micrographs of clean and oxygen-covered tips were taken by a CCD camera (OMA Vision, EG&G Princeton Appl. Res., 512 × 512 pixel, dynamic range ) 18 bit). Real time studies of the oxidation were performed by making use of video-FIM. The surface oxidation was investigated by following local changes of the image brightness at stabilized temperatures ranging from 60 to 1000 K. For this, individual snapshots of the video sequence (time resolution 20/40 ms, depending on the system) were digitized in order to determine the brightness over a well-defined number of pixels, i.e., in local areas of the image. Rh tips were prepared by electrochemically etching a thin wire (φ ) 0.1 mm, 99.99% purity, Johnson Matthey) in a molten mixture (1:4 w/w) of NaCl and NaNO3. Either (001) or (111) oriented samples were obtained in this manner. The samples were cleaned by cycles of low-temperature field evaporation and Ne sputtering at elevated temperatures (750-800 K). Both oxygen (Messer Griesheim, 99.999 vol %) and carbon monoxide (Messer Griesheim, 99.997 vol %) were used without further purification.

3. Results and Discussion The paper is organized in the following manner. We first present data of the field-free interaction of oxygen with Rh tips. Both {011} and {113} crystallographic planes will be inspected in detail. For these studies “classical” FIM was used in order to image the surface reconstruction by field ionization of either Ne or H2 at cryogenic temperatures. We next move to a real-time investigation of the early stages of the field-promoted Rh oxidation at elevated temperatures. Special consideration will be given to surface morphological changes in local areas of the tip including {011}, {113}, {012}, {001}, and (111). The reversibility of the oxidation process will be demonstrated by using a titration technique with CO gas. (i) Oxygen-Induced Restructuring of Rh: General Features and Local Plane Behavior. The FIM ex(26) Gaussmann, A.; Kruse, N. Catal. Lett. 1991, 10, 305. (27) Medvedev, V. K.; Suchorski, Yu.; Block, J. H. Vacuum 1995, 46, 563.

Medvedev et al.

periments to be presented here were performed by exposing a clean [001] oriented Rh tip to oxygen gas at temperatures between 400 and 600 K. This resulted in structural changes of the specimen ultimately associated with morphological shape transformations. Figure 1 exemplifies the restructuring observed after an exposure to 60 L of oxygen (whereby 1 L ) 1.3 × 10-4 Pa‚s) at 550 K. While the clean tip is nearly hemispherical (radius of curvature ∼12 nm, Figure 1a) and neatly divided into a number of high index planes located along the various zone lines between the central (001) pole and the peripheral planes, the end form (imaged in H2 or Ne at different field strengths, Figure 1b-d) resembles a polyhedron. The shape transformation is associated with a strong coarsening of the crystal; i.e., facet structures are formed. Since a detailed account of these observations was recently given elsewhere,24 we concentrate here on local plane reconstructions (see marked areas in Figure 1a). Interestingly, the local reconstruction forms on certain planes seem to be rather independent of the amount of gas dosed to the surface. We mention that exposures of 0.5 L, equivalent to coverages of 0.1-0.2 monolayer, are sufficient to provoke structural changes.28 Parts b and d of Figure 2 demonstrate the occurrence of (1 × 2) and (1 × 3) missing-row type reconstructions on Rh{011} and Rh{113}, respectively. Thus, as compared to clean Rh (Figure 2a,c), only every second or third chain of dense-packed atoms appears to be preserved in the presence of adsorbed oxygen (Figure 2b,d). Note that the distances between atomic chains in the {133} vicinals have not changed at all. However, there seem to be extraatoms in neighboring planes of the 〈100〉 zone lines so that we conclude that the removal of Rh atoms from {011} (and {113}) involves surface diffusion along the troughs.29 The restructuring of the {011} and {113} planes is contrasted by the behavior of the (001) pole. As seen in Figure 2e,f, only the layer edge has visibly changed its structure in the presence of adsorbed oxygen; i.e., a smoothening of the border has taken place. The enhanced mobility of Rh layer edge atoms in the presence of adsorbed oxygen may eventually lead to their liberation and diffusion into terrace regions (see arrows in Figure 2f). Surface reconstructions might be suspected to easily suffer damage while imaging in the presence of a high electric field. However, as shown in Figure 1b,c, field evaporation is only minor when using hydrogen as imaging gas. No displacement of image dots is seen during H2 field ionization when increasing the viewing field from 18.8 to 20.1 V/nm (Figure 1b,c). On the other hand, if Ne is used as imaging gas, some field evaporation may occur and lead to the gradual destruction of the surface reconstruction. Notwithstanding, the missing-row type reconstructions of the {011} and {113} planes are still clearly discernible at a viewing field of 35.5 V/nm (Figure 1d, with the respective planes in the center of the frames). More generally, experiments with further increasing electric fields demonstrated that more than only one layer had to be removed in order to reestablish the original tip morphology. This is expected in view of the rather open facet structures present in Figures 1 and 2. (28) Visart de Bocarme´, T. Diploma work, Free University of Brussels, 1997. (29) One of the reviewers has raised the question of the formation mechanism of missing-row structures on Rh tip surfaces. Accordingly, while the removal of atomic chains seems to be the most straightforward mechanism, buildup by adding atomic chains cannot be excluded a priori. In most of our experiments, however, we have observed that the size of the planes did not change substantially. Recent experiments allowed us to follow the field-assisted removal of individual atoms along the [100] direction of the (011) plane.

Reconstruction and Surface Oxidation of Rhodium

Langmuir, Vol. 14, No. 21, 1998 6153

Figure 1. (a) Field ion micrograph of a clean (001)-oriented Rh tip imaged in Ne at 60 K and 37 V/nm. The arrow shows a surface defect on the (135) surface plane. (b) Same Rh sample after field-free adsorption of 60 L oxygen at 550 K. The image was taken in hydrogen at 150 K and 18.8 V/nm. (c) Same micrograph as in part b at 20.1 V/nm. (d) FIM of the reconstructed tip imaged with neon at 35.5 V/nm.

It should be mentioned that only the substrate Rh atoms or, possibly, Rh-O complexes, contribute to the FI micrographs shown in Figures 1 and 2; i.e., the oxygen adsorbate remains invisible. On the other hand, this allows to straigthforwardly interpret the micrographs in terms of surface structural changes induced by the presence of adsorbed oxygen. The (1 × 3) Rh{113} plane was assembled using ball models. As seen in Figure 3, the most remarkable feature of this reconstruction form is the absence of two adjacent dense packed atomic chains. The facets are of {001} and {111} symmetry. Some ambiguity remains as to the existence of an atomic chain along the trough in the thirdlayer position. The local field strength is too low for imaging in-trough features. We also note that the reported local reconstruction forms were not observed at temperatures below 400 K. Thus we conclude that the process is thermally activated. Moreover, the detailed temperature dependence demonstrated that the (1 × 3) reconstruction of the Rh{113} plane dominates at temperatures T g 500 K while mixtures of (1 × 2) and (1 × 3) were observed between 400 and 500 K. Also the Rh{011} plane was occasionally

observed to reveal (1 × 3) besides (mainly) (1 × 2) reconstruction. It has been shown here that the reconstruction forms of field emitter planes are similar to those of 2D single crystals of the same orientation.10,14 This is remarkable because the difference in size is ∼12 orders of magnitude. From a more general point of view, oxygen adsorption and surface reconstruction have to be considered as steps preceding the Rh oxidation process. Interestingly, Kellogg,16 in an imaging atom-probe study, found onset temperatures for oxidation (at pressures in the millibar range and less) which are similar to those in the present work. Recently, an oxidation process in the presence of an electric field was considered important in studies of the CO/O2 surface reaction.27,30-32 (ii) Morphological Changes during Rh Oxidation in the Presence of an Electrical Field. The oxidation studies to be reported here were performed in an in-situ (30) Medvedev, V. K.; Suchorski, Yu.; Block, J. H. Surf. Sci. 1995, 343, 169. (31) Medvedev, V. K.; Suchorski, Yu.; Block, J. H. Appl. Surf. Sci. 1996, 94/95, 200. (32) Suchorski, Yu.; Imbihl, R.; Medvedev, V. K. Surf. Sci. 1998, 401, 392.

6154 Langmuir, Vol. 14, No. 21, 1998

Medvedev et al.

Figure 3. Ball model for the (1 × 3) missing-row reconstruction of Rh(113). The presence of a Rh atomic chain in the third layer trough position is arbitrary.

Figure 2. (a, c, and e) Images of the clean surface planes (011), (113), and (001), respectively, as marked in Figure 1a. (b, d, and f) The same planes after oxygen adsorption at 550 K.

manner, i.e., during the ongoing reaction. Accordingly, oxygen gas was continuously dosed to the surface while recording the FIM images in a video. The experimental procedure leading to Figure 4a-f consisted in dosing oxygen gas to a [111] oriented Rh tip precovered by CO adsorbate. In this manner we could easily avoid the usual delay times associated with the replacement of one imaging gas (Ne or H2) by another (O2). The beginning of our experiments is thus defined as the time when switching the gas valve from a mixture of CO-O2 to O2 only. Another advantage of this procedure is that the changes in image brightness associated with oxidation and reduction (using CO at the end of the experiments, see below) can be easily compared. Figure 4a shows an O2+-FIM image ∼5 s after introducing O2 only at a gas pressure of 5.2 × 10-5 mbar into the reaction chamber while keeping the Rh tip at 375 K. We note that the radius of curvature of the tip is ∼170 nm. The lateral resolution is ∼1 nm and, thus, lower than in classic FIM investigations. This is mainly due to the relatively high temperature during imaging and, to a lesser extent, to the use of O2 as imaging gas (rather than H2 or Ne as in Figures 1 and 2). Figure 4b-e demonstrates that the continuous interaction of oxygen in the presence of an external field strength of 15 V/nm causes local coarsening of the Rh tip surface and, finally, granule formation with granule sizes between 5 and 10 nm. Quite obviously, the process is associated with changes in the image brightness, i.e., the O2+ field ionization rate. The sequence of images is interpreted as

being due to the beginning of surface oxidation leading to RhxOy aggregates. The direct comparison between Figure 4a and Figure 4b-e demonstrates plane-to-plane variations in the rate of the oxidation process. In particular, the stepped regions between the low-index poles appear with increasing brightness; i.e., they are most susceptible to oxidation. This is different from the behavior of the (111) central plane and the {001} peripheral planes which seem to be rather reluctant to oxidation. Only after ∼4 min of continued interaction with oxygen (Figure 4e), the granulation process has also involved the (111) pole while the {001} peripheral planes remain dim throughout the experiments. Quite interestingly, the {011} planes seem to undergo slower oxidation than the {113} planes. Following the results of previous field-free measurements, both planes are not expected to be reconstructed at 375 K. Figure 4c-e demonstrates the appearance of a rather bright collar in the vicinals of the (111) pole. We suspect the intensive oxidation of these areas to be correlated to the presence of a high number density of kink sites. A careful inspection of the video sequence demonstrated that this bright collar slowly spreads out toward the center of the (111) plane. Thus we conclude that a diffusion process is in operation. It might be speculated that RhxOy cluster species are liberated in favorable positions of the vicinals (like kinks) and move toward the central (111) plane. Consequently, the intrinsic oxidation rate of the (111) plane, in the absence of any diffusion, must be lower than judged from the mere occurrence of granules in Figure 4e. Interestingly, the Rh-surface oxidation process immediately stopped when reintroducing CO gas at 10-5 mbar into the reaction chamber while continuously imaging with O2+ (Figure 4f). During a few seconds only, the original image shown in Figure 4a was obtained again (except for some small bright spots persisting for slightly

Reconstruction and Surface Oxidation of Rhodium

Figure 4. (a) O2+-field ion image of the initial stage of the oxygen interaction with Rh at 375 K. (b-e) O2+-field ion micrographs showing the sequential stages of the interaction of oxygen at 5.2 × 10-5 mbar in the presence of an external field of 15 V/nm; (f) O2+-field ion micrograph showing the result of reducing the oxidized surface by carbon monoxide in the presence of an external field of 15 V/nm.

longer times). Obviously, the previously formed surface oxide reacted off to give CO2 gas. To study the time dependence of the oxidation process in different Rh planes, the video sequence was evaluated for the local O2+ field ionization rates in the (111), (001), (011), (113), and (012) planes. This is demonstrated in Figure 5, parts a and b. One clearly recognizes the occurrence of a considerable time delay preceding the onset of local brightnesses. This is due to the replacement of COad (desorbing as CO2) by Oad. Quite interestingly, this process is associated with a slight decrease of the overall brightness which is in contrast to the behavior observed in studies of the CO oxidation on Pt where Oad regions were found to be imaged selectively by O2+, yielding high brightnesses. The reason for the behavior of Pt as compared to Rh may be found in the different work function changes resulting on CO and O2 adsorption. Details are discussed in ref 32. The time dependence of the surface oxidation shown in Figure 5, parts a and b, demonstrates the high activity of the stepped planes {113} and {012}. In contrast, the lowindex planes are much less active while (011) presents an intermediate case. It is also seen in both figures that the strong initial changes of the brightness with time in open planes are followed by a period of lessening slopes eventually leading to leveling. This observation must be interpreted as being due to increasing kinetic constraints

Langmuir, Vol. 14, No. 21, 1998 6155

during surface oxidation. On the basis of the model that the surface granulation is associated with the formation of RhxOy cluster compounds, it follows that the crystallization into extended phases with (close to) stoichiometric oxide composition is rather slow under our conditions. In comparing the characteristic initial slopes of image brightness as a function of time (Figure 5a,b), it becomes clear that the surface oxidation becomes faster with increasing temperature. The continuously increasing brightness on the (111) plane at 375 K is possibly due to the influence of a diffusion process (see also above, Figure 4) by which RhxOy cluster species move from the vicinals toward the central pole. The surface oxide layer formed at the end of the experiments could be easily reacted off in the presence of CO gas. According to Figure 5, parts a and b, the process is faster than both the initial surface oxidation and the reverse process of replacing COad by Oad (induction period at short times). After a short time of CO interaction with the surface, the original brightness was reestablished again. The asymmetric behavior in the time response to changes in the CO-O2 gas-phase composition is in accordance with results reported on the CO oxidation over noble metal surfaces.33 To gain insight into the kinetics of the oxidation process the reaction temperature was varied between 350 and 483 K under otherwise identical experimental conditions. As seen in Figure 6a-d, the local image brightnesses obtained after a reaction time of ∼200 s differ considerably. While the images at low temperatures, 350-375 K, exhibit a relatively homogeneous brightness with small granulation in regions of high oxidation activity, strong variations of the brightness along with the formation of larger granules are seen at high temperatures. At 483 K, Figure 6d, the granulation in the (111) vicinals is most clearly developed though less extended as compared to the measurement at 417 K (Figure 6c). This observation must be attributed at least partly to the occurrence of field evaporation/desorption, i.e., the removal of RhxOy cluster species while imaging in oxygen during the ongoing reaction. During the course of our studies on the Rh/ oxygen system,23 we have observed that field evaporation does not follow the usual mechanism of “shrinking planes” by removal of species from the layer edge. Instead the process occurs rather irregularly leading to local disorder. We emphasize that the loss of material due to field evaporation/desorption is definitely absent in Figure 6a,b ensuring the correct evaluation of the time-dependent image brightnesses leading to Figure 5a,b. It should also be mentioned here that we observed an immediate increase of the image brightness in the (111) plane when decreasing the sample temperature from 483 to 350 K. After a transition time of ∼5 s, the resulting image was similar to the one shown in Figure 6a. This behavior can be traced back to the presence of considerable local disorder left behind by irregular field evaporation during the preceding experiments at 483 K as described above. It can be easily conceived that a disordered surface provides nucleation centers for the formation of larger granules. Previous FIM work has verified that only a few nucleation centers appear in the respective images after field-free reaction with oxygen at comparable conditions of temperature and pressure,17 in agreement with studies performed for macroscopic Rh crystals.34 Thus we are led to the conclusion that the presence of an electric field of (33) Gorodetskii, V.; Drachsel, W.; Block, J. H. Catal. Lett. 1993, 19, 223.

6156 Langmuir, Vol. 14, No. 21, 1998

Figure 5. (a) Time-dependence of the local O2+-FIM image intensity for (111), (001), (011), (113), and (012) planes. Video sequence was taken in oxygen at 350 K and F ) 15 V/nm. (b) The same as part a, but at 375 K.

Medvedev et al.

the formation of small granules is ∼350 K. The long-time exposure to oxygen under field-free conditions has never led to comparable results.25 Thus we may argue that the presence of an electric field leads to a decrease of the activation barrier for surface oxidation. Assuming pseudo-first-order reaction kinetics, respective rate constants can be estimated from the initial slopes of brightness vs time (Figure 5a,b). We find values of 1.7 s-1 at 350 K and 2.3-2.4 s-1 at 375 K for both (113) and (012) planes. Respective values for (101) planes are smaller but uncertain due to the large scatter of data. A crude estimation of the activation energies leads to values in the range of 5-17 kJ/mol. We feel, however, that a quantitative evaluation of the kinetics necessitates time dependent measurements over a broad range of temperatures in the strict absence of concomitant field evaporation, which is difficult to achieve. We finish this part of the paper by a few more general remarks of studying local effects of the Rh oxidation process by means of FIM. First, a clear-cut differentiation between plane specificity and local variations of the field strength is not always possible. For example, the (local) electric field in more open planes is larger than on closepacked (111) and {001} (for a more detailed discussion of local field strengths see refs 35-36. Thus cross-comparisons on the basis of absolute rather than relative image intensities should only be made for planes of similar inherent roughness. Moreover, the coarsening of the surface during oxidation (granules formation) leads to an increase of the local field strength so that the process tends to self-acceleration. Despite these problems, we think that the careful evaluation of local changes in the surface coarsening and granules formation as followed by real time video-FIM provides interesting information on the initial stages of metal surface oxidation. 4. Comparison with Literature and Conclusions

Figure 6. O2+-field ion images of the Rh surface after an exposure to oxygen for ∼200 s with oxygen at 350 (a), 375 (b), 417 (c), and 483 K (d).

15 V/nm, as applied in the present study, remarkably enhances the Rh surface oxidation rate. We have found in the course of our studies that the onset temperature for (34) Thiel, J. H.; Yates, J. T.; Weinberg, W. H. Surf. Sci. 1979, 82, 22.

This part of the paper is aimed at putting our FIM results into a more general perspective and discussing them in the light of the existing literature data. To begin with, Rh single-crystal surfaces were reported to reveal a rich variety of reconstructions in the presence of an oxygen adsorbate. Experimental methods such as low energy electron diffraction (LEED), scanning tunneling microscopy (STM), helium scattering spectroscopy, and FIM have demonstrated the occurrence of (1 × 2) missingrow forms on Rh{011} and/or {113}.9-14,19,23 For example, in a combined study by LEED/STM on Rh{011} a number of coverage-dependent adsorbate phases could be identified ranging from (2 × 2) p2mg on (1 × 2)-(110) at low coverages to c(2 × 10) on (1 × 5)-(110) at high coverages.10,11,13 The model proposed places the oxygen atoms in 3-fold sites so that -M-O- chains appear in a zigzag form along the [11h 0] crystallographic direction. No such detailed information on oxygen atom positions could be gleaned from FIM alone. On the other hand, FIM allows one to simultaneously investigate the reconstruction characteristics of a large number of small planes forming the surface of a crystal tip and to inspect their mutual communication behavior. In the present study, an adsorbate-induced morphological change was observed transforming the Rh crystal from a hemispherical form into a polyhedral one. Interestingly, the individual {011} and {113} planes were found to exhibit similar missing(35) Schmidt, W. A.; Ernst, N.; Suchorski, Yu. Appl. Surf. Sci. 1993, 67, 101. (36) Suchorski, Yu.; Schmidt, W. A.; Ernst, N.; Block J. H.; Kreuzer, H. J. Prog. Surf. Sci. 1995, 48, 121.

Reconstruction and Surface Oxidation of Rhodium

row reconstructions as the respective macroscopic single crystals. Differences exist insofar as in our case (1 × 3) forms coexist along with (1 × 2). We attribute these differences to the fact that the Rh crystal as a whole seeks to adapt a morphology of the lowest free energy so that the behavior of individual planes loses importance. Thus it is not surprising that the experimental conditions leading to oxygen-induced reconstruction forms are sometimes different from those found in studies with extended single-crystal surfaces (see, for example, ref 19). Chemisorption of oxygen is generally considered the first stage in oxidation. Subsurface diffusion of chemisorbed oxygen atoms can be regarded as a subsequent stage, however, ambiguities remain as to the necessity of critical surface coverages for this to occur. In the course of our FIM studies, see also ref 24, we have obtained clear evidence that subsurface states are occupied after dosing ∼60 L of oxygen to the Rh crystal tip at 550 K. Subsurface diffusion may be followed by nucleation and growth of oxide structures. Kellogg,17 in his combined FIM-atom probe study, reported on the formation of stoichiometric Rh2O3 after heating a Rh crystal tip to 600 K at 1 mbar O2 for times on the order of minutes. Under our experimental conditions (low gas pressures) “deep” oxidation is not expected to occur. Atom-probe studies similar to those performed by Kellogg always indicated Rh/O field desorption/evaporation far below the two-thirds stoichiometric ratio present in Rh2O3.38 However, we have demonstrated here that the surface of a Rh crystal tip undergoes granulation under continuous reaction conditions with O2 gas at 5.2 × 10-5 mbar in the presence of an electric field of ∼15 V/nm at relatively low temperatures (350-483 K). The field-promoted surface granulation is interpreted as being due to the occurrence of strain while oxygen penetrates the surface. In chemical terms, the granulation indicates the nucleation and precipitation of surface oxide clusters, RhxOy. The process is most easily accomplished in the vicinals of the (111) pole meaning that surface steps and kinks increase the initial rate. On the other hand, kinetic constraints to “deep” oxidation are most likely due to the crystallization into extended oxide phases. Since the equilibrium oxygen pressure of Rh2O3 was estimated to be below 10-13 mbar under our experimental conditions, thermodynamic limitations toward deep oxidation are not expected to play any role. Janssen et al.21 studied oxygen interaction with Rh tips over a broad range of temperatures at otherwise low exposures in order to demonstrate the subtle balance between surface oxidation and oxygen subsurface diffusion. The formation of small granules was not observed in the presence of “negative” electric fields as applied in their studies. Interestingly, the presence of oxygen-induced reconstructed planes must not necessarily favor surface oxidation. This conclusion can be drawn on account of parts (37) Lewis, R. T.; Gomer, R. Surf. Sci. 1971, 26, 197. (38) Kruse, N.; Abend, G.; Block, J. H. Unpublished work.

Langmuir, Vol. 14, No. 21, 1998 6157

c and d of Figure 6, which were obtained at temperatures above the critical value necessary for reconstruction (∼400 K). It is seen that the (1 × 2) missing row reconstructed Rh{011} planes remain rather dim indicating slow oxidation (note that we assume that this reconstruction occurs in the presence of an electric field as well as in its absence although the resolution capability of the microscope under imaging conditions at reaction temperatures is insufficient to make visible the atomic surface structure). It may be argued that the inactivity of the plane is dictated by the occurrence of {111} microfacets, which is in line with the observation of a slow oxidation process of the (111) pole of the Rh tip. On the other hand, the {113} planes demonstrate high activity no matter if present in the (1 × 1) bulk-truncated structure (Figure 6a,b) or in the (1 × 3) missing-row form (Figure 6c,d). This latter structure contains alternating {111} and {001} microfacets and is “more open” than the (1 × 2) form of the {011} planes since two adjacent dense-packed rows of atoms are missing The field-promoted surface oxidation is understood as originating from a redistribution of the electron density near the surface.35,36 The presence of an electrostatic field causes a change in the charge distribution of the oxygencovered Rh surface by pushing electrons into the metal. Thus, mechanical strain is built up whereby the spacing between the topmost and the underlying layer is increased such that oxygen atoms can easily penetrate into the Rhsubsurface regions. Clearly, to achieve a detailed microscopic understanding of the field-promoted Rh-oxidation, quantum-mechanical calculations should be performed. Kreuzer et al., using a self-consistent theoretical approach, have recently described the modification of adatom-surface distances in the presence of an electrostatic field.39 Block et al.40 have recently reviewed the importance of high electric fields in various areas of research such as promoter action in heterogeneous catalysis, zeolite chemistry, or nonfaradic electrochemical promotion of the activity of solid catalysts. Studies of the electric field influence on the early stages of metal oxidation can certainly contribute to bridge the gap between basic surface analytic work and studies of corrosion processes occurring on electrode surfaces. Acknowledgment. This work was financially supported by the “Communaute´ Franc¸ aise de Belgique” (ARC, No. 96/01-201), which is gratefully acknowledged. We are also thankful for support by INTAS (Ukraine 95-0186) and a grant (T.V.) by the “Fonds National de la Recherche Scientifique” (FRIA). LA980603C (39) Suchorski, Yu.; Ernst, N.; Schmidt, W. A.; Medvedev, V. K.; Kreuzer, H. J.; Wang, R. L. C. Prog. Surf. Sci. 1996, 53, 135. (40) Block, J. H.; Kreuzer, H. J.; Wang, R. L. C. Surf. Sci. 1991, 246, 125.