J . Chem. SOC., Faraday Trans. 1, 1984,80, 2609-2617 Surface Area and the Mechanism of Hydroxylation of Ionic Oxide Surfaces BY COLIN F. JONES, ROBYN A. REEVE, RUPERT RIGG, ROBERT L. SEGALL, ROGER ST. C. SMART* AND PETER S. TURNER School of Science, Griffith University, Nathan, Queensland 41 1 1, Australia Received 19th April, 1983 The nearly perfect {loo} surfaces of MgO smoke cubes formed in air do not show significant v(0H) absorption in infrared spectra from thin (10 mg cm-2), coherent films exposed to H 2 0 vapour for several hours. It is shown that perfect, five-fold-coordinated sites are not protonated and that the proportion of protonated, low-coordination (i.e. less than five-fold) sites is < 5%. These results are in accord with theoretical predictions for H, adsorption.In contrast, v(0H) [and v(CO,)] absorptions are observed in identical preparations subjected to prior abrasion. The increase in protonated sites is more than ten-fold. Electron microscopy shows that only minor initial alteration of surface structure at edges and corners is caused by abrasion but a major increase in the rate of surface roughening is observed when new, nucleating sites appear and develop in regions across the surface. This process results in a major, time-dependent increase in hydroxylation but it is substantially complete before infrared spectra can normally be obtained. B.E.T. studies, using N, adsorption, do not measure the change in surface area produced by t h s surface roughening. Multilayer water adsorption followed by desorption more than doubles the B.E.T.surface area because of the formation of platelets decomposed from the Mg(OH), brucite surface layer. The particle-size distribution is altered and the B.E.T. method correctly measures this change. The influence of surface structure, through defect (e.g. edge, corner or vacancy) sites, on the reactions that take place at the surface of ionic oxides is well established.l-ll For the adsorption of simple molecules, both experimental and theoretical evidence has illustrated the importance of low-coordination (i.e. less than five-fold) sites. For instance, Boudart et a1.I have shown that the chemisorption (and D, exchange) of H, by MgO depends on the presence of an Mg2+ ion vacancy at a cube corner leaving a micro { 11 l } face consisting of three 0- ions in a triangular array (i.e.a Vl centre). Kunz and Guse,2 using a self-consistent field unrestricted Hartree-Fock approximation, confirmed that the local modification of electronic structure at this Vl centre leads to adsorption of three H atoms. In contrast, a defect-free MgO {OOl) surface is unable to dissociate H, molecule^.^ Extension of the calculations by Mackrodt and coworker^^-^ to consider other defect sites revealed that binding of both CO and H atoms to all low-coordination sites is much higher than to five-fold-coordinated (001 ) surface sites. A recent review by Bermudez' summarises much of the theoretical and experimental evidence for these centres. We are interested in the mechanism of hydroxylation of ionic oxides through the adsorption of H 2 0 molecules from vapour or solution.The dramatic effects of reaction with H,O on surface structure and the consequences of these surface rearrangements for the initial dissolution kinetics in dilute acid solution have been Water, being a triatomic molecule, is a more difficult proposition in computer modelling of adsorption processes and site preferences, and has not yet been 26092610 HYDROXYLATION OF IONIC OXIDE SURFACES studied in this way. Several experimental approaches have, however, been important in relating H,O reactivity to low-coordination sites. For MgO, it is possible to carry out a comparison of material from two preparation methods which result in substantially different proportions of low-coordination ions. MgO smoke, formed by burning Mg in air, produces small (r.m.s.edge length 80 nm), dislocation-free cubes with nearly perfect (001 ] faces.* Thermal decomposition of magnesium hydroxide or carbonate gives high specific surface area crystallites in plate-like pseudo-structures of the starting materials and with highly defective surfaces which contain a number of high-index planes. Collucia et al. lo have compared photoluminescence (and reflectance) spectra of these different surfaces with the different morphology studied by electron microscopy to show that the optical spectra are associated with three- and four-fold coordinated sites in the surface of the oxide. The number of three-fold coordinated sites on MgO smoke is considerably increased by exposure to water vapour as a result of erosion of the edges and corners of the cubes.Further work by Collucia et all1 has shown that this etching of smoke cubes and increase of reactive sites of low coordination does not produce any parallel increase in B.E.T. surface area. This last result has important consequences for any statement of dissolution rate per unit surface area when comparing different surfaces (i.e. with different defect concentrations) resulting from different preparation procedures. In this paper we describe results from infrared, electron-microscope and B.E.T. studies that extend our understanding of the process of hydroxylation and provide a basis for an explanation of several of these effects. We are able, first, to conclude that perfect five-fold-coordinated sites in (100) surfaces do not appear to form OH groups after exposure to H,O vapour whereas defect (i.e.low-coordination) sites do protonate, a result which is the same as that predicted for H, ad~orption.~ In contrast abrasion results in a major increase in v(0H) [and v(C03)] i.r. absorption. Electron microscopy reveals that only a minor alteration of surface structure is caused initially at edges and corners of the cubes by abrasion. The increase in i.r. absorption appears to be out of proportion to (i.e. much greater than) the concentration of new low-coordination sites. However, with time electron microscopy shows that this is caused by an increase in the rate of hydroxylation in new, nucleating regions de- veloping across the surface after this treatment, resulting from surface-structural alteration roughening the surface at the unit-cell level.The B.E.T. surface-area technique does not measure these detailed changes in surface structure. Only when an excess of water vapour, resulting in multilayer formation, is desorbed at high temperature is the B.E.T. surface area doubled because of the formation of platelets decomposed from the Mg(OH), surface brucite formation. The B.E.T. technique in this instance correctly measures the change in size distribution as new surface area is created. EXPERIMENTAL OXIDE PREPARATION MgO smoke was prepared by burning Mg ribbon (B.D.H. laboratory grade reagent) in air. The oxide was variously collected on clean, dry KBr discs, glass plates or e.m. grids held 0.1 m above the ignition. The smoke forms a very coherent, continuous, evenly distributed film on these surfaces, which consist of the nearly perfect cubes studied previously8 by transmission and scanning electron microscopy.INFRARED STUDIES For infrared studies, the sample was transferred, either deposited on KBr discs or after scraping from the glass plate and pressing between two KBr discs, to a Specac variable- temperature vacuum cell. A second method of preparation involved slurrying unmodified orc. F. JONES et al. 261 1 pretreated MgO smoke with carefully dried propan-2-01, followed by drying at 80 "C onto AgCl discs, before transfer to the infrared cell. [There was no evidence of v(CH) absorption in any spectra after this treatment.] The second method allowed control of the mass per unit area in the beam.Considerable care was taken to ensure that all oxide films were continuous and evenly distributed. The infrared cell could be evacuated under conventional vacuum conditions to Pa and pure H,O vapour admitted to the cell. Infrared spectra were recorded on a Perkin-Elmer 62 1 spectrometer using reference-beam attenuation to compensate for the low transmission of the samples (usually < 7% at 2000 cm-l). Adequate spectrometer response generally gave a resolution of ca. 9 cm-l at 4000 cm-l with slow scan rate. Experiments using MgO smoke prepared under glove-box conditions will also be described. This system was flushed with a N,+O, mixture (20 min) and sealed with P,O, as a dessicant for 12 h. The H,O content in the box was estimated to be < 100 v.p.m., but some H,O vapour was clearly present since the formation of the cubic particle shape is dependent on its presence.The MgO samples, prepared by resistive heating to ignition in the box, were pressed between KBr discs and mounted in a vacuum cell similar to that of Cant and Little1, fitted with a windlass system for removing the sample from the i.r. beam, before removal from the box. ELECTRON-MICROSCOPE STUDIES Specimens for electron microscopy were obtained by dispersing the powder ultrasonically in acetone or petroleum ether and holding clean gold grids in the spray above the dispersion. Grids were inserted into the microscope vacuum as quickly as possible (usually < 1 min) and kept in the vacuum until all micrographs had been recorded. Samples prepared in the glove box were deposited directly on to e.m.grids, kept under vacuum and treated identically to the samples for i.r. examination before transfer to the microscope. A JEM-100C instrument was used operating at 100 kV; micrographs were recorded at magnifications of lo5 or higher, with slight underfocus to reveal surface details by phase contrast. SURFACE-AREA STUDIES B.E.T. surface areas were determined using the Perkin-Elmer model 212D sorptometer with N, adsorption (a = 0.162 nm2) at 77 K and have errors of < f 10%. The values quoted here for the same conditions are similar to those of Coluccia et al." determined with a Sartorious microbalance 4102 using N, adsorption at 77 K. RESULTS The fresh, unattacked MgO smoke particles have been carefully characterised. They are nearly perfect cubes with some small cubic projections a few unit cells in dimensions on the (100) faces.899v13 The degree of roughening from these few projections depends on the ambient water vapour pressure during ignition but they are generally ca.5 nm apart. A few crystals show anomalous forms, such as plates or twins,14 the concentrations again being dependent on water vapour pressure. Dislocations occur in < 1 in lo6 particles. The particle-size distrib~tionl~ is extremely asymmetric with most particles having edge lengths in the range 45-85 nm but ca. 1 in 1 O4 having edge lengths > 1.5 pm. INFRARED STUDIES When MgO is deposited directly onto a KBr plate as an even, continuous film to 10 mg cm-2 and transferred into a vacuum cell in < 10 min, we find no measurable absorptions in the infrared spectra between 4000 and 750 cm-l.v(0H) and v(C0,) only become measurable after the sample has been standing in air for periods > 12 h. If the thickness of the deposited film is increased until total absorption from MgO vibrations is found below 800 cm-l (i.e. ca. 50 mg cm-2), very weak absorptions (i.e. absorbance increase above background of c 0.05) are seen at 3500 and 1 600- 1 400 cm-l.2612 HYDROXYLATION OF IONIC OXIDE SURFACES 4 000 3500 31 1 I I I I \ 5 1600 1L 00 1000 6 wavenum ber/cm-' Fig. 1. Infrared spectrum of fresh MgO collected on a glass plate scraped onto a KBr plate, with a second KBr plate used to spread the MgO (10 mg cmP2) evenly by gentle rotation and the two plates clamped together. Discontinuities in the spectra are caused by changes in reference- beam attenuation.4 000 3500 3000 1600 1400 1200 1000 Fig. 2. Ratio-recorded infrared spectra of (i) sample beam containing lightly abraded (mortar and pestle) MgO smoke solvent-dried, as described in the text, onto a KBr plate and (ii) reference beam containing the same thickness of fresh, unabraded MgO smoke solven t-dried on to a KBr plate. wave number /cm 0 In contrast, when MgO smoke is deposited on to a glass plate, scraped off and then dispersed into an even, continuous film of 10 mg cmP2 between two KBr windows, a spectrum like that in fig. 1 is found. Two broad v(0H) absorptions16--'o are evident near 3650 and 3500 cm-' with broad v(C0,) absorptions16-20 at 1650,1520,1380,1250, 1000 and 850 cm-l.J .Chem. SOC., Faraday Trans, 1, Vol. 80, part 10 Plate 1 Plate 1. Phase-contrast electron micrograpns or ivigu srnoKe C U D ~ S LaKen ai uiiiereni Limes after lightly grinding in petroleum ether. (a) Initially, within minutes of preparation, all regions of the specimen examined show smooth (100) faces with a few broken cube corners (arrowed). (b) After 10 min in the microscope vacuum (< lop5 Pa H,O) all cubes exhibit the onset of surface roughening; this is most readily visible in phase contrast on steeply inclined faces (as at arrows). (c) After a further 15 min all cubes show a high degree of roughening. C. F. JONES et al. (Facing p . 26 12)J. Chem. Soc., Furaduy Trans. 1, Vol. 80, purt 10 Plate 2 Plate 2. MgO cubes exposed to H,O vapour for 20 h at 24 "C and outgassed at 1200 K.The original cubes (marked A) ae substantially eroded. The eroded material forms sheets of hydroxide which are reduced to finely divided MgO (arrowed) upon heating. The dark regions such as the one marked B arise from strong Bragg diffraction.c . F. JONES et al. 2613 The only processes likely to have caused this change were the scraping and pressing operations involving particle contact. To test this possibility, fresh smoke was very lightly ground in a mortar and pestle and examined by both i.r. and e.m. techniques immediately after drying off the alcohol. The effect of this treatment on surface hydroxylation can be seen in the infrared spectrum of fig. 2, where the lightly ground smoke (sample beam) is run against the unmodified smoke (reference beam) using the same number ofmg cm-, in each beam.The samples were dried down from propan-2-01 slurries as described in the Experimental section. There has been a substantial change in the background below 1400 cm-l giving rise to the steep slope evident in the spectrum but, in addition, clear differences in absorption are seen at 350&3400, 1600 and 550 cm-l with a shoulder at 1000 cm-l. If the smoke surfaces were prepared in the glove box and disordered by scraping and pressing between KBr plates, weak v(0H) absorption was still found in the i.r. spectra [although no v(CO,) was detected] despite the low H,O vapour content (< 100 v.p.m.) and short time of exposure to the glove-box atmosphere (i.e.< 5 min). This evidence slfggests that there has been a major increase in the concentration of sites at which protonation and CO, adsorption can take place as a result of the abrasion process. ELECTRON-MICROSCOPE STUDIES It is tempting to assume from these results that the cubes have been damaged by abrasion so as to expose extensive areas of low-coordination sites compared to the virgin smoke. The electron microscope reveals that the effects of abrasion are more complex. Although there is some damage to cube corners and edges, initially the cube surfaces appear almost unaffected [plate 1 (a)]. After relatively short times ( i e . < 10 min) in the microscope vacuum (< Pa H,O vapour), the cube surfaces begin to show changes in surface structure or roughening effects [plate l(6) and (c)].In this experiment it was important to examine different areas of the specimen at successive times and to ensure that these areas had not been exposed to the electron beam while the roughening process proceeded. This is because the electron beam can itself induce surface damage on water-exposed smoke.21 With this precaution, however, we consistently observed new regions of surface alteration developing across the cube surfaces with time. This structural alteration introduces a high concentration of small projections and intrusions, still based on {loo} facetting, of a few unit cells in any dimension. The result of this process is a substantial increase in the number of surface sites which can be hydroxylated (or adsorb CO,). In view of this evidence, it is not surprising that the increase in v(0H) and v(CO,) is much greater than that expected from the damage inflicted by the abrasion process alone.This initial damage appears to be catalytic in inducing much more rapid surface alteration. This alteration is qualitatively similar to that found in our work immediately after immersion in water and before any dissolution occurs, i.e. ' castellation '.8 SURFACE-AREA STUDIES A surface area of 10.5f 1 m2 g-l is consistently found for the virgin smoke. A reproducible reduction of 15 % (i.e. to 9.0 1 m2 g-l) is found for all samples of lightly ground smoke. There has been major alteration in surface structure but, of course, the atomic-scale rearrangements described in plate l(b) and (c) do not alter the particle-size distribution in any significant way.We believe that this result illustrates the inability of the B.E.T. adsorption process to detect the atomic detail of the disordering, but its ability to detect the comparatively unimportant change in the particle-size distribution caused by some pressure-induced wringing together of particle surfaces.2614 HYDROXYLATION OF IONIC OXIDE SURFACES 1 I 00 3500 3 000 25 11800 1600 1400 1200 1000 wavenumber/cm -' Fig, 3. Infrared spectrum of MgO smoke exposed to 3 dm3 of H,O vapour at 2 kPa for 25 h, heated to 900 "C for 4 h and solvent-dried on to a KBr plate. Further evidence for this contention is provided by two more experiments. We have repeated the experiment of Collucia et al." They exposed smoke to a fixed quantity of H,O vapour for 20 h at 300 K and outgassed at 1200 K for 1 h.This has been shown to produce substantial surface disordering.1° We confirm that this treatment does not change the B.E.T. surface area despite the considerable change in surface structure. This is true also for the smoke after exposure to a limited quantity of H,O and evacuation at 300 K but before 1200 K outgassing; again the B.E.T. surface area is unchanged. The i.r. spectrum shown in fig. 3 confirms the increase in adsorption sites for H,O and CO,, leading to the conclusion that low-coordination sites are involved. In our experiments we estimated that the total quantity of H,O vapour available to the surface to produce this change corresponds to less than two monolayers. The second experiment of interest involved exposure of the smoke particles to a continuous supply of H,O vapour from bulk liquid in a closed dessicator over 20 h at 24 "C.The particles were then outgassed at 1200 K. The new surface area was found to be 23 f 1 m2 g-l. It is interesting that the sample after this exposure to H 2 0 vapour followed by evacuation at 24 "C, i.e. before 1200 K outgassing, gave the same surface area of 23 1 m2 g-l. Electron microscopy after 1200 K outgassing reveals (plate 2) the presence of extensive sheet-like particles amongst the heavily eroded MgO particles. In this experiment the MgO platelets have formed from decomposition of Mg(OH), in the layered structure of the brucite lattice. It appears that extensive hydroxide layers have been formed on the surfaces as multilayers of water are adsorbed.The particle-size distribution is significantly shifted towards smaller particles with the reduction in size of the original cubes and the addition of the small, very high surface area platelets. [As expected, the infrared spectrum shows very intense v(0H) absorption on exposure of the specimens to air.] The increase in measured area is associated with the shift in the particle-size distribution towards smaller particles. In all of these experiments the evidence is that the changes in B.E.T. surface area reflect only the changing particle-size distribution rather than the detailed atomic-scale disordering.c . F. JONES et al. 2615 DISCUSSION DEFECT CONCENTRATIONS ON MgO SURFACES The lack of v(0H) absorption on unattacked smoke surfaces and the very slow rate of hydroxylation in air can be explained in terms of the concentration of low- coordination sites on these surfaces.An estimate of the concentration of three- and four-coordinated sites on virgin MgO smoke surfaces can be made with a few simplifying assumptions. If we assume a 65 nm edge-length cube, with cubic projections of two-unit-cell dimensions spaced 5 nm apart across the (100) faces, this, including edge and corner sites on these projections plus those on the cubes themselves, gives ca. 5% of surface sites with less-than-perfect coordination. If we use (@, which has a value of 120 nm from the size distribution, this estimate reduces to 4% of all surface sites. If these are the sites that become protonated, as suggested by the theoretical StudieF and the work of Collucia et al.,lo7 l1 then we can also estimate the likely v(0H) infrared absorption from these sites.Assuming a relatively large extinction coefficient for the hydrogen-bonded v(0H) absorption near 3400 cm-l of ca. 13 mol-1 m2 and an absorbance increase (above background) of ca. 0.01 as a measurable limit in a conventional infrared spectrometer, then we need ca. 8 x lo-* mol of OH groups rn-, of infrared beam. For MgO smoke with a lattice parameter 4.20 A and a surface area of 10.5 m2 g-l dispersed as 10 mg ern-,, if we assume that 5% of all surface sites are OH groups, then we arrive at an estimate of 9.5 x a01 m-, for the actual MgO films. From these estimates it is clear that we are very close to the limit of detectability of infrared absorption from OH groups associated with low-coordination sites but, more importantly, it is obvious that if all surface sites were protonated we would detect the infrared absorption at 10 mg cm-, dispersion without difficulty.The estimates are also consistent with the observation of weak absorption when the dispersion thickness is increased to ca. 50 mg crn-,. We have now seen that major surface alteration is catalysed by the abrasion process although this process produces relatively little initial damage. It is not clear from our results whether the surface roughening occurs first (followed by protonation of the new defect sites produced) or whether protonation itself is enhanced by the process resulting in surface reordering to accommodate the changes in site energies.The i.r. results appear to indicate that the number of hydroxyl groups on the surfaces has increased at least ten-fold because of the surface disordering since we observe an absorbance increase > 0.1 in v(0H) absorption. This increase is easily achieved by the creation of pits and protrusions of one- and two-unit-cell dimensions on the sur- face since each edge or corner oxygen site in such defects becomes a centre for ready prot~nation.~, We may speculate that the mechanism for production of surface sites may involve hydroxylation of a five-fold site by pressure. It is certain that physically adsorbed H,O molecules will be present over most surfaces of the smoke cubes shortly after production of the 2o We surmise that pressure-induced electrostatic forces on an H,O molecule are sufficient to hydroxylate a surface oxygen ion.Then even light grinding is sufficient to produce many active sites across a smooth surface, which in turn acts to nucleate the hydroxylation, and roughening, of the surface in times very short compared with the time it takes undamaged smoke exposed to H,O vapour to be totally roughened. and to explain their e.s.r. results of the number of 0; ions observed. For 170 nm perfect cubes, the size used in their estimates, 4.5 x 1014 g-' three-coordinated sites are expected from corner sites.ll This figure is at least one order of magnitude less than It is now possible to consider some estimates in the work of Coluccia et2616 HYDROXYLATION OF IONIC OXIDE SURFACES the amount of 0; actually found on unmodified smoke, although these ions are believed to form on only a fraction of the three-coordinated sites actually present.For a particle-size distribution such that (P): = 120 nm,ll the number of three-coordinated sites from the cube corners is only ca. 2 x 1 OI4. Now, ifwe include the same two-unit-cell projections spaced ca. 5 nm apart on these surfaces as discussed above, this estimate increases to > lo1’ g-l. This increase can explain the 0; concentration observed.6 Attack by water vapour is found in the work of Coluccia et al. to increase the population of these sites by at least a factor of ten, which is in accord with our infrared results. SURFACE STRUCTURE AND SURFACE AREA The results from these studies strongly suggest that conventional B.E.T. surface-area measurements cannot detect, the atomic-level rearrangements involved in the surface roughening associated with hydroxylation and abrasion. There are two possible reasons for this insensitivity.The first is that the N, molecule with an area of 16 A2 is sufficiently large that steps of one- and two-unit-cell dimensions make little difference to the total count of adsorbed N, molecules. It is conceivable that the physically adsorbed N, molecules, which will have some significant lateral quadrupolar inter- actions, are acting like a coherent skin over the surface rather than as site-specific adsorbates. The second interesting possibility arises from the recent work by Colbourn and Ma~krodt,~ where it has been shown that ions at the corners and edges of small cubes or projections relax inwards whilst ions at the base of steps and ledges relax outwards (by 0.1-0.4 lattice parameter).This relaxation has the effect of partly ‘smoothing’ the surface and may contribute to the insensitivity of the B.E.T. measurements to these changes. CONCLUSIONS The nearly perfect surfaces of MgO smoke cubes formed in air containing a very small concentration of water vapour do not show significant v(0H) absorption in i.r. spectra (at 10 mg ern-,) because the proportion of surface oxide ions that are pro- tonated (i.e. less than five-fold-coordinate sites) is < 5 % . The five-fold-coordinate sites on perfect (100) surfaces are not protonated, and thus there are too few OH groups in the infrared beam.In contrast, v(0H) [and v(CO,)] are observed on specimens identical to those above apart from prior abrasion through pressure or light grinding. A more than ten-fold increase in active, low-coordination sites is observed. Electron microscopy shows that this difference results from a major change in the rate of hydroxylation and extensive surface roughening because of the presence of new, pressure-induced active sites and it is not caused by any difference in surface structure arising directly from abrasion. B.E.T. studies, using conventional N, adsorption, do not measure the change in surface area produced by this major alteration in surface structure . An excess of water vapour, resulting in multilayer formation, followed by desorption gives more than double the surface area, because of the formation of platelets decomposed from the Mg(OH), brucite structure.The size distribution is altered and new surface area created. The B.E.T. method correctly measures this change in particle- size distribution. Support from the Australian Research Grants Scheme is gratefully acknowledged.c. F. JONES et al. 261 7 I M. Boudart, A. Delbouille, E. G. Derouane, V. Indovina and A. B. Walters, J. Am. Chem. SOC., 1972, 94, 6622. A. B. Kunz and M. P. Guse, Chem. Phys. Lett., 1977, 45, 18. E. A. Colbourn and W. C. Mackrodt, Surf. Sci., 1982, 117, 571. E. A. Colbourn and W. C. Mackrodt, Solid State Ionics, 1983, 8, 221. E. A. Colbourn, J. Kendrick and W. C. Mackrodt, Surf. Sci., 1983, 126, 550. J. Kendrick, E. A. Colbourn and W. C. Mackrodt, Radiat. Eff., in press. R. L. Segall, R. St. C. Smart and P. S. Turner, J. Chem. SOC., Faraday Trans. I , 1977, 73, C. F. Jones, R. L. Segall, R. St. C. Smart and P. S. Turner, Proc. R. SOC. London, Ser. A, 141. lo S. Coluccia, A. J. Tench and R. L. Segall, J. Chem. SOC., Faraday Trans. I , 1979, 75, 176' l 1 S. Coluccia, A. Barton and A. J. Tench, J. Chem. SOC., Faraday Trans. I , 1981,77, 2203. l 2 N. W. Cant and L. H. Little, Can. J. Chem., 1968, 46, 1373. l3 A. F. Moodie and C. E. Warble, J. Cryst. Growth, 1971, 10, 26. ' V. Bermudez, Prog. Surf. Sci., 1981, 11, 1. 1710. 98 1, 374, l4 R. R. Cowley, R. L. Segall, R. St. C. Smart and P. S. Turner, Philos. Mag., Ser. A, 1979, 39, 163. l5 C. F. Jones, R. L. Segall, R. St. C. Smart and P. S. Turner, Philos. Mag., Ser. A , 1980, 42, 267. lCi L. H. Little, Infrared Spectra of Adsorbed Species (Academic Press, London, 1966). l7 M. L. Hair, Infrared Spectroscopy in Surface Chemistry (Edward Arnold, London, 1967). l9 J. V. Evans and T. L. Whateley, Trans. Faraday SOC., 1967, 63, 2769. 2o S. J. Gregg and J. D. Ramsay, J. Chem. SOC., 1970, 2784. 21 C. F. Jones, R. L. Segall, R. St. C. Smart and P. S. Turner, Radiat. Eff., 1982, 60, 167. R. St. C. Smart, T. H. Slager, L. H. Little and R. G. Greenler, J . Phys. Chem., 1973, 77, 1019. (PAPER 3/627)