J . Chem. Soc., Faraday Trans. 1, 1982, 78, 2649-2659 Spectroscopic Study of Anatase Properties Part 5.-Surface Modifications Caused by K 2 0 Addition BY CLAUDIO MORTERRA,* ANNA CHIORINO AND GIOVANNA GHIOTTI Istituto di Chimica Fisica dell’universita di Torino, Corso Massimo d’Azeglio 48, 10125 Torino, Italy AND EMILIA FISICARO Centro Ricerche Sibit, Gruppo Montedison, 15047 Spinetta Marengo (Al), Italy Received 10th September, 1981 Infrared spectroscopy has been used to investigate the surface properties of anatase gels containing up to ca. 1 % K,O, one of the additives most commonly employed to give TiO, pigmentary characteristics. It was found that K,O mostly collects at the surface of the material, as revealed by the spectral behaviour of surface hydroxyls and of surface sulphate contaminants.During the thermal treatment leading to the pigmentary material, K,O progressively modifies TiO, surface acidity, so that on the low-surface-area final product no Ti ions are revealed by suitable admolecules (pyridine, CO), but only coordinatively unsaturated K ions, acting as weak Lewis centres. The surface modifications caused by K,O are slowly reversible on contact with water vapour. Our previous studies (Parts 1-41-4) have been devoted to the characterization of the surface properties of pure anatase and to the identification of the nature and activity of various anionic impurities that arise from the manner of preparation. We considered these studies as being preliminary to an investigation of the role played by the various additives that are needed to give TiO, its well-known pigmentary characteristics.This work deals with the addition of K20, whose concentration in TiO, pigments ranges between 0.05 and 0.4 wt % and whose presence is thought to be essential for adequate particle growth during the thermal treatment leading to the production of white pigment. In this work i.r. spectroscopy has been used to elucidate the changes produced by the presence of K20 and revealed by modifications to both the background spectrum of the solid and the spectral features of the interaction of the solid with suitable test molecules. EXPERIMENTAL The anatase used here as a starting material is a ‘ via-sulphate preparation’ (hereafter referred to as TS) coming from the H,SO, hydrolysis of pure titanyl sulphate, whose characterization has been reported elsewhere.2* As the addition of K,O to TiO, pigment preparations occurs in the early stages of anatase-gel production, gel samples containing K,O (hereafter referred to as TSK) were studied after various thermal treatments at temperatures ranging from 298 K (i.e.after ambient-temperature drying) to 1073 K (i.e. just before the anatase-rutile phase transformation). Two K,O concentrations were investigated, namely 0.2 and 1.0 wt %, but all data reported refer to the latter concentration unless otherwise stated : this concentration highlights the 86 2649 FAR 12650 I.R. STUDY OF THE TiO,/K,O SYSTEM differences between TS and TSK, and no qualitative differences were observed between the two preparations of different K 2 0 content. 1.r.spectra were run on a Perkin-Elmer 580/B double-beam spectrophotometer. All details concerning experimental procedures and instrumentation have been reported previously. 1v All gases and chemicals were of the highest-purity grade available and were further purified in situ using standard techniques RESULTS AND DISCUSSION No appreciable differences between TS and TSK were observed in the spectral features either of the surface hydroxyls during dehydration-rehydration experiments or after contact with CO, of specimens dehydrated in various ways. In the former case the lack of resolution at i j 2 3000 cm-l, caused by severe scattering,, can also be invoked as a factor limiting the spectroscopic selectivity, whereas in the latter case, based on spectra run in a well-resolved range, the addition of up to ca.1 % K,O does not affect the scarce surface basicity of the anatase revealed by CO, chemi~orption.~ Significant differences between TS and TSK were observed in the low-frequency background spectrum, where sulphate impurities absorb, and in the activity towards acid-revealing test molecules, such as pyridine (Py) and CO. The relevant results are described separately. BACKGROUND SPECTRUM I N THE 1000-1800 Cm-' RANGE Fig. 1 compares this spectral range of a TS and TSK starting samples outgassed at up to 673 K, and then rehydrated and outgassed again at ambient temperature. This is the spectral range of coordinated molecular water (having a band at ca. 1630 crn-l)l* and of sulphate impurities., Three facts can be observed.(i) On passing from TS to TSK there are no changes of shape, spectral position, thermal stability, and reversibility of the band owing to coordinated water. This indicates that, at least in the early stages of thermal treatment (Le. in the temperature interval needed to remove coordinated molecular water), no surface acidic sites of those species responsible for water coordination are either ascribable to or appreciably affected by the presence of potassium. (ii) Although the level of sulphur contamination remains unchanged on passing from TS to TSK gel, on TSK samples there is a definite increase in the amount of surface sulphates that are typical of TS., This is particularly evident for samples treated at T > 473 K, characterized by a surface sulphate of highly covalent character absorbing at ca.1375 and ca. 1130 cm-l. Moreover, this sulphate is definitely more loosely held on TSK than on TS [cf., e.g., the band at 1375 cm-l in spectra 4 and 5 of fig. 1 (a) and (b)]. (iii) On TSK samples there is a new sulphate species whose low-8 SO, stretching mode presumably lies in the unresolved band at ca. 1130 cm-l but whose high-v mode is at ca. 1275 cm-l on samples activated at low temperatures and moves to 1300 cm-l following activation at T 2 473 K. The reversibility of this spectral behaviour on rehydration (spectrum 6 of fig. 1) indicates that this sulphate species, obviously ascribable to the presence of potassium, is also at the surface, at least in the early stages of the thermal treatment considered here.The assignment of sulphates which are characteristic of TSK samples, both at high and low dehydration stages, is not straightforward, and requires a close comparison with the behaviour of pure TS samples. The high-ij SO, stretching mode observed on TSK, at both dehydration stages, isC. MORTERRA, A. CHIORINO, G. GHIOTTI AND E. FISICARO 265 1 wavenumber/cm-' FIG. 1.-The 1050-1725 cm-' spectral range of a previously untreated sample of TS (a) and TSK (b), respectively. S1-5: the sample was treated in vacua at 298, 373, 473, 573 and 673 K, respectively. S6: the sample of spectrum 5 was rehydrated and further dehydrated at 298 K. compatible with that of sulphato complexes of Czv symmetry, such as chelating and bidentate sulphato c~mplexes.~ On TS only, the sulphates typical of low dehydration stages had frequencies compatible with that modeL2 It is also recalled that, on TS, dehydration causes the reversible transformation of chelating or bidentate sulphato complexes into covalent sulphates,6 characterized by a higher splitting of the two SO, stretching modes.The similar spectral behaviour of TSK sulphates on dehydration- rehydration suggests that the sulphate species typical of K-containing preparations undergoes the structural changes previously suggested for TS, following the overall polarity of the surface layer. Thus on highly dehydrated materials, characterized by a low polarity of the surface layer, there can be two types of covalent sulphate,6 of which the one containing potassium exhibits a smaller splitting between the two SO, stretching modes, owing to the more polar character induced by the alkali-metal ion(s).2652 I.R.S T U D Y OF THE TiO,/K,O SYSTEM ADSORPTION OF PYRIDINE The use of pyridine as a selective probe molecule to characterize both qualitative and quantitative aspects of surface acidity is widespread [see, for instance, ref. (7) and (8) and references therein], and the suitability of 8 a-8 b and 19a-19 b ring-stretching modes as analytical tools has been previously suggested.** The reasonable number of data available in the literature, concerning both homogeneous and heterogeneous Py compounds and complexes, allowed us to outline the 1400- 1700 cm-l spectral range as reported in fig. 2: in it the width of the schematic bands only concerns the range of occurrence, rather than the actual bandwidth, whereas the height only indicates the difference in intensity between fundamental and combination modes.wavenum ber/cm-' FIG. 2.-Schematic representation of the spectral range of some Py modes in various Py-containing homogeneous and heterogeneous compounds. (No reference is made in the scheme to any actual intensity ratio, but rather to the difference between fundamental and combination modes.) PYRIDINE INTERACTION W I T H HIGHLY HYDRATED MATERIALS Fig. 3(a) shows the interaction of Py with a TSK starting sample activated at ambient temperature (i.e. still possessing most of the surface sulphate impurities and highly hydrated), whereas fig. 3 (b) shows the interaction of Py with a sample that was first freed from all sulphate contaminants* and then rehydrated and outgassed at ambient temperature.Most of the spectral features previously reported for TS evacuated at the same temperature3 are found for TSK as well, including the formation of PyH+ species when in the presence of sulphate contaminants, shown by the band * It was previously shown2 that the complete elimination of sulphate contaminants can be achieved upon a prolonged outgassing at temperatures as high as 873 K. In the present work another technique has been adopted, leading to the elimination of sulphate at lower temperatures and in shorter times, and thus with a less severe surface-area loss. It consists of the vacuum decomposition at ca. 773 K of previously chemisorbed Py : the reducing atmosphere owing to the pyrolysis of the organic admolecules causes an easier elimination of sulphates as well as partial reduction of TiO, 1,2 that is completely reversible upon 0, treatment at T 2 673 K.2C.MORTERRA, A. CHIORINO, G. GHIOTTI A N D E. FISICARO 2653 I 1650 1600 1550 1500 wavenumber/cm-' FIG. 3.-Interaction of Py with a TSK sample dehydrated at 298 K. (a) Virgin sample, (b) the same sample after the elimination of sulphate contaminants and rehydration. S1: background; S2: after contact with Py (266 N mP2); S3: after Py evacuation at 298 K. at ca. 1545 cm-l and by the intensity of the band at 1640 cm-l (see scheme of fig. 2). Also the almost complete ligand displacement of undissociated molecular water, which has been shown3 to lead to the formation of a Py species strongly Lewis-coordinated to coordinatively unsaturated Ti ions (Ticus ions, 8a mode at 1608-1614 cm-l), is observed on TSK, in both the presence and absence of residual sulphates.The following facts can also be observed. (i) There is a band at ca. 1600cm-l, ascribable to the 8a mode of Py hydrogen-bonded to surface hydroxyls, whose poor2654 I.R. STUDY OF THE TiO,/K,O SYSTEM intensity is nearly equal to that of the same species on TS outgassed at 298 K and that is nearly unchanged on passing from the sample of fig. 3(a) to that of fig. 3(b). (ii) Py species that are easily removed by evacuation at ambient temperature (namely, the liquid-like species with an unresolved 8a-8b mode at 1582 cm-l and the hydrogen-bonded species with an 8a mode at 1600 cm-l) possess a well-resolved 19a component at ca.1484 cm-l, whereas the sum of the 19a modes owing to both Lewis and Brarnsted species lies at ca. 1494 cm-l. (iii) There is a weak band at 1588- 1590 cm-l, not observable on TS,3 that is probably caused by the 8a mode of a new Py species, whose 8b mode is believed to lie in the unresolved 8b envelope at 1578 cm-l. The scheme of fig. 2 suggests the assignment to Py coordinated to weak Lewis-acidic sites, and the absence of such a species on TS allows us to ascribe it to Py interacting with coordinatively unsaturated K ions (K,,, ions). The proposed assignment is consistent with the i.r. and Raman frequencies of 8 a-8 b modes of Py Lewis-coordinated to K ions in zeolites.lO*ll Also, the small splitting between the two v8 ring modes is consistent with the weak electrostatic field produced by ions characterized by a large radius and low charge.1° (iv) On passing from fig.3(a) to fig. 3(b) there is no appreciable change in intensity of the band owing to Kc,,-coordinated Py species. It is thus argued that surface sulphates, which the spectral features of fig. 1 clearly indicate to be affected by the presence on TSK of K ions, when removed do not leave many K,,, ions capable of interacting with Py. PYRIDINE INTERACTION WITH DEHYDRATED MATERIALS Fig. 4 shows the 1400-1700 cm-l spectral range of Py adsorbed at ambient temperature and desorbed at various temperatures on a TSK sample dehydrated at 673 K, i.e. almost complete dehydration,2 and from which all sulphate impurities had been removed.The following observations can be made. (i) The very few residual hydroxyls2 still involve some Py by hydrogen bonding, yielding an 8a mode band at ca. 1600 cm-l of much the same intensity as that produced on the highly hydrated TSK sample of fig. 3, whereas on TS after outgassing at temperatures as low as 498 K hydrogen-bonded Py could no longer be observed in the spectrum. This suggests that potassium does not induce in anatase hydroxyls a higher capacity to undergo An-- 1650 1603 1550 1500 14 wavenumber/cm-' FIG. ,4.-Interaction of Py with a sulphate-free TSK sample dehydrated at 673 K. S1: background; S2: after contact with Py (266 N m-2); S3 and S4: after Py evacuation at 298 and 373 K, respectively.C. MORTERRA, A. CHIORINO, G.GHIOTTI A N D E. FISICARO 2655 hydrogen bonding with Py, but, unlike TS, such hydroxyls are desorbed last during thermal activation. (ii) The 19a mode of the easily removable Py species (1484 cm-l) is far more intense than on the sample activated at 298 K: as this increased intensity cannot be ascribed to hydrogen-bonded Py, whose intensity is at most unchanged, it must be caused by the liquid-like Py species (8a-8b modes at 1582 cm-l), thus confirming that this species cannot be thought of as being physically adsorbed, but that a weak specific interaction of the acid-base type must also be involved in this case, as previously s~ggested.~ (iii) The 8a mode owing to Py Lewis-coordinated to K,,, ions is much stronger than on the 298 K sample, and is completely desorbed within 373 K (spectrum 4 of fig.4). It was previously shown that Py strongly Lewis-coordinated to Ti,,, ions (8a mode at 1608-1614 cm-l) adsorbs at almost its maximum concentration even on samples activated at ambient temperature, through a ligand-displacement mechanism at the sites that coordinate water in the undissociated form. This is also true for TSK samples [see fig. 3(a) and (b)], so that the almost complete coordinative saturation of K ions on a 298 K sample, both in the presence and in the absence of residual sulphates, cannot be primarily ascribed to undissociated coordinated water. It is thus deduced that, as far as the production of activity towards Py is concerned, the thermal treatment of TSK samples in the 298-673 K range enhances the concentration of sites responsible for the adsorption of liquid-like Py species (as does the thermal treatment of TS samples) and creates Lewis-acid centres coming from the production of K,,, ions and that this is somehow connected with the surface dehydroxylation process rather than with the desorption of coordinated water or of surface sulp hate contaminants.PY R I D I N E INTER ACT I O N FOLLOW I N G HI GH-TE M PER A T U R E TREAT MEN T When the thermal treatment of pure TS anatase is carried out at temperatures higher than 673 K, i.e. at temperatures above complete dehydration, there is a rapid surface-area loss; this can be revealed by B.E.T. measurements2 and, upon Py chemisorption, by a severe weakening of the spectrum, although the overall features t c ... m '2 ;=" 2 i 6'XI 1600 1550 l650 l6bo 1550 1650 6;oo 1550 wavenumberlcm-' FIG.5.--Interaction of Py with a TS sample outgassed at 923 K (a), with a TSK sample outgassed at 923 K (b), and with a TSK sample outgassed at 1073 K (c). S1: background; S2: after contact with Py (266 N m-"; S3: after Py evacuation at 298 K.2656 I.R. STUDY OF THE TiO,/K,O SYSTEM remain almost unchanged. [See fig. 5(a), relative to a sample treated at 923 K: only the liquid-like Py species exhibits a decreased relative intensity.] If the same thermal treatment is carried out on TSK and Py is allowed to contact the sample, fig. 5(b) shows that the overall shape of the spectrum changes drastically. (i) The band owing to the 8a mode of K-coordinated Py possesses a much higher relative intensity with respect to all other components, both in the presence of Py and after Py evacuation at ambient temperature.As for the absolute intensity of the band at 1590 cm-l, it is higher on a sample treated at 923 K [fig. 5(b)] than on a sample treated at 673 K (fig. 4) or at 773 K (not shown in the figures), despite the surface-area loss. This fact indicates that the complete elimination of surface hydroxyls (very few OH groups remain on the 673 K sample and none on the 773 K one, on the basis of the i.r. data) is not the sole mechanism through which K,,, ions, acting as Lewis-acidic centres, are produced in this temperature range. (ii) If the intensity of the band envelope at 1582 cm-l(8a-8b mode of liquid-like Py) and at 1578 cm-l(8b mode of all other species) is assumed as a rough estimate of the total amount of chemisorbed P Y , ~ ~ * fig.5(a) and (b) tell us that on the relevant samples the total amounts of Py are comparable, but after Py evacuation at ambient temperature much less Py remains adsorbed on TSK. In particular, the number of Py species Lewis-coordinated to Ti,,, ions is halved by such a treatment, whereas it is slightly decreased on TS. The quicker reversibility of this species on TSK is made particularly evident in fig. 5(c), showing that the small amount of Py still Lewis-coordinated to Ti,,, ions on a TSK sample activated at 1073 K is fully reversible at ambient temperature. Also, the fraction of Py coordinated to K,,, ions reversible at ambient temperature is higher the higher the sample activation temperature.The above experiments suggest that, even at temperatures above 1073 K (the temperatures employed to prepare TiO, pigments), all Lewis acidity owing to Ti,,, ions and characteristic of the early stages of thermal treatment would be destroyed by the presence of potassium. REVERSIBILITY OF SURFACE ACIDITY MODIFICATIONS The reversibility of TSK surface acidity, modified by high-temperature treatment, has been checked as follows. A sample treated as in fig. 5(c) and freed from all chemisorbed Py was kept in long contact with water vapour, and further dehydrated at various temperatures. After each treatment the acidity was checked by Py ad- sorption at ambient temperature, and the relevant results are summarized in fig. 6. Even after dehydration at ambient temperature, the ratio of Py Lewis-coordinated at K,,, sites to Py Lewis-coordinated at Ti,,, sites is higher than on the starting material.The incomplete reversibility on rehydration of the acidity changes is made even more evident by parts (b) and (c) of fig. 6: on the sample treated at 673 K the 8a band at 1590 cm-l is already stronger than the 8a band at 1610 cm-l, and after thermal treatment at 823 K we obtain the same profile for the Py bands and rapid reversibility which are typical of previously untreated samples treated at higher temperatures [see fig. 5(b) and (c)]. Note also that the Py 8 b mode at 1578 cm-l and 19a mode at ca. 1490 cm-l (not reported in fig. 6) are similar in the three sections, showing that no further surface-area changes have occurred, and that the overall amount of chemisorbed Py does not vary appreciably with activation temperature.It is thus confirmed that, both before and after complete dehydration, the Lewis acidity of the K,,, ions is produced at the expense of the Lewis acidity of the Ti,,, ions. Prolonged contact with water at ambient temperature of TSK samples freed fromC. MORTERRA, A. CHIORINO, G. GHIOTTI AND E. FISICARO 2657 1650 1600 1650 1600 1550 wavenurnber/cm-' FIG. 6.-Interaction of Py with a TSK sample outgassed at 973 K, rehydrated at 298 K (14 h) and further outgassed at: (a) 298, (b) 673 and (c) 823 K. S1: after contact with Py (266 N m-z); S2: after Py evacuation at 298 K. all sulphate contaminants (see previous footnote) previously outgassed at high temperatures restores the ratio between stronger and weaker Lewis-acid sites only partly.In fact, on passing from a sample rehydrated at ambient temperature after outgassing at 773 K to a sample rehydrated after outgassing at 973 K, there is a strong decrease in the ratio of the total amount of coordinated water to the amount that remains coordinated after evacuation at ambient temperature : the ratio between the absorbances at the H,O bending maximum passes from ca. 0.40 to ca. 0.16. ADSORPTION OF CARBON MONOXIDE In a previous paper3 is was shown that TS samples still carrying residual sulphate impurities behave differently towards CO chemisorption from anatase samples having undergone different preparation paths. After only a thermal elimination of sulphates2 very little can be said about the activity towards CO, as the high temperature required brings about such a severe surface-area loss that all the i.r.activity of CO adsorption is lost. In contrast, the thermal-chemical procedure adopted in the present work to eliminate sulphate impurities (see previous footnote) allows TS samples to maintain a sufficient surface area to yield well-detectable i.r. bands on CO chemisorption. Fig. 7(a) shows that sulphate-free rehydrated TS samples behave like all other anatase sample^,^' l2 giving rise to one CO band (2184-2187 cm-l) after activation at T -= 473 K and two CO bands (2188 and 2208 cm-l) after activation at T 2 473 K. These bands reach maximum intensity following complete surface dehydration, and for still higher temperatures slowly decline in a parallel way owing to sintering of the material.Fig. 7(b) shows that a sulphate-free rehydrated TSK sample behaves quite differently: after dehydration at 473 K there is only the 2188 cm-l CO component, as on TS, and with an intensity comparable to that of TS. However, this CO band remains the sole component even on activation at higher temperatures; its intensity starts to decline long before complete surface dehydration, and reduces to zero for CO adsorption after thermal treatment at T 2 923 K. Volumetric measurements indicate that the decrease in surface area with activation temperature is much the same2658 t % e E 0 ." ." E u I.R. STUDY OF THE TiO,/K,O SYSTEM 2250 2150 2250 2150 wavenumber/cm-' FIG. 7.-Interaction of CO (5.3 x lo3 N m-*) with a sulphate-free TS sample (a) and a sulphate-free TSK sample (b) after outgassing at: S1, 473; S2, 673; S3, 773 and S4, 973 K.on TS and TSK, so that the above data suggest the following. (i) No CO adsorption occurs on the K,,, Lewis-acidic sites, whose production occurs at temperatures > 298 K and continues above the temperature of complete dehydration. This is not unexpected, in view of the weak polarizing field produced by alkali-metal ions. (ii) Consistent with what has been suggested by Py adsorption, the production of K,,, Lewis-acidic sites progressively destroys the strong Ti,,, Lewis-acidic sites ; this process starts at medium activation temperatures for the fraction of Ti,,, sites responsible for the 21 88 cm-1 CO band, but is complete at any activation temperature for the Ti,,, sites responsible for the 2208 cm-l CO component, i.e.for the sites that are formed last upon TS dehydration and for which a more polarizing field causes a higher upward shift of the CO stretching frequency. CONCLUSIONS The present study indicates that K,O, added in small amounts to anatase gels, mostly collects at the surface at any stage of thermal treatment. In fact the spectrum of surface sulphate impurities is modified to an extent that cannot be justified by a 1 wt % concentration of K,O, if homogeneously distributed. Although surface hydroxyls on TSK do not reveal appreciable spectral differences from those on TS, owing to poor resolution in the relevant spectral region, they show in different ways their interaction with added K,O, thus confirming the surface location of potassium ions.In particular, unlike TS, all surface OH groups must be eliminated for hydrogen bonding with adsorbed Py to disappear, and reasonable amounts of surface hydroxyls must be eliminated for potassium ions to reveal themselves through direct interaction with suitable test molecules.C. MORTERRA, A. CHIORINO, G. GHIOTTI AND E. FISICARO 2659 The surface location of K20 is confirmed by its easy extraction from anatase powders in water suspension. At the concentration levels employed here and following the thermal treatments which we adopted, no K20 segregates as a separate phase. If this were in fact the case, changes in the activity towards CO, adsorption would be expected, and these were never observed.The presence of K20 mostly affects, in a destructive way, strong surface Lewis acidity which was previously ascribed to Ti,,, ions. Even if K 2 0 is present at the surface at any stage of the preparation, high-temperature treatment is needed to affect appreciably strong Lewis acidity. In fact Ti,,, acidity, mostly produced by the cationic centres responsible for undissociated water coordination, manifests itself in the ‘normal’ way of a K-free sample after a thermal treatment at low temperatures, but at higher temperatures is progressively replaced by a weaker Lewis acidity, owing to K,,, ions, while the remaining part is rendered more and more vulnerable to evacuation. The interaction with CO, that can distinguish between the two Ti,,, families of sites whereas Py can not, and the interaction with water are vital in order to confirm that the elimination of Ti,,, strong Lewis sites occurs for the strongest ones first.An extrapolation of the present spectroscopic data seems to indicate that, on the final product, no Ti,,, ions are exposed, and that the entire surface layer should be thought of as a layer of potassium titanate, partly reversible to water vapour in the absence of other additives. A rough calculation can be attempted in order to follow this process. Complete dehydration of anatase was shown to occur with the elimination of ca. 8 water molecules per nm2,3 leaving as many coordinatively unsaturated cations at the surface. If we assume that the K 2 0 is all at the surface and that the Ti02-K titanate surface transformation occurs through the direct substitution of Ti cations for K cations, a K,O concentration of 1 % and a surface area of some 100 m2 g-l (at ca.773 K) would mean 16% K surface ions, and a surface area of some 15 m2 g-l (at ca. 1023 K) would mean 100% of surface K ions. The roughness of this calculation is evident, especially in comparison with the spectral data, which indicate a small residual Ti,,,/Py activity after treatment at temperatures as high as 1023 K, but the overall trend is confirmed. As the final pigmentary product, ready for the grain-coating process (to be dealt with elsewhere), has a surface area of ca. 2-4 m2 g-l, this calculation tells us that a K,O concentration of ca. 0.2% (the K 2 0 Concentration usually adopted) is sufficient for a complete surface-layer transformation. Finally, the present results indicate what is the most probable surface chemical composition of the material undergoing the next preparation step (i.e. the coating step), and also suggest that what actually controls the mechanism of crystal growth and the process of grain-shape modification during preparation is the nature and concentration of the surface acidic centres. C. Morterra, A. Chiorino, A. Zecchina and E. Fisicaro, Gazz. Chim. Ital., 1979, 109, 683. C. Morterra, A. Chiorino, A. Zecchina and E. Fisicaro, Gazz. Chim. Ital., 1979, 109, 691. C. Morterra, G. Ghiotti, E. Garrone and E. Fisicaro, J. Chem. Soc., Faraday Trans. I , 1980,76,2102. C. Morterra, A. Chiorino, F. Boccuzzi and E. Fisicaro, 2. Phys. Chem. (N.F.), 1981, 124, 21 1 . K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordination Compounds (J. Wiley, New York, 1978), p. 241. L. J. Bellamy, Advances in Infrared Group Frequencies (Methuen, London, 1968), p. 224. H. Knozinger, Adu. Catal., 1976, 25, 222. C. Morterra, A. Chiorino, G. Ghiotti and E. Garrone, J. Chem. Soc., Faraday Trans. I , 1979,75,271. C. Morterra, S. Coluccia, A. Chiorino and F. Boccuzzi, J . Catal., 1978, 54, 348. lo J. W. Ward, J . Catal., 1968, 10, 34. l1 T. A. Egerton, A. H. Hardin and N. Sheppard, Can. J . Chem., 1976, 54, 586. M. Primet, J. Bandiera, C. Naccache and M. V. Mathieu, J. Chim. Phys., 1970, 67, 535. (PAPER 1 / 1425)