The Modification of Fine Powders by Inorganic Coatings+

This review describes the modification of fine particles by the formation of an inorganic coating. It shows how a technology, which has been developed in, but mainly confined to, the titanium dioxide pigment industry, has considerable potential in other areas. Inorganic coatings may be used to encapsulate a solid or modify its dispersion properties. They may also be used to impart desired properties, e.g. electrical conductivity or to obtain specific optical effects. In ceramics, inorganic coatings offer a way of distributing minor components, such as stabilisers for zirconia or toughening agents for alumina, in a way that does not depend on the vagaries of mixing. Examples of different coating procedures are presented, and for two of these - dense silica and silica-alumina - the mech anism is briefly described. The uses and limitations of gas adsorption, electron microscopy, elec trophoresis, ESCA and SIMS for coatings characterisation are summarised. It is shown that each is useful for coating characterisation but none can unambiguously differentiate between uniform and patchy coating. Finally, specific examples of improved dispersion, improved barrier performance and beneficial powder properties are presented.


Introduction
Fine powders may be used to modify the properties of a continuous organic medium -for example titanium dioxide can be an opacifier in paints and inks, or aluminium hydrates a flame retardant in plastics. Alternatively, powders may be the constituents from which ceramic or other components are fabricatedfor example barium titanate for multi-layer capacitors or silicon nitride for the fabrication of turbo-charger rotors. In each case the powder is selected because of its desirable bulk properties -the high refractive index of TiOz, the endothermic evolution of water from aluminium hydrates, the dielectric properties of barium titanate or the high temperature strength, coupled with low density, of silicon nitride. However, to optimise its opacity the titanium dioxide must be dispersed in the paint and resist flocculation during film formation. The aluminium hydrate must be incorporated into the polymer. Barium titanate may have to be dispersed into an ink to allow screen printing, and the silicon nitride requires a sintering aid to promote densification of the fired component. Hence, as is normally the case, the processing of these powders is controlled by their surface properties. Modification of powder dispersion by the use of organic surface active molecules is widely practised, but this review describes an alternative means of surface modification-the deliberate formation of an inorganic surface layer, or coating. It also demonstrates that modification of particle surfaces by the formation of inorganic coatings has many uses other than dispersion control and some of these applications are demonstrated by the following examples. Although many of the examples above have been drawn from the literature of pigment technology, the aim of this review is not to discuss pigment technology. Rather, it is to draw the readers attention to the possibilities that inorganic coatings may provide in many other application areas.

Electrically Conducting Coatings
In contrast to the two previous examples in which the coating is a barrier, it may itself contribute to a desirable property. Thus, inorganic coatings are used to manufacture electrically conductive powders, suitable as anti-static pigments or for the formation of conductive layers on paper. The challenge is to manufacture a product that is conductive but not black. Antimony-containing tin oxide is conductive but it cannot by itself be produced as a powder with uniform particle size and good dispersibility. Also, as the antimony content is increased the colour becomes bluish and less white. One solution to this problem has been to coat titanium dioxide with antimonycontaining tin oxide and thus combine conductivity, dispersion and whiteness. The resistivity of a 10% (Sb+Sn)02 coated powder is five orders of magnitude less than that of a physical mixture of the components (8).

Novel Optical Effects
Coatings also contribute directly to desired effects in pearlescent, or 'effect' pigments which are manufactured by forming an iron oxide or titanium dioxide film of carefully controlled thickness on substrates of mica or aluminium flakes. A portion of incident light is reflected at the coating surface and a further portion is reflected at the substrate. The consequent interference gives rise to an appearance which depends on the thickness of the coating and on the angle from KONA No. 16 (1998) which it is viewed. If the substrate is mica, different colours may be seen in reflection and transmission and if the coating itself absorbs light, e.g. Fe203 on aluminium flake, the highest colour brilliance is observed when the interference colour is close to the absorption colour of the coating. It is important that coatings of this type are relatively smooth and if they are formed by precipitation they may be subsequently calcined (9). Alternatively, the coating on, e.g, aluminium flakes may be deposited in a fluidized bed by chemical vapour deposition (10).

Electrorheology
Certain suspensions of polarizable particles dispersed in insulating oils exhibit a rapid reversible change in the viscosity following the application of an electrical field. This phenomenon is called electrorheology. Composite particles consisting of an acrylic polymer core coated with titanium hydroxide particles have been shown to give excellent electrorheological suspensions (14).

Applications in Ceramics
Coatings of precipitated hydrous metal oxides have been found to be very effective binders for ceramic powders. They give high compacted densities and higher green strengths than those obtained with organic binders (11).
The control of stabilising additives is a key to the successful production of partially-stabilised zirconia ceramics. Inorganic coatings have been found to be a very efficient way of controlling the amount and distribution of additions, such as yttria or ceria (12). During the firing process the stabilizer migrates from the surface of the ceramic into the zirconia particle. The derived ceramics have high strength, and sinter at 1350°C to >97% theoretical density (13). Such coatings may be formed by precipitation from aqueous salt solutions or by hydrolysis of alkoxides.

The Preparation of Coated Powders
The most usual means of forming a coated powder is by the precipitation of an inorganic hydrous oxide onto the surface of the particles to be coated. For reasons of convenience the coating constituents are usually designated by such terms as 'alumina', 'silica', 'zirconia' etc. This should not be take to imply the existence of a discrete oxide phase -it is merely a convenient shorthand for the hydrous oxide, silicate or phosphate that has been precipitated. Unless the coated particles are subsequently deliberately calcined, or dried at high temperatures, the coatings are heavily hydrated and it is therefore not surprising that they are non-crystalline. In some cases an anhydrous layer may be formed on a substrate particle, e.g. by a solid gas reaction in a process of chemical vapour deposition. However, even in these cases the coating is usually amorphous. Figure 1 shows a typical coating on the surface of crystalline titanium dioxide. The contrast between the crystallinity -exemplified by the clear lattice images -of the substrate and the amorphous coating is clear.
The following sections provide an introduction to a few of the many methods that have been used to modify the surface of fine inorganic powders, and for the two most important classes a brief description of the coating mechanism is included.

Dense Silica coatings
Perhaps the most widely applicable method of applying an inorganic coating to a substrate is Iler's 'dense-silica' method. (16,17,18). The original patent describes examples of the formation of dense silica on attapulgite, haloysite and kaolin clays, chrysotile and crocidolite asbestos, vermiculite and talc, iron powder, nickel and aluminum flake, glass fibres, cotton and nylon cloths, and on titania pigments. The method is illustrated by the following two examples. High resolution electron micrograph of a rutile crystal coated with silica-alumina type coating. The contrast between the crystalline substrate and the amorphous coating is evident. 48 Dense Silica on Aluminium Flake Aluminium flake was pre-treated in 0.25% sodium silicate solution and was then suspended in water at 2.4% solids concentration. Sufficient sodium sulphate was added to make the solution 0.2M in sodium ion. The suspension was then heated to 95°C and 1000 cm 3 each of a sodium silicate solution (4% Si02 and 1.25% Na20, equivalent to 0.33 gms of silica per gram of aluminium) and 1.57% sulphuric acid solution were added simultaneously, but separately over a period of 2 hours during which the pH hardly changed (9.3 at start, 9.5 at end). The final free-flowing product was 15.6% Si02 and had a surface area of 40 m 2 /g (compared with 51 m 2 /g. initially). One criterion of the applied coating being 'dense', i.e. not porous to nitrogen, is that coated particles whose core density is comparable with that of silica should have a surface area that is the same as, or lower than, the un-coated material.

Dense Silica on Titania
In another example, a dense silica coating was applied to a 200 g.p.l dispersion of rutile in water at pH 10. The slurry was then heated to 95oC and to the vigorously stirred suspension llitre each of sodium silicate (4% Si02 and 1.25% Na20) and of 1.57% sulphuric acid were added separately but simultaneously at a rate of 15 ml/min. During this addition the pH fell to 8.7. The suspension was filtered, washed and the pH adjusted to 6.5 before further washing and air drying. An electron micrograph of a titanium dioxide particle coated by a dense silica method is shown in Fig. 2. It shows clearly the dense skin of substantially uniform thickness bound to the core particle and composed of amorphous hydrated silica. Iler demonstrated that a key advantage associated with this skin is the reduction of the photo-activity of the resulting titanium dioxide pigments.

Mechanism of Dense Silica Coating
Although the steps in this mechanism are broadly understood their relative importance probably varies between different modifications of the process. However, it is probable that the first step in the coating is the adsorption of silicate on the surface of the substrate particle (e.g. in the coating of aluminium flake) and that this adsorbed material provides the surface on which subsequent coating occurs.
It is generally considered that the next step in the mechanism of the dense silica coating process is the generation of silicic acid, Si(OH)4. This may be formed by the addition of silicon tetrachloride to water, by the ion exchange of sodium silicate solution, or by the addition of acid to sodium silicate solution. However, at pH 9 the solubility of Si(OH)4 is only ca 0.01% (expressed as Si02) at 25oC (ca 0.04% at 90°C) and if its concentrations exceed a few hundredths of one percent it polymerises to form silica particles (19). However, provided that the particle growth is not allowed to continue, this polymerisation is reversible and Iler considered that the 'active silica' which is a precursor to dense silica coating contains both monomeric silica and particles up to 1-2 nm diameter. Further, since silica particles smaller than ca 5 nm are more soluble than large particles, large particles-e.g. substrate particles with an initial layer of adsorbed silica-would be expected to grow relative to discrete 2 nm particles by a process of Ostwald ripening -that is surfaces with a small curvature (large particles) grow at the expense of particles with a high curvature (small particles).
Therefore, putting these observations together, Iler concluded that dense silica coating occurred as a result of molecular deposition of Si (OH)4. If the sodium silicate and acid are added too quickly the 2 nm silica particles will grow, the reverse reaction will become more difficult, and at ca 5nm the driving force for Ostwald ripening will be lost. KONA No.l6 (1998) Depending on the ionic strength and the pH the silica particles would then either form stable colloidal silica or form a gel and, at best, this will represent an inefficient use of the coating silica. The aim is to have sufficient super-saturation to coat the substrate without having so much super-saturation as to form a silica sol.
The choice of reaction conditions follows. The pH of 9 is sufficiently high to minimise the tendency of the 2 nm particles to form gels but is below the level, 10.5, at which the silica will redisolve as silicate ions. Further, at pH9 the base catalysed depolymerisation and condensation reaction rate is at its maximum and this will increase the rate of Ostwald ripening. Also, the rate of Si (OH)4 transfer is accelerated at high temperature because of the increased solubility of silica. Iler (17) gives formulae relating the maximum permissible silica deposition rate to the temperature and the total surface area to be coated : where A is the specific area of the substrate (m 2 1 g) n= (T-90)/10 where Tis the temperature (K). and S is the amount of Si02 added (g/hr per g. of substrate).

Developments of the Dense Silica Method
This 40 year old method still forms the basis of a wide range of coatings and has been developed extensively. In this summary I restrict myself to its DuPont descendants. Alexander (20) has described an extension of the method to produce silica coatings which contain chemically combined polyvalent metal ions. For example aluminium may be incorporated into a silica coating by simultaneously adding sodium silicate and sodium aluminate to a suspension of fine particles kept at a pH between 10 and 11. Zinc or copper may be incorporated by dissolving zinc sulphate or copper sulphate in sulphuric acid and then simultaneously adding the silicate and acid solutions. Other examples describe the incorporation of lead, magnesium and iron. Werner (21) has described the preparation of a titanium dioxide pigment with the improved durability resulting from a dense silica coating, but which, as a consequence of a subsequent precipitation of alumina, does not suffer from the poor dispersion associated with single component silica coatings.
In this case the dense silica was deposited by first adding sodium silicate and subsequently raising the pH. The process described by Werner is slow and West (22) has described a faster method, suitable for continuous production. The essence of this faster process is increased pH and higher temperatures.
Other patents in this family claim the benefits of incorporating small amounts of boron at the dispersion stage (23,24). Many of the other developments of the dense silica coating procedure focus on making the process more convenient at an industrial scale or on optimising pigmentary properties other than durability.

Silica Alumina Coatings
The molecular deposition of silica is believed to be central to the formation of dense silica coatings. However, the deposition of alumina-silica coatings can occur by a quite different route.

The Mechanism of Silica Alumina Coating
Howard and Parfitt (hereafter HP, 25) and subsequently others (26,27) have provided insight into the mechanism of silica/ alumina coating precipitation. The first stage in this process is the adsorption of polysilicate anion. HP described the addition of sodium silicate solution (containing the equivalent of 100 gm SiOz/Q) to a slurry of Ti02 at pH 3. Silicate adsorption continued throughout the addition as the pH increased to 10.5 but at maximum adsorption (0.5% SiOz) only one quarter of the added silica was adsorbed on the substrate. This adsorption caused no significant change in surface area but was demonstrated by a decrease in the isoelectric point of the substrate from pH 3.5 to 2.7. Since the non-adsorbed silica was present in amounts considerably greater than the known solubility of silica at these pH sit was inferred that much of the unadsorbed silica was present as multimeric anions in stable dispersion.
The second stage of this coating process is the acidification of the titanium dioxide/silicate slurry by the addition of an aluminium salt. In the absence of aluminium salts, the acidification of the Ti02/sodium silicate slurry simply causes the progressive desorption of silicate species from the titanium dioxide surface. In the absence of silicate, the addition of aluminium sulphate to an alkaline titanium dioxide slurry rapidly lowers the pH. At pH 10 the slurry becomes saturated with respect to aluminium salt and, therefore, during further addition of aluminium sulphate practically all the added aluminium is precipitated. This continues until the pH had dropped to ca. 3. At this pH the aluminium becomes soluble once more and therefore as the pH falls from 3.0 to 2.8 the additional aluminium sulphate is neither precipitated nor adsorbed -even though for a total of 2.8% (based on TiOz) most of the aluminium is added over this range. Figures 3 and 4, derived from HP's results, summarise these changes 50 and contrast them with the addition of aluminium salts to the titanium dioxide/silicate slurry. The addition of aluminium nitrate or aluminium sulphate to the silicate/titanium dioxide slurry caused the precipitation of all the aluminum ions, and ca 50% of the silicate, from solution at pH 9 to pH 10.
The remaining silica is precipitated from colloidal solution at pH 3.8 to 4, the point at which soluble aluminium cationic species become stable. HP con-  The Adsorption of Silica accompanying the reduction of pH during the second stage of Silica Alumina Coating. When the pH is reduced with acid, previously adsorbed silicate desorbs. If the pH is reduced by the addition of aluminium sulphate there is considerable adsorption of silica. eluded that to a first approximation the precipitation of the aluminium is governed by the solubility of aluminium hydroxide. The first stage of this precipitation is the formation of hydroxy aluminium polymeric species and these heterocoagulate with the anionic (poly-)silicate species. Thus both silica and alumina are precipitated. The maximum in the silica precipitation may correspond to the formation of higher molecular weight polymeric aluminium species at pH 6 to pH 4. Similarly, when the pH is lowered below 3.5, the polymeric aluminium cations, particularly those with nitrate counter-ions, de-polymerise to form soluble cationic aluminium species and release some of the hetero-coagulated silica. (To demonstrate this point, Fig. 3 and 4 include points for which the pH was lowered below 3 by nitric acid or sulphuric acid addition.) The resulting sol particles, aggregated at the titanium dioxide surface, form, after drying, a surface coating. The electrophoretic mobility of samples extracted from suspension at different stages of the coating process suggest that the external surface is silica rich.
During the third stage of this coating the pH is raised by adding alkali. Precipitation of the alumina and small amounts of silica in solution is essentially completed between pH 4.5 and 5.0 and the surface area shows a minimum at ca. pH 4. Further increases in pH remove sulphate from the coating, presumably as a result of hydroxide ion exchange, and increase surface area. After 'neutralisation' to pH 8.5 these silica-alumina coated titanias have surface areas of 16-18 m 2 I gm, significantly greater than those of the dense-silica analogues.
A recent study (28) of coprecipitation of hydrous alumina and silica precipitated in an aqueous neutral suspension of titanium dioxide showed that the surface area decreased from 40 to 30 m 2 /gm as the preparation time increased from 1 to 15 minutes.

Silica-Alumina Coating on Titania.
In a representative example of this procedure(26) sodium silicate solution (SiOz: Naz0=3.2) was added to a 300 g.p.l. suspension of TiOz particles at 60°C. The stirred slurry was held at this temperature for 30 minutes during which adsorption of silicate occurred. Aluminium was then added as a solution of aluminium sulphate. During the initial stages of this addition the pH fell to ca 4 and complete precipitation of the residual silica and of the added alumina occurred. Further aluminium sulphate addition lowered the pH further, without further alumina precipitation but with dissolution of a small portion of the precipitated silica. The KONA No. 16 (1998) slurry was then neutralised with NaOH and this caused complete precipitation of alumina and silica in solution.

Other Silica-Alumina Coatings.
A very large number of silica-alumina coating variations have been described and used and it not possible to summarise them all here. The most important variation is the use of sodium aluminate in place of aluminium sulphate and mechanistic studies of this coating route have been described (26,27). High pH probably facilitates the formation of the pseudobohemite phase in the coating alumina (29). Other variants have include the incorporation of titania or zirconia in the coatings (4) and the use of phosphate dispersants rather than sodium silicate (30).

Alkoxide-derived Coatings.
The methods described above have all made use of conventional inorganic reagents, aluminium sulphate, sodium silicate etc. It is also possible to coat from alkoxides of e.g., aluminium, silicon or titanium, and methods of this sort have been attractive for academic studies because they eliminate the counter-ions which can otherwise increase the ionic strength of the suspension (and hence cause flocculation of the coated particles) or else be incorporated into inorganic coatings (as with the sulphate ions derived from aluminium sulphate).
The coating of gold sols by a two stage process involving first the hydrolysis of an alkoxide (a Stober silica step) and subsequently an Iler type dense silica step has been described, together with the optical properties of the resulting sols (31).

Coatings on Ceramic Particles.
A number of studies of dense silica deposition on alumina have been published (18,34) but there has been much less effort directed towards understanding the mechanism of other coatings on ceramic particles and the coating procedures are much less well developed.
In a very simple example (ll), focused on production of zirconia toughened alumina, alumina (Sumitomo AKP 30) was ultrasonically dispersed in water and the pH adjusted to 0.8 by the addition of nitric acid. Sufficient zirconyl sulphate solution to give 15% zirconium oxide was then added drop by drop, with vigorous stirring. The pH was raised to 8.5 with ammonia solution and was held at this level for 20 minutes. During the addition hydrous zirconia was precipitated on the alumina particles. The coated product, after that the coating was well distributed but very fluffy and density measurements by gas pycnometry showed the density of the coated particle to be significantly lower than the uncoated substrate. It is clear that the voluminous coating is heavily hydrated and this was confirmed by both thermogravimetric analysis and surface area measurements after calcination at increasing temperatures. The initially amorphous coating became crystalline after calcination and microscopy showed a number of zirconia islands distributed over the surface of each alumina crystal. Similar experiments have been reported using zirconyl chloride as the starting reagent (32). A preparation using zirconium alkoxide hydrolysis also led to a fluffy coating and increased surface area from an initial 12 m 2 I gm to 100 m 2 I gm for the coated sample (nominally, 20% zirconia) (33). As with the samples prepared by hydrolysis of zirconyl sulphate, the alkoxide-derived samples showed significant weight loss during firing and continued to lose weight up to soooc (35).
The coating of zirconia particles with stabilisers such as yttria and ceria has been much more successful (12,13). Ceria stabilised zirconia powders have been coated with yttria or alumina using a homogeneous hydrolysis process in which the pH change is induced by heating an acidic solution of reactants containing urea (15).

Vapour Phase Coating or Chemical Vapour Deposition (CVD)
The previous four sub-sections have described the formation of inorganic coatings by controlled precipitation of the coating components from an aqueous phase onto the surface of fine particles in an aqueous suspension, or slurry, An alternative coating routeparticularly suited to the treatment of particles which have been formed in the gas phase -is by deposition from the vapour phase.
As might be expected there is a voluminous patent literature (e.g. 37) but the elements of the method are summarised in recent papers (38,39) which describe the gas-phase coating of titania with metal oxides. The coating precursors (e.g. SiC14. AlCb, ZrCI4) may be introduced, either singly or in combination, into the vapour phase down stream of a TiCI4 oxidation reactor. Introduction of SiCI4 at 1500°( gave relatively uniform silica coatings 5-20 nm thick. Similar uniform coatings could be achieved at 1300oC if some water vapour was also added to the reactor. Results were reported for silica, alumina, zirconia, silica/alumina and zirconia/ alumina coatings and simple models of the competing processes were proposed (40). In general, the gas phase oxidation of SiCk AlCb and ZrCI4 is slower than that of TiCI4 and it is likely that the coating is achieved by reaction of the coating precursor with the already formed TiOz surface rather than by initial gas phase oxidation of SiCI4 to SiOz followed by deposition on the TiOz surface. At lower temperatures, exchange reactions of the type AlzCI6 + TiOz -+ AIOCI + TiCI4 have been shown to occur by X-ray identification of the AIOCI (41) and the kinetics studied.
Commercial application of these methods would eliminate the need for first wetting and then drying titanium dioxide pigment. However, despite the apparent attractiveness of a simpler process, as described in numerous patents (37,42,43), the route has not been commercially exploited, most probably because of the practical difficulties of neutralising the acid surface that results from the oxidation of metal chlorides, the only economical coating precursors for bulk chemicals. However routes involving more expensive coating precursors -e.g. decomposition of alkanoyl oxysilanes (44) -have been studied for specialist applications.
An interesting recent variant on vapour phase coating is a report of the in situ synthesis of palladium powders coated with, e.g. ZnO and CaO by a spray pyrolysis method. (45).

The Characterization of Coated Powders
An ideal coating process would lead to each particle of material A acquiring a uniform coherent coating of component B. Alternatively B will coat A, but will do so unevenly, so that, instead of a uniform coating, islands of B will form and uncovered, bare patches of A will persist. By contrast, a totally unsuccessful coating procedure will lead to the precipitation of B as a separate phase and will leave the entire surface of A uncoated. (See Fig. 7) The possibilities become more complex if a multi-component coating is required, and it is therefore not surprising that the characterization of coated powders remains a significant challenge for the materials scientist.

Surface Area and Adsorption
For the development technologist the simplest method of characterization is usually surface area determination and the most useful is electron microscopy. The specific surface area (SSA) of a spherical particle of radius rand density p is given by If a truly coherent non-porous coating has been achieved the SSA of the coated particle will, to a first approximation, be (SSA)coated = 3/(r. p [ l+x/300]) where x is the amount of coating, expressed as weight %, and it is assumed that the density of the coating is equal to that of the underlying particle. In theory, the coated sample will thus have a lower surface area than the uncoated parent material as exemplified by 40 m 2 /gm for dense silica on 51 m 2 /gm aluminium flake. Dense silica coatings on 7 m 2 /gm TiOz typically have a surface area of ca 10 m 2 I gm as measured by nitrogen adsorption. By contrast silica alumina coatings, of the type described earlier, often have areas in the range 15-20 m 2 I gm. An increase of surface area of 10 m 2 /gm can easily result from 5% of coating. It follows that the coating must have an area of ca. 200 m 2 I gm and most of this must be internal area-that is, the coating must be porous to nitrogen. It is therefore unsurprising that the changes in surface area accompany the incorporation and replacement of sulphate ions during the neutralisation of silica alumina coatings (Fig. 8).
The use of adsorptives other than nitrogen gives additional information. Thus it has been shown that, for n-pentane, the B.E.T. 'C constants' (a measure of

Electron Microscopy
The primary method of assessing coating quality is electron microscopy. Figure 2 shows a transmission electron micrograph of a silica coated titanium dioxide. The coherent coating layer can be seen clearly. However, because transmission microscopy of this type measures through the particles it averages thicknesses which may vary over the depth dimension. Thus, the appearance of uniformity is enhanced. Fig. 9 shows a scanning electron micrograph of a pigment which has been coated with both dense silica and alumina. The underlying, darker areas, in this micrograph are believed to be due to silica which has formed a coherent coating on the TiOz. The lighter patches are due to alumina which has formed patches on the underlying silica even though the alumina is present in amounts which, if evenly distributed, would form a coherent overcoat over the silica. Both the silica only and the silica plus alumina coatings were prepared using a dense silica method of the general type described above. 54 Fig. 9 Scanning Electron Micrograph showing alumina patches on a dense silica coating.
The major limitation of the electron microscopy is that only a very few crystals are examined. One gram of a fine powder contains about 10 14 crystals of 100 nm radius, but it is highly unlikely that more than a few hundred crystals will be examined in a high resolution t.e.m. study. Therefore, although it is tempting to interpret micrographs, such as Fig. 2, as demonstrating a successful coating, they can easily mislead the optimistic development scientist. They are particularly difficult to interpret if the coating level is low. However, microscopy will detect both changes in the texture of the coating and the formation of bulk precipitates.

Electrophoresis and Point of Zero Charge
The next characterization method in the coating technologist's armoury is the measurement of electrophoretic or acoustophoretic mobility of the coated particle as a function of pH. Unlike the other methods considered here, this technique does not need the sample to be exposed, with risk of surface modification, to vacuum conditions. Surface ionization and isoelectric points vary from oxide to oxide and therefore this measurement may be used to infer surface composition. Further, because the surface potential controls the colloid stability and hence the rheological behaviour of the modified particles, the measurement is directly relevant to their behaviour in many applications, particularly if the measurements can be made at high particle concentration, e.g. by using acoustophoretic methods.
Pure silica has a much lower point of zero charge (p.z.c.) than pure titania. Therefore it would be expected that a silica coating would lower the p.z.c. Figure 10, based on results for silica on a pure rutile (48) shows that this is indeed the case. Deposition of 2.5% dense silica on pure titania of 20 m 2 I gm gave isoelectric points judged to be consistent with pure silica. However, to obtain reversible mobility /pH curves which did not exhibit coating-solubility of the coating at low pH, a 5% coating was necessary.
The variation of isoelectric point with coating composition has also been reported and Figure 11 shows measurements of the point of zero charge, taken from Losoi (27)  KONA No. 16 (1998) as a function of the fraction of alumina in a silica alumina coating. Both sets of results show a general increase in the p.z.c. as the fraction of alumina in the coating is increased. Losoi's results also show that the coating sequence can be important: silica followed by alumina, gave a higher p.z.c. than alumina followed by silica, giving credence to the belief that some layering of the coating had resulted from the sequential deposition. The p.z.c plateau for the four Parfitt coatings that were richest in alumina is probably a consequence of the increase in weight fraction being almost entirely controlled by a decrease in silica-the actual alumina analyses were all close to 2.2%. Zeta potential curves for alumina coated with zirconia were similar for 1:8 and 1:1 zirconia alumina compositions and were taken to provide good evidence that the zirconia coats the surface of the alumina (32) but the possibility that the measurements were simply dominated by the zirconia because of its much higher (x8) surface area was not explicitly considered.
In summary, there is good evidence that electrophoretic measurements can differentiate between major changes of coating type.

Surface Analytical Methods
Since none of the classical surface characterization methods outlined above is able to answer all of the questions that can be asked about coating quality, a number of attempts have been made to use modern surface analytical techniques. These include X-ray Photo-electron Spectroscopy (XPS) often known as ESC A, and Secondary Ion Mass Spectrometry (SIMS). A limited number of Auger studies have also been carried out (26).
XPS measures the energy of electrons that are ejected from the solid by incident X-ray photons. The electron energy is characteristic of the element from which the electron is ejected, and the probability, P, that an electron originating from a depth, t, within a solid will reach a detector at an angle a to the normal to the solid surface is given by where K is a constant and A. is a characteristic escape depth (of the order of 1-2 nm for the cases of interest). As approximately 95% of the emitted electrons originate from a depth of 3A. (3-6 nm) from the solid surface, ESCA signals are dominated by the surface. For typical coatings the ratio of (coating element)/ (substrate element) may be two orders of magnitude higher than those calculated from bulk analyses (27,50). If the coating had simply precipitated as particles similar to the substrate particles the XPS ratio would approximate to the bulk ratio. Therefore, a high ratio confirms that the a coating rather than a bulk precipitate has formed.
However, for a 200 nm diameter particle, the outermost 5nm accounts for 15% of the mass. Since most coatings are less than 15% of the substrate mass, they are equivalent to less than 5 nm thick. (Because the coatings are porous the geometrical thickness may be greater than this, but there is a corresponding decrease in atom density.) As the ESCA sampling depth is typically 3-6 nm, it follows that at normal coating levels the underlying substrate can contribute significantly to the measured signal even when perfect encapsulation has been achieved. This conclusion has been confirmed experimentally for dense silica coatings on both alumina (18,34) and titania (50). Thus although ESCA is highly surface sensitive, it is not sufficiently sensitive to readily differentiate between, e.g., a 3% coating spread evenly over a 200 nm particle, and the same amount of material covering only 50% of the surface with a coating that is twice as thick. (If A.=2 and the density of the coating is half that of the core, the core signal for the fully coated case falls to ca. 56% of the uncoated value, whilst the signal for the half coated core is 37% of the uncoated value) For planar surfaces greater surface specificity may be achieved by collecting the electrons that escape at large angles from the normal -that is by increasing the path length of solid through which they must travel. However, for an assembly of fine particles this angular dependent surface enhancement is negligible because of the complex geometry of the real surface (50). An alternative approach has been to use an ion beam to etch away the surface layer and measure the increase of substrate signal and decrease of coating signal as the upper layers are removed. This approach also is complicated for fine powder assemblies. The process is analogous to slicing the top from a box full of oranges -not matter how deep you slice, some skin will always be apparent. However, theoretical etching profiles have been calculated and compared with experiment (51). Despite these sophistications the overall conclusion remains: ESCA, like the classical techniques, will show gross coating differences but is less well suited to monitoring the practically important small deviations from perfect encapsulation by a thin coating.
Secondary Ion Mass Spectrometry measures the ions sputtered from a surface by a beam of incident ions. The technique is significantly more surface 56 sensitive than ESCA and for a 4% silica coating on a titanium dioxide signal the SIMS Si:Ti ratio was 50:1 whilst the corresponding ESCA Si:Ti ratio was 10:1 Therefore SIMS is probably better suited than ESCA to studies of coating integrity. Although quantification of SIMS is extremely difficult it has been used with some success in studies of the mechanism of coating alumina with silica (18).

Examples of Specific Effects
In the introduction to this review various applications of coatings were indicated. In this final section three of these examples are described in more detail.

Coatings for Ceramics
The simplest example of the effect of surface modification on powder properties may be observed in the properties of the powders themselves. Both the bulk density and the cohesivity of TiOz are modified by the formation of a surface layer of, e.g., alumina or silica and similar effects are observed when zirconia is coated with yttria.
A beneficial application is found in the increased strength of compacts of coated particles in comparison with the uncoated starting materials (11). The first stage of ceramic processing is the preparation, by pressing, of a 'green' body with the form of the desired final article. These bodies must have a strength adequate to withstand the mechanical disturbance associated with transport within the factory, loading into furnaces etc. Typically the powders are compacted with a polymer binder, e.g. PVA, in order to obtain the required strength. However, these binders must be burnt out during firing and this can be a rate determining step in the kiln cycle. When zirconia-coated alumina ceramics, of the type described earlier, were isostatically pressed at 400 MPa the resulting compacts had strengths, measured in 4point bend, of 13.5 MPa. This is significantly higher than the figure of 0.8 MPa for the same alumina coated with a PVA binder. Uncoated powder compacted at 400 MPa was too weak to be measured by this method. Similar results have been demonstrated for both titania and magnesia powders.

Dispersion
Coating affects the dispersion of particles in three ways. First it reduces the Van der Waals interactions which are responsible for the attractive forces between colloidal particles. For a constant electrostatic repulsion and steric stabilisation, this leads to a greater energy barrier to flocculation and therefore stability is increased. The effect is particularly significant for core particles of a high Hamaker constant, e.g. iron oxide or titanium dioxide, coated by a material of low Hamaker constant, e.g. silica or alumina. For such particles in a 5mM 1:1 electrolyte, calculations (assuming a constant surface potential of 30m V) show that coatings of even 1 or 2 nm can increase the colloid stability by 10-20 kT and can therefore turn an unstable system into a stable one. Of course if one material is coated by another the surface potential at fixed pH will not remain constant because the number and type of ionizable groups will be altered, as demonstrated above for both silica ( Figure 10) and silica-alumina ( Figure 11) coatings. Therefore adsorption of surfactants and dispersing agents will also be changed and at a given pH adsorption of an anionic surfactant will drop to zero if the isoelectric point of the adsorbing surface is below the suspension pH, whilst the opposite trend will occur with a cationic detergent (52). In a key study of the effect of coatings on titanium dioxide pigments (53) it was shown that in alkyd paint systems the surface potential increased with increasing fraction of alumina in the coating. However, the magnitude of the increase was insufficient to account for the observed increases in opacity of the pigment. On the basis of adsorption from solution, it was demonstrated that the alumina coatings were more basic than the silica ones and, it was suggested, they could interact only with the few (two or three per polymer molecule) acid sites in the alkyd molecules. These molecules therefore adopted a looped configuration on the pigment surface. The more acidic silica surface was able to interact with the many basic groups (hydroxyls, esters) on the molecule which was therefore constrained to lie flat on the pigment surface. It was concluded that the increased pigment opacity was a result of steric stabilisation associated with the changed polymer conformation. Hence, in this example the chemistry of the particle surface -specifically its acid base natureis also affecting the particle dispersion.

Control of Photoactivity
Photocatalytic oxidation of alcohols by titanium dioxide has been widely studied. The reaction is believed to involve the UV photo-generation of electrons in the conduction band, and holes in the valence band of titanium dioxide. The holes migrate to the oxide surface where they combine with surface hydroxyl anions to form hydroxyl radicals, OH'. These radicals then abstract a hydrogen from the isopropanol. Hence, if the surface is modified by coating, KONA No. 16 (1998) the photocatalysis should be reduced because of a barrier between the hydroxyl radicals and the isopropanol molecules. Figure 12 shows how the activation energy for the reaction is increased as increasing amounts of coating are applied to a titanium dioxide catalyst. The increase in activation energy from ca. 48 to ca 75 kJ/mol occurs between 0.04 and 0.09% coating/m 2 of substrate and this range stradles the figure (0.07% coating/m2) at which Furlong et a!. (48) found that the p.z.c. of a silica coated pure rutile becomes relatively independent of coating level, as shown in Figure 10. This figure corresponds to coating thicknesses of ca 0.3 to 0.5 nm which may be compared with a distance of 0.16 nm for the Si-0 bond length in silica. Yet again the profound effect of very thin coatings is demonstrated.

Acknowledgements
The electron micrographs used to illustrate this review were kindly made available by, and are reproduced by permission of, Tioxide. Coating Level (%) Per Unit Area Fig. 12 The Activation Energy for Photocatalyzed Isopropanol Oxidation. The increased activation energy occurs at ca 0.06% silica/m 2 , the same value at which the p.z.c. became constant (Fig. 10).