J. MATER. CHEM., 1994, 4( l), 119-124 Chemomechanical Polishing of Lithium Niobate using Alkaline Silica Sol and Alkaline Silica Sol modified with Hydrogendifluoride Anion Margaret Beveridge," Laurence McGhee," Scott G. McMeekin,bt Max I. Robertson,b Alexander Rossb and John M. Winfield*" a Department of Chemistry, University of Glasgow, Glasgow, UK G12 8QQ Logitech Ltd, Old Kilpatrick, Dunbartonshire, UK G60 5EU Lithium niobate undergoes a slow reaction with anhydrous HF at room temperature to give LiNbOF,, which is readily hydrolysed to give Nb205 as one product. This sequence of reactions is believed to be responsible for the superior performance of aqueous alkaline silica sol modified by the addition of [HFJ-for the chemomechanical polishing of LiNbO, wafers, compared with silica sol alone.Lithium niobate is an important material for optoelectronic applications;' for example the fabrication of planar optical waveguides by hydrogen-lithium exchange processes involv- ing LiNbO, single-crystal wafers has received considerable attention.24 For many applications the device performance is determined by the quality of the surface finish. The trend towards smaller devices requires wafer surfaces whose rough- ness, the peak-valley variation, should approach 1 nm as measured by a stylus instrument. Careful control of surface polishing is therefore necessary. Polishing of optical materials has a long history, many of the guiding principles were formulated by Lord Ra~leigh,~ but in many respects it is still as much art as science.Studies of LiNbO, polishing have either focussed on practical aspects or have emphasised mechanical interactions between abrasive particles and the LiNbO, surface.8p'' It is increasingly appar- ent, however, from the large number of polishing studies involving silica and other oxide glasses, that chemical reactions at the surface may be just as important as mechanical action in determining the quality of the surface fini~h,'~.'~ hence the term 'chemomechanical polishing'. We have examined the effect on LiNbO, polishing of the hydrogendifluoride anion by comparing the results obtained using an alkaline silica sol'' (Syton) with those obtained under identical polishing conditions but with [HF,] -added to the reagent.The latter reagent results in a faster process without loss in surface quality. The chemical events occurring at the LiNb03 surface have been investigated using model experiments and the results used to suggest a mechanism for the polishing reaction. Experimental Materials Congruent lithium niobate wafers, X-, Y-, or Z-cut, (Pilkington Electrooptics) were cut to the required dimensions using a diamond saw and were edge lapped to prevent mechanical fracture during subsequent face-polishing operations. Congruent LiNb0, powder (Johnson Matthey) was used as received. The aqueous alkaline silica sol was a commercial material (Syton, Monsanto; nominal particle size 0.04 pm) Potassium hydrogendifluoride was prepared using the method of Clark et ~1.'~Anhydrous potassium fluoride (General Purpose Reagent BDH; 174g 3.0mol) was added to distilled water (300 cm3).The solid dissolved on stirring, then glacial acetic acid (Pronalys A.R, May and Baker, 300 g, t Present address: Department of Electronic and Electrical Engmeering, University of Glasgow. 5 mol) was added slowly with stirring. The resulting solution was allowed to stand at room temperature overnight, during which time a colourless crystalline solid formed. This was filtered and dried at 120°C. The total solid product obtained after 72 h was 35g, 30% based on the KF taken. (Found: HF, 23.6; KHF, requires: HF, 25.85%). Potassium hydrogen- difluoride (1.0 g) was dissolved in alkaline colloidal silica sol (125 cm3), stirring the mixture at room temperature until all the solid had dissolved.The pH was in the range 7-8. The reagent appeared to be stable, i.e. a colloidal solution, in use over 4 h. After longer periods however, coagulation occurred but the reagent was easily restored to its former state by vigorous shaking at room temperature. Retention of HF by silica gels has been noted previously." SEM examination of the solid obtained by evaporation of the colloidal solution showed it to be very similar to that obtained by evaporation of the silica sol alone, the only difference being a greater proportion of larger particles. However much of the fluorine may have been lost during the evaporation process." General Instrumentation IR spectra were recorded using PE 983 and Nicolet 5DXC spectrometers using a SpectraTech collector for diffuse reflec- tance spectra.Powder XRD measurements were made using a Philips PW1050-35 diffractometer with a vertical goniometer and Co-Kcr radiation, samples being mounted on adhesive tape. BET areas were determined using N, as adsorbate. Lithium in polishing solutions was determined by atomic absorption spectroscopy (AAS) using a Perkin-Elmer 1 100 spectrometer with a lithium hollow cathode lamp. Standard solutions were prepared from Spectrosol lithium nitrate solu- tion (1000 ppm) to give lithium concentrations in the range 0.25-2.00 ppm. Three integrated absorbance measurements were made for each solution and the relationship between absorbance and Li' concentration was linear over the range studied.Absorbances from the analyte solutions were deter- mined similarly and the lithium concentrations measured from the experimentally derived concentration us. absorbance relationship. Analyte solutions were prepared by taking an aliquot (25 cm3) of material collected from the polishing process and digesting it with aqueous HF (4070,10 cm3) in a PTFE beaker. The resultant solution was heated to dryness, digested with HNO,, evaporated to dryness and finally made up as a standard solution using aqueous HNO,. Duplicate analyses were carried out throughout. Polishing Procedure A set of 12 wafers (2 cm x 1cm) was cut from a large LiNb03 wafer, edge lapped and wax bonded in a cruciform arrange- ment to a glass puck, normally 83 mm diameter and 6 mm thick.This was attached to a Logitech PP5 polishing jig by means of a vacuum chuck. The polishing jig ensured that the samples were held parallel to the lapping and polishing plates. The samples were lapped on a Logitech PM2 precision polishing instrument which was fitted with a grooved, cast- iron plate, at 30-40 rpm using 3 pm a-alumina in water as the abrasive. The jig loading was ca. 3.2 kg. The lapping time was dependent on the initial height and condition of the samples, lapping being continued until all the samples were in a single plane. Surface-roughness measurements of lapped wafers were in the range 100-140 nm. Samples were polished using the same instrument with an expanded polyurethane plate (grade LP87) at 10 rpm, with 1.5 kg loading and alkaline silica sol with or without added KHF,, as the polishing fluid.Feed rates were ca. 100 cm3 h-l. The progress of a polishing reaction was followed by determination of stock removal, surface roughness and Li analyses of polishing fluids at regular intervals. Stock removal was calculated using a contact gauge to measure the change in thickness at the centre point of each wafer. The precision of these measurements was limited by the resolution of the gauge (0.1 pm). Surface roughness determinations were made on a Rank Taylor Hobson Talystep instrument, data being presented as the amplitude parameter, R, (defined as the arithmetic mean of the departures of the roughness profile from the mean line).R, values were calculated from data taken over a 0.5 mm trace at magnifications of 1 x 105-2 x lo6 depending on the degree of lapping damage that remained. The standard stylus was used for all traces, except after the final polishing time interval when the fine stylus was used. The quantity of polishing reagent used over a specified time interval was collected quantitatively and to this volume was added the washings from the wafers and polishing plate. Analytical data are expressed as total Li, i.e. the product of the total solution volume and the Li concentration as deter- mined by AAS. Replicate experiments indicated that there was no marked dependence of total Li determination on the volume of solution collected, thus the solubility of the Li-containing product was not a determining factor.The quality of surface finish obtained and the time required for polishing with alkaline silica sol alone proved to be very dependent on the procedure used for conditioning the expanded polyurethane pad. A grooved pad required frequent reconditioning to prevent a film of silica being formed on the surface, inhibiting transport of the reagent. A satisfactory, reproducible procedure was to scroll the pad on a lathe, using a 45" diamond tool to form a 0.5 mm deep spiral groove. This procedure was less important when using the silica sol-KHF, reagent but was usually adopted in order to achieve an objective comparison. Fluorine-18 Measurements and the Reaction of LiNbO, with Anhydrous HF Fluorine-18 (rl,, = 110min, p' emitter) was prepared by neu- tron irradiation of Li,CO, at the Scottish Universities Research and Reactor Centre, East Kilbride. The irradiated material was converted to H18F and then to solid Cd8F using inactive CsF as a carrier and the solid was transported to Glasgow.16 Aqueous H18F was prepared by addition of a small quantity (ca.5 mg) of solid Cs18F to 40% aqueous HF (20 cm3) contained in a PTFE beaker. The mixture was warmed to equilibrate the solution, cooled and diluted with 40% aqueous HF until an aliquot (0.5 cm3) counted in an FEP (perfluoroethylene propylene) tube gave a count of lo4 in 30s. Anhydrous HF was labelled with [''F] by exchange with solid Cs18F at 200°C in a Monel metal pressure vessel J. MATER.CHEM., 1994, VOL. 4 connected to a calibrated Monel vacuum line. The specific count rate (count minmmol-l) of each batch of H18F pre- pared was determined by condensing a measured quantity of gaseous Hi8F onto a sample of calcined ;'-alumina contained in an evacuable FEP tube equipped with a Monel or PTFE valve. The interaction of H18F vapour with LiNbO, wafers, cu. 5 x 20 mm, and with congruent LiNbO, powder (0.5 g) was studied by expanding a measured quantity of H18F into an evacuable FEP counting tube. In some cases the sample was pumped for 24 h before Hl'F admission. Count rates were determined at regular intervals using a Scaler ratemeter and an NaI well scintillation counter (NE and Ecko).The counting geometry used was such that ["F] activity determined arose mainly from the activity deposited on the solid sample. Calibration experiments using counting tubes containing H18F vapour only, confirmed this and thus ["F] activity originating from the gas phase could be discounted without serious error. The variation of count rate with time for a sample defined the ['*F] growth curve for the solid. When apparent equilib- rium was reached, volatile material was removed by pumping at room temperature for ca. 10min. The solid count rate corresponded to ["F] activity permanently retained by the wafer or powder. This count rate, combined with the specific count rate of H18F used, enabled the uptake of HF to be determined.For some samples the effect of addition of further quantities of H18F or the addition of inactive HF was deter- mined using a similar procedure. Congruent LiNbO, powder (4g) was added to an FEP tube and connected to a Monel vacuum line by Swagelok fittings. Anhydrous HF (3 cm3) was added by distillation and the mixture was allowed to react for 20 h. Volatile material was removed by distillation and the solid pumped to remove remaining traces of HF. The solid material was washed with distilled water through a plastic frit, the colourless washings evaporated, and the resulting solid dried at 120"C. Analysis of the solid gave: F, 31.8; Li, 3.55%; Li: F = 1:3.2. A second analysis carried out later gave: F, 25.0; Li, 3.6%; Li :F =1:2.5. XRD and vibrational spectroscopic data obtained are dis- cussed below. Results Polishing Experiments Several series of polishing experiments carried out under identical conditions, demonstrated the beneficial effect of adding KHF, to alkaline silica sol for the polishing of Y-cut LiNbO, wafers. The criteria used were stock removal measure- ments, surface-finish determinations as measured by R, data (an average value determined from two traces taken on five of the twelve wafers polished in any one experiment) and the lithium contents of the polishing reagents collected over 0.5 h time periods during polishing.Stock removal data are com- pared in Fig. 1, data averaged from two separate experiments for each reagent.After 120 min polishing ca. 8 pm of material had been removed using silica sol-KHF, whereas only cu. 3.5 pm was removed using silica sol alone over the same time period. Subnanometre surface roughness was achieved in under 1 h using silica sol-KHF, whereas ca. 2 h was required using silica sol alone, see Fig. 2 (data averaged from two experimental runs). Using either reagent, final R, values were in the range 0.5-1.0 nm. Lithium content, determined as the product of the Li concentration in the reagent collected over a 0.5 h period and the reagent volume, was always greater using the silica sol-KHF, reagent. However, data obtained from polishing with silica sol alone depended markedly on the degree of J. MATER. CHEM., 1994, VOL.4 I I I I I I I 0 20 40 60 80 100 120 140polishing time/min Fig. 1 Stock removal from LiNbO, during polishing with alkaline silica sol (a)and alkaline silica sol-KHF, (0) 60 I 50 .i 40 E5 30cr" 20 10 0 0 20 40 60 80 100 120 140 poIis h in g tim e/m in Fig. 2 Surface roughness measurements on LiNb0, made during polishing with alkaline silica sol (a)and alkaline silica sol-KHF, (0) polishing pad conditioning. Data in Fig. 3 are typical of those obtained, however Li contents from silica sol polishing experi- ments could be increased by up to a factor of two, particularly in the first hour of polishing, by careful attention to pad conditioning. The effect of anisotropy in LiNbO, on polishing was investigated by Li determinations in polishing fluids. Similar- sized pieces of X-, Y- and Z-cut LiNbO, were lapped and polished using identical conditions after 1.5 h and Li contents 0.7 0.6 0.5 0.4 h -I-a c0 0.3 0.2 10.1 0.0 1 I 2 5 6 7 8 time intervaV0.5 h Fig.3 Total Li determined from polishing fluids during successive 0.5 h periods during polishing of LiNbO, with alkaline silica sol (dark columns) or alkaline silica sol-KHF, (light columns) determined (Table 1). The orientation of the LiNbO, wafer had an effect on the quantity of soluble Li determined, particularly when silica sol alone was used. Addition of KHF, produced the greatest enhancement of the polishing reaction when Y-cut material was polished, but appeared to have no effect for X-cut. In view of the mechanical effects described above, any correlation of the small differences in the lithium removal with the different atomic arrangements of the three surfaces is not justified.Interaction between Anhydrous HF and Silica The concentration of [HF,] -in the fluoride-modified silica- sol reagent was 0.1 mol dmP3, this being the upper limit for avoidance of rapid coagulation of the reagent. In order to model the silica sol-[ HF,] -interaction, silica powder obtained by evaporation and vacuum drying of alkaline silica sol was exposed to fluorine-18 labelled anhydrous hydrogen fluoride vapour whose [18F] specific count rate had been determined at room temperature. Exposure resulted in an immediate [18F] count rate which increased slowly thereafter.The interaction between H1'F and the solid appeared to be essentially complete within a few minutes. On the removal of excess H"F by pumping, the [l'F] count rate of the solid corresponded to an HF uptake of 0.30f0.01 mmol g-' This was indicative of a significant interaction, although SEM examination showed no difference in particle size as a result of HF vapour treatment. Table 1 Total Li determined from polishing fluids after 90 min polishing of LiNb0, Li determined/mg after polishing with wafer orientation silica sol silica sol-KHF2 X-cut 0.26 0.25 Y-cut 0.05 0.3 1 Z-cut 0.18 0.42 J. MATER. CHEM., 1994, VOL. 4 Interaction between LiNbO, Wafers and Water A large Y-cut LiNbO, wafer was cut into twelve 2 cm x 1 cm wafers; these were polished with alkaline silica sol and their transmission IR spectra recorded.The spectra contained a complex band envelope in the v(0H) region, v,,, =3485 cm-' with shoulders at 3465 and 3502cm-'. These were very similar to previous observations which have been taken to indicate the presence of three types of hydroxy group in the material." Band integrations were uniform from sample to sample, indicating there was no gross difference in chemical action over the total face of the larger wafer as a result of the polishing process. Irradiation of an LiNbO, wafer-water mixture in a commer- cial microwave oven led to the incorporation of additional hydrogen. The IR spectrum of a polished wafer irradiated for 30 min in liquid r2H]-H2O contained a complex absorption between 2650 and 2560cm-' due to v(02H) vibrations in addition to the v(0'H) absorption.The latter was essentially unaffected by the irradiation process, indicating that the new bands arose from the incorporation of additional hydrogen rather than being the result of 2H/'H isotopic exchange. After 16 days under ambient laboratory atmosphere the v(02H) absorption could not be observed. In a separate experiment the v(02H)absorption was observable after 20 min irradiation and the band area increased significantly after a further 20 min irradiation period. Attempts to follow the decrease in the v(02H)band area quantitatively with time were not entirely successful; however, it had decreased significantly after 11 days.Irradiation of an LiNbO, wafer in the presence of H20 for 35 min resulted in the v(0H) band area increasing by a factor of two. After a total of 1.5 h irradiation the band intensity had increased by a factor of three. In all cases it appeared that additional hydroxy groups were formed during the irradiation process and that the 'extra' hydrogen incorpor- ated was lost from a wafer under ambient conditions. Prism- coupling experiments on an LiNbO, wafer irradiated for 1.5 h in H20 indicated that the sample behaved as a monomode waveguide, i.e. the extraordinary refractive index of the wave- guiding layer was greater than that of bulk LiNbO,. By analogy with previous work in which multimode planar waveguides have been fabricated by reactions of LiNbO, wafers with protonic acids,2" it is likely that lithium is lost from the wafer during this process.The observed behaviour was unexpected from previous studies in which hydrogen ('H or 2H) diffusion into LiNbO, was observed only at high temperatures.18 It is, however, related to the facile 'H/2H isotopic exchange processes that have been observed between water vapour and partially exchanged Li, -,H,NbO, materials, and implies that the reversible reaction (eqn. 1)19 is possible 02-(surface)+H20 (liquid)+20H- (surface) (1) Interaction between LiNbO, and Anhydrous HF This was examined by two methods, by exposing LiNbO, to H18F vapour at room temperature and by physico-chemical examination of the products from the reaction between LiNbO, and liquid anhydrous HF.Admission of HI8F vapour to LiNbO, powder at room temperature resulted in significant uptake of ["F] by the solid (Fig. 4). The bulk of the radio- activity (ca. 80%) appeared to be retained on pumping at room temperature and the uptake of HF, determined by using H"F of measured specific count rate, was in the range (0.19-0.35)kO.Ol mmol g-'. The [18F] growth curve indi- cated that the reaction was relatively slow. An apparent equilibrium was reached after 100 min; however, removing excess H18F at this stage then adding a further aliquot of 1000 3 900 F I .-E E 800 1 81 Q,c s! c1 8 700 LL 7 600 II I Ill 0 20 40 60 80 100 120 140 t im e/m in Fig.4 Growth in ['*F] count rate in LiNbO, on exposure to HI8F vapour at room temperature H18F, led to a second rapid uptake.The [18F] activity was reduced by exposure of the labelled solid to inactive HF vapour at room temperature, suggesting that some fluoride was labile and all ["F] appeared to be removed completely by washing with H,O. Aqueous H18F was placed on an LiNbO, wafer and left for 0.5 h; after rinsing with H20 all the activity was removed. The BET area of LiNbO, powder was 0.3 m2 g-'. Assuming the van der Waals radius of the HF molecule is 2.55 x lo-'' m then the uptake of HF required to form a monolayer on the surface of LiNbO, is 0.002mmol g-'. The observed uptakes were far higher, suggesting that the reaction involved bulk material.The IR transmission spectrum of the solid recorded after ["F] activity had decayed, was similar to that of untreated LiNbO,. In particular, there was no evidence for new bands in the region expected for Nb-F stretching vibrations. Diffuse reflectance IR Fourier transform spectroscopy (DRIFTS) of LiNbO, powder and LiNb0, powder treated with anhydrous HF produced good quality spectra. The main features in the LiNbO, spectrum (4000-650 cm-l) were envelopes in the hydroxy group stretching and deformation regions and a strong band at 960 cm-'. Treatment of LiNbO, powder with HF vapour followed by subsequent exposure to moist air, resulted in a significant increase in the intensity of the hydroxy group stretching mode envelope.Before HF treatment this occurred at 3700-2750 cm-' with vmax= 3468 cm-'; the corre- sponding data after treatment were 3725-3000 cm-' and v,,,=3528 cm-'. In addition, the profile of the 960 cm-' band was changed. There was no evidence in the spectra for the presence of (HF), oligomers nor for bands attributable to Nb-F stretching modes. The XRD powder pattern of LiNbO, that has been treated with liquid anhydrous HF overnight [Fig. 5(a)] showed the presence of LiNbOj2' and LiNbOF4.21 This product was partially soluble in water and the solid isolated from solution after evaporation and drying contained LiNbOF42' and Nb20520 as shown from its XRD powder pattern [Fig.5(b)]. The main features in its Raman spectra were strong bands at J. MATER. CHEM., 1994, VOL. 4 76 68 60 52 44 36 28 20 12 4 20ldegrees Fig. 5 XRD powder results from (a)the solid product from LiNb03 and anhydrous HF (liquid); (h)the water soluble material from this reaction; ( W) LiNbOF,, (A)LiNb03, (a)Nb,O, 810 and 607cm-’. Its fluorine content decreased with time presumably owing to hydrolysis. The behaviour of LiNb0, towards HF may be compared with the reaction between LiNbO, and liquid BrF,. At room temperature a mixture of LiNbOF, and LiNbF, is formed; conversion to LiNbF, is not complete even at 1260C.21Under more extreme conditions, for example using molten hydrogen- difluoride salts, LiNbO, can be fluorinated to give LiF and M,NbOF6 (M=NH, or K).,, Discussion Both processes studied in this work can be described as chemomechanical, in which chemical reactions are induced at the lithium niobate surface via mechanical energy generated from forces between the polishing pad and the wafer surface.The results obtained are inconsistent with a simple grinding process involving silica particles for two reasons. First, a water soluble lithium species is produced and, secondly, addition of the hydrogendifluoride anion has a beneficial effect. The latter would not be expected for a purely mechan- ical process because of the destabilizing effect of [HF,]- on the negatively charged silica particles (cf. ref. 15). The lithium- containing product is unlikely to be simply LiNbO, since the latter is insoluble in water and can be dissolved only with difficulty in aqueous HF.The reaction between LiNbO, and [HF,] -under polishing conditions does not lead to retention of fluorine-containing species on the surface. Removal of LiNbO, is enhanced however (Fig. 1 and 3) but without loss of surface finish (Fig. 2). Simple etching of LiNbO, in aqueous HF, in which the major species is the ‘tight’ ion pair H30+F-,23 destroys the surface finish. The most obvious way of moderating an etching reaction for polishing purposes is via the formation of an insoluble product (a passivating layer) which accumu- lates in the valleys on the surface (cf. ref. 12, 13). Continued reaction on the peaks eventually produces a plane surface from which the passivating layer is removed by mechanical wiping.The observations made in the polishing of LiNbO, by KHF,-silica sol and the results of the model reaction between LiNbO, and anhydrous HF, particularly the XRD data (Fig. 5), lead us to propose the following model. Hydrogendifluoride anion reacts with the LiNbO, surface to give water-soluble LiNbOF,, which undergoes hydrolvsis to give hydrated Nb205 and LiF. Hydrated Nb205 accumulates on the surface and acts as a passivating layer. Further reactions involving Nb205 must be considered, since it is known to form the [NbOF5I2- anion in dilute aqueous HF,,, and to react with MHF, (M=K or NH,) giving M,NbOF,.25 However [NbOF,I2- and [NbOF4( H20)] -exist in equilib- rium with each other in water,,, hydrolysis of [NbOF,I2- in alkaline media is reported to give hydrated Nb205” and [NH,], [NbOF,] gives [NH, J[NbOF,] on thermal decomposition.22 In reality therefore, LiNbOF, and hydrated Nb205 can be considered to be two of the key member.; (eqn.2 and 3) in a complex series of reactions occurring on the LiNbO, surface. LiNbO3+4[HF,]--+LiNbOF,+4F-+2H20 (2) 2LiNbOF, +6H,O+Nb,O, +2LiF +6HF (3) Attempts to identify the presence of NbV 0x0 or oxofluoro species by vibrational spectroscopy were unsuccessful, because of the dominant effect of LiNb03.27 There were, howeher, no bands that could be obviously attributed to [NbF,]-or [NbF7I2-. The lack of evidence for LiF in the XRD powder data from the model reaction (Fig.5) can be attributed to the dominant effect of the heavy NbV atoms. The activation of the surface observed when an LiNbO, wafer is exposed to microwave radiation in the presence of water (eqn. l), leads us to suggest that the chemical reaction between LiNbO, and alkaline silica sol under polishing con- ditions may involve the attack of OH- or -Si-0- maieties on the surface. The products expected would be LiOH and Nb,O,. The absence of additional reactions (eqn. 2 and 3) results in a slower polishing process in agreement with the observations made. The authors thank staff at the Scottish Universities Research and Reactor Centre, East Kilbride for assistance with the neutron irradiations. Financial assistance from the Department of Trade and Industry and Logitech Ltd, under the auspices of the LINK Nanotechnology programme, is gratefully acknowledged.Prism-coupling experiments were carried out by Dr. A. Loni of the Department of Electronic and Electrical Engineering, University of Glasgow. References 1 A. Rauber, Current Topics in Materials Science, ed. E. Kaldis, North-Holland, 1978, vol. 1, ch. 7, p. 481. 2 J. L. Jackel. C. E. Rice and J. J. Veselka, Jr., Appl. Phys. Lett., 1982, 41, 607; C. E. Rice, J. L. Jackel and W. L. Brown, J. Appl. Phys., 1985,57,4437. 3 A. Loni, R. M. De La Rue and J. M. Winfield, J. Appl. Phys , 1987, 61, 64; A. Loni, G. Hay, R. M. De La Rue and J. M. Winfield, J. Lightwave Technol., 1989,7,911; A. Loni, R. W. Keys, R. M.De La Rue, M. A. Foad and J. M. Winfield, IEE Proc. Part J, 1989, 136,297;A. Loni, R. M. De La Rue, J. McCaig and J. M. W infield, J. Appl. Phys., 1990,67, 3968. 4 J. T. Cargo, A. J. Filo, M. C. Hughes, V. C. Kannan, F. A. Stevie, J. A. Taylor and R. J. Holmes, J. Appl. Phys., 1990,67,627. 124 J. MATER. CHEM., 1994, VOL. 4 I. T. Savatinova, M. Kuneva, B. Jordanov and D. Kolev, J. Mol. 18 R. Gonzalez, Y. Chen, K. L. Tsang and G. P. Summers, Appl. Struct., 1990,219, 165. Phys. Lett., 1982, 41, 739; N. Schmidt, K. Betzler, M. Grabs, V. A. Ganshin and Yu. N. Korkishko, Opt. Commun., 1991, 86, S. Kapphan and F. Klose, J. Appl. Phys., 1989.65, 1253. 523. Lord Rayleigh, Proc. Optical Convention, 1st Convention, London, 19 20 W. Bollmann, Phys.Status Solidi A, 1987,104. 643. Joint Committee for Diffraction Standards, International Centre 1905, p. 73; see also, F. Twyman, Prism and Lens Making, 2nd edn, for Diffraction Data, LiNbO,, 20-631; Nb205.18-91 1. 8 9 10 11 12 13 14 15 16 17 Institute of Physics, 1988, ch. 3, p. 49; G. Fynn and W. J. A. Powell, Cutting and Polishing Optical and Electronic Materials, IOP, Bristol, 1979;W. J. Rupp, Optica Acta, 1971,18, 1. J. Noda and I. Ida, Rev. Electr. Commun. Lab., 1972,20, 152. B. Furch, E. Bratengeyer and H. Rauch, J. Opt. Commun., 1983, 4, 47. E. Neumann and H. Schulz, Cryst. Res. Technol., 1985,20, K115. S. D. Poulsen, FerroeIectrics, 1987,75, 79. N. J. Brown, Precision Engineering, 1987,9, 129. L. M. Cook, J. Non. Cryst. Solids, 1990, 120, 152.J. H. Clark, J. Emsley, D. J. Jones and R. E. Overill, J. Chem. Soc., Dalton Trans., 1981, 1219. E. M. Rabinovich and D. L. Wood, Muter. Rex Soc., Symp. Proc., 1986.73, 251; E. M. Rabinovich, D. M. Krol, N. A. Kopylov and P. K. Gallagher, J.Am. Ceram. Soc, 1989,72, 1229. K. W. Dixon and J. M. Winfield, J. Chem. Soc., Dalton Trans., 1989,937. J. R. Herrington, B. Dischler, A. Rauber and J. Schneider, Solid 21 22 23 24 25 26 27 E. G. Rakov, M. V. Melkumyants and V. F. Sukhoverkov, Russ. J. Inorg. Chem., 1990, 35, 632. A. I. Agulyanskii, Yu. I. Balabanov, \I. A. Bessonova, A. G. Babkin and P. T. Stangrit, Iza. Akud. Nauk SSSR, Neorg. Muter., 1985,21,98; Chem. Abstr., 1985, 102, 124587r. P. A. Giguere and S. Turrell, J. Am. Chem. Soc., 1980, 102, 5473; D. Mootz, U. Ohms and W. Poll, 2. Anorg. Allg. Chern., 1981, 479,75; J. Khorami, R. Beaudoin and M. Menard, Can. J. Chem.. 1987,65,817. J. A. S. Howell and K. C. Moss, J. Chem. SOL‘.,A, 1971,2481. A. I. Agulyanskii, S. S. Pochivalov and V. M Mel’nikova, Russ. J. Inorg. Chem., 1989,34, 1567. Yu. A. Buslaev, E. G. Ilk, V. D. Kopanev and V. P. Tarasov. J. Struct. Chem., USSR, 1972, 13, 865. J.-M. Jehng and I. E. Wachs, Chem. Muter, 1991,3, 100; J. H. von Barner, E. Christensen, N. J. Bjerrum and B. Gilbert, Inorg. Chem.. 1991,30, 561. State Commun., 1973, 12, 351; L. Kovacs, V. Szalay and R. Capelletti, Solid State Commun., 1984,52, 1029. Paper 3/0348 1J; Receiccii 17th June, 1993