Journal of Crystal Growth 54 (1981) 546—550 North-Holland Publishing Company
THE CZOCHRALSKI GROWTH OF SINGLE CRYSTAL LITHIUM ALUMINATE, LIAIO2 B. COCKAYNE and B. LENT RSRE, St Andrew’s Road, Ma/vem, Worcestershire, UK Received 27 January 1981; manuscript received in final form 12 March 1981
The existence of a congruently melting compound corresponding to a composition LiAlO2 within the Li20--A1203 system is confirmed and the melt growth of orientated single crystals is demonstrated. The compound grown from the melt has the -y-forni reported for flux-grown crystals. Optical and physical property measurenients for the melt-grown material are reported.
1. Introduction Lithium aluminate, LiAIO2, is reported to exist in three allotropic forms cw, ~ and ‘y which have hexagonal, monoclinic and tetragonal crystal structures respectively [11. The tetragonal y-phase has been assigned the space-group P41212 (or the enantiomorph P43212) [21 which is acentric and suggests that the material should have piezoelectric properties. The piezoelectric properties form the basis of a separate study  whilst this paper reports the procedures developed for the growth of single crystals of ‘y-LiAlO2 from the melt using the Czochralski technique and presents measurements made of general optical and physical properties. Small single crystals of y-LiAJO2 have been produced by earlier workers [4,5] using a high-pressure flux technique but most published work has been concerned with the production of powders and ceramics . Some confusion in the nomenclature of the vanous phases occurs in the literature. Most workers [1,2,4,5] refer to the tetragonal phase as y and the hexagonal phase as ci, whereas others  reverse this classification. In this paper, the tetragonal phase is referred to as ‘y throughout. 2. Experimental 2.1. Melt preparation
LiA1O2 was prepared from a 1 1 stoichiometric mixture of Li2CO3 (Johnson Matthey Chemicals Ltd, 0022-0248/81 /0000—0000/$02 .50 © 1981 North-Holland
Grade 1) and A1203 (BDH Ltd, Optran Grade) which was contained in a platinum crucible and heated in air for 16 h at 1100°C in order to decompose the carbonate and reduce the volume of material. This treatment yielded a loosely sintered mass with 98% of the anticipated weight loss due to the decomposition. X-ray powder diffraction measurements on this material gave an identical pattern to that produced subsequently in single crystal form. 2.2. Crystal growth
Using the sintered material prepared as above, single crystals were grown by the Czochralski technique in the apparatus employed for many oxides and described elsewhere [7,8]. The iridium crucible used to contain the melt was heated inductively by a 450 kHz radio-frequency generator and magnesia refractories were utilised as heat shields to surround both the crucible and the space above the melt into which the growing crystal was withdrawn. Initially, a randomly orientated crystal was obtained from polycrystalline nucleation on to the end of an iridium rod by restricting the diameter of the crystalline material so that only one crystal could grow. Subsequent crystals grown on to orientated seeds cut from this initial crystal, were used to produce c-, a- and  -axis crystals. The diameter of the crystal was controlled using a signal derived from crystal weight as the monitoring parameter, as described previously . A gas ambient of argon/2 oxygen flowingrotaat a 3 min~ provedvol% suitable. A crystal rate of 500 cm
B. Cockayne, B. Lent I Czochralski growth of LiA1O
Fig. I. Single crystals of y-LiMO2 grown under identical thermal geometries; 1, at(a) which c-axis rate crystal loss of grown the alkali at 0.3component cm h~ to give occurs; chemithe decanted cally stable endgrowth of thisconditions; crystal partially (b) [1101-axis cleaved and crystal has been grown removed at 0.1 (the cm hsmall squares have a side = 0.1 cm).
tion rate of 10 rpm was used and the growth rate was optimised at 0.3 cm h~.At a lower growth rate of 0.1 cm h’, considerable loss of the volatile alkali component occurs, probably as Li 20, during a typical growth cycle of 16 h leading to precipitation which imparts a milky appearance to the crystals, particularly in the core. At a higher growth rate (>0.5 cm h’~)the crystals develop a core of voids, characteristic of cellular structures in oxide crystals grown under conditions of constitutional supercooling [7,10]. A c-axis crystal of LiAIO2 grown at the optimised rate is shown in fig. la and compared with a axis crystal grown under conditions of alkali loss in fig. lb. The former crystal is transparent and has the smooth surface characteristic of chemically stable growth whilst the latter soon develops the furrowed appearance characteristic of growth under conditions where chemical decomposition of the melt induces surface etching at a rate dependent upon crystal composition, 2.3. Crystal polishing Sections of oxide crystals required for optical study can usually be prepared successfully after diamond sawing by grinding the surface flat on 600 grade carborundum, followed by polishing on a solder lap using successively finer grades of diamond paste to the 3 pm grade. When the y-LiA1O2 single crystals were processed in this way, severe pitting of the surface ensued. The normal polished surface was obtamed by replacing the solder lap with a wax lap and by the use of an alumina/ethane diol suspension in
place of the 3 pm diamond stage. Subsequent p01ishing using Syton produced highly-polished pit free surfaces. 2.4. Crystal quality The interferogram, fig. 2a, shows clearly that crystals can be grown which have a high degree of optical perfection and complete extinction between crossed polars can be observed. In contrast, crystals~grownat rates sufficiently slow for melt decomposition effects to become apparent, exhibit refractive index changes which can be observed as distortions in the fringe pattern of fig. 2b. Inspection between crossed polars (fig. 3a) shows that this distortion arises from a sub-structure which is identical to that observed in a number of anisotropic Czochralski-grown single crystal refractory materials  and known to be due to the presence of low-angle boundaries. Further substructures can develop within these crystals when the precipitate causing a milky appearance forms in the crystal core. In c-axis crystals, this precipitation produces strain within the material which has the four-fold symmetrical distribution characterised by fig. 3b. Further evidence for this strain is the presence of substructures in the centre of the crystal (fig. 4a) and the propagation of twins from the core to the edges of the crystal (fig. 4b). The precipitates are too small to be resolved individually by optical microscopy. Their chemical identity has not been established but they are most likely to be associated with deviations from stoichiometry arising from loss of Li20 at the melting point of LiAIO2 these deviations in melt
I ic. 2. 1 is vman —Creen interferogrant taken
1 cm of )—LiAlO
2 I rum: (a) —,tsis ~ stal crown at the optimised rate I 0) 10 j—a~iscr~stal grunn under conditions of alkali loss )crysLil diameter = 1.4 cub.
1 ig. 3. Strain patterns of 7—LiAlO9 crystal sections taken through crossed pulars showing: (a) the suhstrueture in the crystal section from fig. 20 (crystal diameter = 1.4 L~ I: (0) the four—told strain pattern associated svith the onset of precipitation in a IraOs— serse section of a c—axis crystal grown under conditions ol alkali loss (crystal diameter = 1.4 cm).
I ig. 4. Strain patterns of ~—LiAlO2er~stal seCtions taken through crossed polars. showine: (a) the substructure asso~iated is lilt the unset of precipitation as seen in the cr stal section of fig. 30 (bar marker I (.( (5 cm ) : lb ) the t pe ot tis itt siru~ture which dci el— ups at the oitset of precipitation as seen in the cr\ stal seLtion ot fig.30 (ha r marker = 0.1(5 cm I.
B. Cockayne, B. Lent
/ Czochralski growth of LiA1O2
stoichiometry must lead to compositional changes in the solid which lead to a metastable state and subsequent precipitation.
3. Crystal properties 3.1. Melting point
The melting point was determined using an optical pyrometer with an operating range of 750—3000°C and calibrated using other oxide melts such as A1203 and Y3A15012. In order to determine the melting temperature, the charge was heated until a small stable solid island, some 0.2—0.3 cm in diameter, was just visible and the temperature noted. The power into the crucible was then raised until the island just disappeared and the temperature was recorded again. In general, the difference between these temperatures was less than 20°C and the mean was arbitrarily chosen as the melting point. On the basis of five separatw determinations, a melting point of 1700 ±20°C was measured. 3.2. Thermal expansion coefficient
Dilatometric measurements were made upon a single crystal sample measuring 7.55 cm (c) X 5.45 cm (a) X 5.45 cm (a) with mutually perpendicular faces cut parallel to (100) axes. The expansion characteristics were determined at a linear heating rate of 5°C min’ under a vacuum of 106 Torr using an automatic dilatometer described in an earlier paper . The temperature range investigated was RT to 700°C.Theexpansion coefficients, derived 60C1 by andlinear cia = regression analysis were ct~= 15 X 10 7.1 X 1060C1. 3.3. Density
tent with the space-group of P41212 (P43212): this technique does not distinguish between the two enantimorphs. Approximate lattice spacings can be derived from the precession photographs and gave a = 5.17 A, c = 6.26 A which agress well with the more accurately determined values of a = 5.1687 A and c = 6.2679 A obtained on flux grown crystals of y-LiAlO2 . 3.5. Optical transmission
At room-temperature the compound is transparent to radiation throughout the wavelength range 0.2 to 4pm followed by a steady drop in transmission to zero at 5.5 pm. The crystals are colourless which is consistent with the absence of absorption bands in the visible spectrum. The absence of hydroxyl absorption bands in the infra-red range for a compound containing a strong hydroxide forming component such as Li20, is also worthy of note. 3.6. Refractive index
The refractive index of a 30°prism cut from one of the crystals with the c-axis parallel to the roof of the prism was measured at a number of wavelengths using the minimum deviation method, the ordinary and extraordinary rays of this birefringent material being separated by means of a rotatable~polarizer. The results are presented in table 1. Conoscopic examination of a c-axis slice using a Bertrand lens and a quartz wedge showe that the crystals were negatively birefringent, i.e. n0 > ne.
Table 1 Refractive indices of single crystal -y-LiAIO 2
The density of four different crystals was measured by standard pyknometry and determined to be 2.62 ±0.01 which is in substantial agreement with the value of 2.615 published for flux-grown material. 3.4. Crystal structure
An X-ray structural investigation using the Buerger precession technique gave systematic absences consis-
___________________________________________ 4593 4678 4799 5085 5890
1.6141 1.6128 1.6118 1.6101 1.6039 1.6014 1.6006
1.6345 1.6324 1.6317 1.6289 1.6225 1.6197 1.6182
—0.0204 —0.0196 —0.0199 —0.0188 —0.0186 —0.0183 —0.0176
B. Cockayne, B. Lent/Czochralski growth of LiAIO
4. Conclusions It has been shown that despite the chemical reactivity of Li20 and the high melting point of A1203, a melt corresponding to the composition LiAIO2 can be made sufficiently stable to allow single crystals of y-LiAIO2 to be grown by the Czochralski technique. Between the competing influences of void defects, induced by fast growth rates, and precipitates, induced by slow growth rates, conditions can be delineated which permit the growth of transparent crystals with good optical properties. These crystals have the same crystal structure as those produced previously at a lower temperature by flux techniques . Acknowledgments The authors wish to thank I.M. Young (RSRE) for verifying the crystal structure and Dr. I.R. Harris (University of Birmingham) for his advice and assis-
tance with the dilatometry. Copyright Controller © 1-TIMSO London 1981.
References  K. Kinoshita, J.W. Sim and J.P. Ackerman, Mater. Res. Bull. 13 (1978) 445.  Marezio, Aeta B. Cryst. 19 (1965)  M. R.W. Whatmoxe, Cockayne et al.,396. unpublished work.  J.P. Remeika and A.A. Ballman, App!. Phys. Letters 5 (1964) 180.  M. Marezio and J.P. Remeika, J. Chem. Phys. 44 (1966) 3348.  C.H. Chang (1968) 2020.and G.L. Margrave, J. Am. Chem. Soc. 90  B. Cockayne, D.S. Robertson and W. Bardsley, Brit. J. Appl. Phys. 15 (1964) 1165.  B. Cockayne, M. Chesswas and D.B. Gasson, J. Mater. Sci. 2 (1967) 7.  W. Bardsley, B. Cockayne, G.W. Green, D.T.J. Hurle, G.C. Joyce, J.M. Roslington, P.J. Tufton and H.C. Webber, J. Crystal Growth 24/25 (1974) 369.  B. Cockayne, J. Crystal Growth 42(1977)413.  B. Cockayne, J.G. Plant and R.A. Clay, J. Crystal Growth, submitted.  J.S. Abel, K.G. Barracloygh, I.R. Harris, A.W. Vere and B. Cockayne, J. Mater. Sci. 6 (1971) 1084.