Synthesis and structure of Ba,InO,X (X = F, C1, Br) and Ba,ScO,F; oxide/ halide ordering in K,NiF,-type structures Richard L. Needs,' Mark T. Weller,*' Ulrich Schelerb and Robin K. Harrisb 'Department of Chemistry, University of Southampton, Southampton, UK SO1 7 1BJ bDepartment of Chemistry, University of Durham, Durham, UK DH1 3LE The structures of three indium and one scandium complex layered oxide halides have been determined from time-of-flight powder neutron diffraction data. The indium phases Ba,InO,F, Ba,InO,Cl and Ba,InO,Br all crystallize in the space group P4/nrnm and exhibit complete oxide halide segregation, producing alternating sections of halide [ BaF] and oxide [ BaO] ions separating infinite InO, layers. However, Ba,ScO,F shows only partial oxide/fluoride ordering with the fluoride ions occupying the two apical sites on scandium equally with oxide ions, in the space group I4/mrnm.19FMAS NMR data have been collected from Ba,InO,F and Ba,Sc03F, and confirm the structural analysis in terms of a single type of fluoride ion site in each material. The structural chemistry of complex oxide halides has become the subject of renewed interest following the recent reports of high-temperature superconductivity in various layered oxide halide cuprates. There appear to be at least three routes to these p-type oxide halide superconductors: interstitial halogen doping, cation substitution and anion substitution. Interstitial fluorine doping of Sr,CuO, affords the 46 K superconductor Sr,CuO,F,+,,' in which the fluoride ions occupy solely the apical sites about the copper, giving effective CuO, square planes and an interstitial site in the new SrF rocksalt layer.Cation substitution of calcium by sodium in both C~,CUO,C~,~ and C~,CU,O,C~,~,~ at high pressure results in the formation of ( Ca,Na),CuOzClZ5 and (C~,N~),C~CU,O,C~,~ with T,values of 26 and 49 K, respectively. Anion substitution of chloride by oxide in (Sr,Ca)3Cu204C12 at high pressure gives the 80 K further 48 h. The resultant ceramics were grey-green in colour and were checked for phase purity by powder X-ray diffraction. The indates were of low crystallinity but were pyre, and wer? indexed on a primitive tettagonal cell of a = 4.22 A, c = 15.03 A and a = 4.24 A, c= 15.48 A for the oxide chloride and oxide bromide respectively.The scandate Ba,ScO,F was indexe4 on a body-centred tetragonal cell of a=4.15 A and c= 13.54 A. The end member of this series is the oxide iodide Ba,InO,I, and the synthesis of this phase was attempted; owing to the hygroscopic nature of BaI,, as far as possible all procedures were carried out under argon, but after transfers to and from the glove box and powder X-ray analysis under Mylar film some hydrolysis occurred, resulting in the evolution of iodine vapour upon reheating. Reactions at all temperatures in the range 700-1050 "C produced complex barium indium oxides. superconductor Sr2.3Cao.7Cuz04+ ,C1, .7 ,, -These examples The reaction was hampered further by the volatile nature demonstrate clearly how the design of new superconductors with high transition temperatures can be achieved uia the use of oxide halides to achieve both layering and segregation.Apart from copper chemistry, there are relatively few struc- turally characterised layered oxide halides. K2Nb0,Fs and Sr,FeO,Fg have disordered apical fluoride oxide ions. Sr,Fe,05Cl,,10 Sr3Fe205Br2,11 their barium indium analogues Ba,In,05C1212 and Ba31n205Br2,13 and S~,CO,O,~~C~~'~ crys-tallize as Ruddlesden-Popper phases and as such are structur- ally similar to the La, -xSr,Cu206+ al' superconducting phases. As indium and scandium in complex oxides often form struc- tural analogues of cuprates, a study of their oxide halides was initiated and B~,IIIO,F~~ and Ba,In,05F,17 were reported.Both contain fully ordered segregated oxide and fluoride ions, with Ba,InO,F exhibiting a new K,NiF, superstructure, similar to that of the T* phases of cuprate superconductors.lS In this paper we report the synthesis and structure, as determined from analysis of time-of-flight powder neutron diffraction data, of the indates Ba,InO,Cl and Ba,InO,Br and the scandate Ba,ScO,F, and confirmation of the structure of Ba,InO,F as reported previously. Experimental The synthesis of Ba,InO,F has been reported previously.16 The other barium indates and scandates were all synthesised by high-temperature sintering of well ground mixtures of appropriate molar ratios of BaCO, (99.95%), BaF, (99%), BaCl, (99.9%), BaBr, .2H20 (99% pre-dried at 500 "C), In,O, (99.9%) and Sc203 (99.5%).The powders were fired at 850 "C for 24 h, reground and fired in the range 1025-1075 "C for a of BaI,. The NMR experiments were performed on a Chemagnetics CMX-200 spectrometer operating at 188.288 MHz for fluorine nuclei. A Chemagnetics double-resonance high-speed magic- angle spinning (MAS) probe was used, with spin rates of 12.5 and 16.0kHz. The MAS technique was applied in order to average to zero the chemical shift anisotropy of the fluorine nuclei. Homonuclear (F,F) dipolar coupling also can be aver- aged by MAS, provided that high spin-rates are used. In order to reduce the influence of the deadtime of the probe, which would show up in distortions of the baseline of the spectra, a spin-echo pulse sequence with detection at the echo maximum was applied.To avoid phase distortions generated by the overlap of rotor and spin echoes, the latter were synchronised with the MAS period. The echo times were 160 and 125 ps at the 12.5 and 16.0 kHz, respectively. The other experimental parameters were: fluorine 90" pulse duration, 3.5 ps; spectral width, 100 kHz. The chemical shifts were referenced by replace- ment to the signal of CFC1,. Structure Determination Powder X-ray data were collected using a SiemensoD5000 diffractometer employing Cu-Ka, radiation (1.5406 A) over the 20 range 20-120". Powder neutron diffraction data were collected on the POLARIS medium-resolution diffractometer at the Rutherford-Appleton Laboratory over the time-of-flight range 3-19 ms.Data from the high-resolution back-scattering bank were used for the refinement. The data were first normal- ised to the vanadium standard, a background pattern was deducted and the data were then analysed using the GSAS1' J. Mater. Chern., 1996, 6(7), 1219-1224 1219 Table 1 Final refined atomic coordinates for Ba21n03F at 298 K, estimated standard deviations (e s d s) are gven in parentheses anisotropic temperature factors/A2 Jw1) 2c 0 25 0 25 0 38075( 10) 0 97(4) 0 97(4) 093(5) Ba(2) 2c 0 25 0 25 0 10282(9) 0 95(4) 0 95(4) 0 85(5) In 2c 0 75 0 75 0 23245( 10) 0 M(2) 0 44( 2) 0 70( 6) O(1) 4f 0 75 0 25 0 25701(8) 1 07( 3) 048(2) 190( 4) O(2) 2c 0 75 0 75 0 08239(10) 193(4) 193(4) 0 58(4)F 2c 0 75 0 75 0 42745( 15) 3 02(6) 3 02(6) 1 97( 7) x2=209, R,= 147%, Rp=2 74% Space group P4/nmm a=4 1640(2) A,c= 13 9439(8) A Table 2 Final refined atomic coordinates for Ba21n0,C1 at 298 K, e s d s are given in parentheses anisotropic temperature factors/A2 -atom site X Y Z B11 B22 B33 Wl) 2c 0 25 0 25 0 35253(8) 0 72( 2) 0 72( 2) 0 73(5) Ba(2) 2c 0 25 0 25 0 09469(8) 0 93(3) 093(3) 0 96(5)In 2c 0 75 0 75 0 21405( 10) 049(2) 0 49( 2) 0 61(6) O(1) 4f 0 75 0 25 0 23777( 6) 108(3) 0 59(2) 132(4) O(2) 2c 0 75 0 75 0 07545( 10) 2 06(3) 2 06(3) 0 48( 5) Cl 2c 0 75 0 75 0 42591(5) 1 ll(1) 1 ll(1) 156(4) x2=3060, RWp=141%, Rp=287% Space group P4/nrnm a=42208(2)A, c=150308(2)A Table 3 Final refined atomic coordinates for Ba,InO,Br at 298 K, e s d s are gven in parentheses anisotropic temperature factors/A2 atom site X Y Z Bll B22 B33 Ba(1) 2c 0 25 0 25 0 34054( 12) 0 71(5) 0 71(5) 121(9) W2) 2c 0 25 0 25 0 09274( 12) 1 30(6) 1 30( 6) 0 67(8) In 2c 0 75 0 75 020681(14) 0 72(5) 0 72(5) 0 19(I) (31) 4f 0 75 0 25 0 22980( 7) 1 37( 5) 0 40(5) 161(7) O(2) 2c 0 75 0 75 0 07263( 12) 203(5) 203(5) 0 82(8) Br 2c 0 75 0 75 0 42450( 10) 124(4) 124(4) 2 13(9) x2=331, RWp=127%, R,=235% Space group P4/nmm a=42365(3)& c=154805(1)A Table 4 Final refined atomic coordinates for Ba2Sc03F at 298 K, e s d s are gven in parentheses anisotropic temperature factors/A2 atom site X Y z Bl1 B22 B33 ~ ~~ Ba sc 4e 2a 0 0 0 0 0 36140(9) 0 0 88(2) 0 52( 1) 0 88(2) 0 52(1) 116(5) 4 33(6) O(1) 0(2)/F 4c 4e 0 0 05 0 0 0 16383(8) 0 88(3) 197(3) 0 25(2) 197( 3) 2 10( 6) 1 22( 5) x2=227, R,=l 97%, Rp=3 55% Space group I4/mmm a=4 1480(2)& c=13 5441(8)w package with neutron scattenng lengths and absorption cross- peak shape, zero point and lattice parameters Further param- sections taken from Koster and Yelon '' eters included were both anion and cation positions, isotropic The quality of the powder X-ray data collected for Rietveld thermal parameters followed by an absorption correction, to analysis of both Ba,InO,Cl and Ba,InO,Br was poor, owing allow for the high neutron-absorption cross-section of indium, mainly to their hygroscopic nature The collection of time- and anisotropic thermal parameters Close inspection of the of-flight powder neutron diffraction data on the high-flux profile fits revealed a small amount of impunty in all three medium-resolution diffractometer POLARIS at RAL in indates This impurity level increased from oxide fluoride to vanadium cans produced good quality data Rietveld quality bromide and, by inspection of d spacings, was found to be the powder X-ray diffraction data were collected for Ba,ScO,F Rudlesden-Popper-type series Ba,In,O,X, The impurity level and no evidence of any primitive peak was observed, indicative in Ba,InO,Br was high enough, cu 5% Ba3In2O5Br,, to be of oxide halide disorder The initial model for refinement of included as a second phase using the crystal data reported this phase was the coordinate set of Sr2Fe0,21 with all anion previously l3 The neutron diffraction data from Ba,ScO,F positions refined as oxide Refined parameters included the were refined using the X-ray coordinates as an initial model background, lattice parameters, peak shapes, atomic coordi- and proceeding as for the indates, with no evidence for a nates and thermal parameters lowering of the symmetry from 14/mmm to P4/nmm observed The initial model for powder neutron data refinement of all The anion distribution was determined by bond-valence calcu- three indates was that of Ba,In03F, as derived from refinement lations using the refined neutron data anion positions There of powder X-ray data and published earlier The refinements is no quoted value for the bond-valence parameter,, r, for an proceeded smoothly mth the initial inclusion of background, Sc-F system so it was calculated, using ten scandium fluorides 1220 J Muter Chem, 1996, 6(7),1219-1224 from ICSD, as 1.756.The model selected was that of K2Nb03F and Sr,FeO,F with the indium apical anion positions occupied equally by oxide and fluoride. This structure gave the best results for the valence of both metal ions, compared with both fully and equatorially disordered models. The results of all powder neutron diffraction data refinements are in Tables 1-4 with important bond lengths and angles in Tables 5 and 6 and the bond-valence calculations of Ba,ScO,F in Table 7. Final profile fits for Ba,In03F, Ba,ScO,F and the I I I I I II-'-I 'I:8l6-I f4l2 two-phase refinement of Ba,InO,Br are shown in Fig.1-3, respectively. Discussion The phase Ba,InO,F has been reported previously,16 and its anion environment has been derived from bond-valence con- siderations owing to the similar X-ray scattering powers of oxide and fluoride. This problem is not alleviated by using neutron diffraction as the scattering amplitudes of oxide and fluoride are very similar, but the actual positions of the anions, surrounded by the powerful X-ray scatterers barium and indium, are determined more easily by neutron diffraction data analysis. As shown in Table 1 and Fig. 1, there is no evidence for any anion disorder as the profile fit is very good, the positions are stable with low e.s.d.s and the thermal parameters mainly are satisfactory, confirming the structure determined from X-ray studies.16 However, owing to the slightly high thermal parameters of the fluoride ion site partial oxide/ fluoride disorder was considered. 19FMAS NMR studies show only one resonance at 6, -60.5 f.1 (Fig.4) consistent with Table 5 Important derived bond lengths and bond angles for Ba,In03F, Ba,In03C1 and Ba,InO,Br distance/& angle/degrees bond Ba,InO,F Ba,InO,CI Ba,InO,Br Ba( 1)-O( 1) x 4 Ba( 1)-X x 4 Ba( 1)-X x 1 Ba(2)-0( 1) x 4 Ba(2)-O(2) x 4 Ba(2)-O( 2) x 1 In-O( 1) x 4 In-O(2) x 1 In-X x 1 O(1)-In-O( 1) 2.704( 1) 3.0156( 6) 2.674( 2) 2.9928( 2) 2.9582( 2) 2.582( 2) 2.1100(3) 2.092 1 (2) 2.719( 2) 161.32 2.726( 1) 3.18 13 (5) 3.329(2) 3.01 22 (9) 2.9986( 2) 2.557( 2) 2.1406( 4) 2.0817(9) 3.186(2) 160.70 2.725( 1) 3.2657( 9) 3.638( 3) 2.998( 1) 3.01 17( 3) 2.559( 3) 2.1479( 4) 2.077( 2) 3.380( 3) 160.9(2) Table 6 Important derived bond lengths for Ba,ScO,F bond dis t ance/A Ba-O(2)/F x 4 2.9 529 (3) Ba-O( 2)/F x 1 2.676(1) Ba-O( 1) x 4 2.7974( 8) sc-O( 1) x 4 2.0740( 1) SC-O( 2)/F x 2 2.219( 1) I I 1 c.s23456789fO 0 I1 I 1 I I I I I I I1 \ 2 10 .-c@I ii ILl I_J ! --L f0 11 12 13 14 15 16 f7 18 19 time of flight/ms Fig. 1 Final profile fit for the refinement of powder neutron diffraction data of Ba,InO,F. Upper solid line, calculated pattern; crosses, observed pattern; tick marks, allowed reflections; lower continuous line, difference plot.the proposed anion ordered model. Any disorder over these anion positions would produce two quite different fluorine sites which should be detectable as two resonances owing to the large spectral range of "F; however, the linewidth is substantial (ca. 10 ppm). The compounds Ba21n03C1 and BazIn03Br have been syn- thesised and structurally characterised by the analysis of powder neutron diffraction data. Both are isostructural with Ba,InO,F and as such contain infinite layers of In02 sheets separated by alternating layers of halide and oxide rocksalt layers. The structure and indium coordination polyhedra of all three indates are shown in Fig. 5 and 6. In the Ba,InO,X series the lattice parameters increase linear5 with halide ion radius; the increa2e in a is small (0.07 A), while that in c is much larger (1.54 A).These a5e very similar to the increases in Ba,In,O,X, C0.07 and 3.03 A (two In-0,X layers) for a and c, respectively], from X=F to X= Br in each case.12,13*17 The coordination geometry of the InO, square pyramid remains nearly constant in all three phases, with the large expansion in c arising from the increasing size of the Ba-X layer (Fig. 4 and 5). Again, these observations are similar to those seen in the series Ba31n205X2 (X =F, C1, Br) and Sr3Fe20SX2 (X =C1, Br). A structural consequence of these trends is that both the In-X and Ba-X bonds increase as a Table 7 Final bond-valence calculations for Ba,ScO,F apical anion disorder equatorial anion disorder full anion disorder cation valence X(l)=O X(2)=O/F X(l)=O/F X(2)=0 X( 1)=O/F X( 2) =O/F 2.177 0.649 2.826 sc 1.920 0.736 2.656 2.044 0.69 1 2.735 1.001 0.882 1.881 Ba 0.877 1.010 1.887 0.946 0.942 1.888 X( 1) and X(2) refer to the anion sites O(1) (4c)and 0(2)/F (4e) as given in the table of atomic coordinates (Table 4).J. Mater. Chem., 1996, 6(7), 1219-1224 1221 I I I I I I I I I 345678970 I I I I I I I I I] 5 I-h01 7i L-l--I I I 12.LU-10 11 12 13 14 15 16 17 18 19 time of flightlms Fig. 2 Final profile fit for the refinement of powder neutron diffraction data of Ba,ScO,F Upper solid line, calculated pattern, crosses, observed pattern, tick marks, allowed reflections, lower continuous line, difference plot function of halide size, along with the level of structural distortion from the body-centred coordinates of the ideal K,NiF, structure (Tables 1-3) As shown in Fig 5, the In-Oaplcal bond length decreases slightly from Ba,InO,F to Ba21n03C1 and very slightly to Ba,In03Br There are three possible reasons for this First, there may be a small residual In-X interaction, decreasing from fluoride to bromide, thus requiring a greater bonding interaction between the indium and apical oxygen Secondly, the ordering of the two apical sites is not complete, leading to the observation of two average sites, one mainly fluoride and the other mainly oxide, although the I9F MAS NMR data described above suggests otherwise Finally, the In- Oequatorlal bond increases from oxide fluoride to oxide bromide owing to the small lattice expansion in a, thus requiring a greater In-Oaplal interaction to satisfy the indium valence A similar trend is observed in the B?31n20,X2 system, where In-Oaplcal bonds decrease from 2 07 A in the oxide fluoride to 2 05 A for both oxide chloride and bromide In all three in4ates there is one rather short Ba(2)-0(2) bond pf ca 2 56 A, this is not unprecedented, with a value of 2 53 A seen in Ba31n20, 23 A companson of the indium coordination environments between Ba,InO,X and Ba31n,05X, shows the In-X interactions to be longer in the Ba,InO,X series The scandate Ba,ScO,F is not isostructural with Ba,InO,F, it crystallises in the normal K2NiF4 space group with dis- ordered apical anions forming a distorted octahedron, as in K,NbO,F The 19FMAS NMR spectrum of Ba,ScO,F shows only one resonance at 6, -790+1 (Fig 7), consistent with the proposed structure, although as for Ba21n03F the linewidth is substantial In both spectra there are some very minor peaks, in addition to the spinning sidebands suggestive of small amounts of impurity Anion ordering is to be expected in the indium oxide chloride and bromide where the two halide anions have very different sizes compared with oxide, and this ordering is clear from refinement of the powder neutron diffraction data In the case of the oxide fluondes either 1222 J Muter Chem, 1996, 6(7), 1219-1224 IuL L U A L - 1U 11 12 13 14 15 16 17 18 19 time of flight/ms Fig. 3 Final profile fit for the two-phase refinement of powder neutron diffraction data of Ba21n03Br Upper solid line, calculated pattern, crosses, observed pattern, upper tick marks, Ba,In,O,Br, allowed reflections, lower tick marks, Ba,InO,Br allowed reflections, lower continuous line, difference plot -605 I d,d100 0 100 6 200 -300 Fig.4 "F MAS NMR spectrum of Ba21n0,F collected at a spin rate of 16 kHz Some very weak spinning sidebands are visible 129 transients were accumulated using a recycle delay of 4s complete ordering (Ba,In03F) or apical anion disorder (Ba,ScO,F) is possible The difference between the K2NiF4 structure and Ba,InO,F superstructure is shown in Fig 8 The change in structure between Ba,InO,F and Ba,Sc03F is probably due to the change in size and coordination requirements between In3+ and Sc3+ The ionic racii of In3+ and Sc3+ in six-fold coordination are 0 80 and 0 745 A, respect-ively In general, the larger the metal ion the greater the coordination number, therefore the expected coordination number of indium is either similar to or greater than that of scandium However, as has been shown above (Fig 8), the coordination of indium is effectively square pyramidal and that of scandium is distorted octahedral in these oxide fluorides A major difference between the two ions is their electronegativ- ity Scandium is significantly less electronegative than indium, the Pauling electronegativity values for indium and scandium Fig.5 Structures of the layered segregated indium oxide halides Ba,InO,X (X=F, C1 or Br) showing the InO, sheets separated by Ba-0 and Ba-X rocksalt slabs ""(I! -& Fig. 6 The In0,X coordination environment of (a) Ba,InO,F, (b)Ba,InO,Cl and (c) Ba,InO,Br r ii Fig. 7 19FMAS NMR spectrum of Ba,ScO,F collected at a spin rate of 12 kHz. The echo technique was not used in this case, which accounts for the baseline distortion. Some weak spinning side bands are visible. 256 transients were accumulated, using a recycle delay of 4s. being 1.78 and 1.36, respectively; this has a noticeable effect on their bonding requirements. In crystal structures, scandium is often found in six-fold c~ordination~~with some examples of seven-, eight- and nine- coordination kno~n.,~?,~ Examples of lower coordination are limited to a few cases with very bulky ligands, whereas indium is often found in less than six-fold coordination.In general, electropositive elements form ionic compounds with many, longer, non-directional bonds, i.e. Na and Mg form ionic bonds with relatively large coordination numbers, rarely less than six. Similarly, less electropositive species form fewer, more covalent interactions, which are shorter and directional in nature; tetrahedral environments are common and octahedral geometry is the largest coordination environment found com- monly. Hence, in crystal chemistry scandium, which is the more electropositive, will form preferably coordination spheres K2NiF4 Ba21 no3F Fig.8 Comparison of the body-centred structure of K,NiF4 and the primitive superstructure of Ba,InO,F. K,NiF, structure: large light spheres K; small black spheres, Ni; medium grey spheres, F. Ba,InO,F structure: large grey spheres, Ba; small black spheres, In; medium grey spheres, 0;medium light spheres, F. which are more ionic in nature than those formed by indium. A comparison of the structures of the triiodides, InI, and ScI,, illustrates this difference in bonding character. 11-11, is a low- melting-point dimer consisting of pairs of distorted InI, tetra- hedra.27 ScI,, a high-melting-point ionic solid, crystallises in the Bi1328 structure, alternate-edge-sharing octahedra forming layers linked by van der Waals bonds.In the Ba2M03F compounds (M=Sc, In) this difference is manifested by the adoption of a distorted octahedral environment for scandium and a square-pyramidal geometry for indium. This large differ- ence in electronegativity can be attributed to the different electronic configurations as the indium d-electron subshells shield the outer electrons poorly. In conclusion, we have synthesised and structurally charac- terised a series of complex barium indium oxide halides, whose structures are similar to the T* cuprate superconductors, which contain infinite sheets of InO, layers separated by alternating rocksalt slabs of Ba-0 and Ba-X. The structure of the equivalent scandium oxide fluoride has been determined in the normal K,NiF4 space group with disordered apical anion sites containing an equal mix of oxide and fluoride ions.This relatively unusual difference between the crystal chemistry of a complex layered indate and the corresponding scandate has been explained by consideration of their respective electronega- tivities and the consequent degree of covalency exhibited in their bonding. This wide area of relatively unexplored chemis- try has yielded already interesting structural modifications of known structure types. Therefore, the investigation of indi- um and scandium oxide halides has been widened to include both other metals, such as iron, copper, titanium and tin, and related structure types, e.g. perovskite and Sr,Ti,Ol,-type Ruddlesden-Popper phases.We wish to thank the EPSRC for a grant in support of this work and the Department of Chemistry at Southampton for partial studentship funding for R. L. N. References 1 M. Al-Mamouri, P. P. Edwards, C. Greaves and M. Slaski, Nature (London), 1994,369,382. 2 Von B. Grande and H. K. Muller-Buschbaum, 2. Anorg. Allg. Chem., 1977,429,88. J. Mater. Chem., 1996, 6(7), 1219-1224 1223 8 9 10 11 12 13 14 15 16 T Sowa, M Hiratani and K Miyauchi, J Solid State Chem , 1990, 84,178 J Huang, R D Hoffmann and A W Sleight, Mater Res Bull, 1990,25,1085 Z Hiroi, N Kobayaashl and M Takano, Nature (London), 1994, 371,139 Y Zenitani, K Inan, S Sahoda, M Uehara, J Akimitsu, N Kubota and M Ayabe, Physica C, 1995,248,167 C Q Jin, X J Wu, P Laffez, T Tatsuki, T Tamura, S Adachi, H Yamauchi, N Kosbzuka and S Tanaka, Nature (London), 1995,375,301 F Galasso and W Darby, J Phys Chem , 1962,66,1318 F Galasso and W Darby, J Phys Chem, 1962,67,1451 W Leib and H K Muller-Buschbaum, 2 Anorg Allg Chem, 1984,518,115 W Gutau and H K Muller-Buschbaum, 2 Anorg Allg Chem, 1990,584,125 M Abed and H K Muller-Buschbaum, J AIloys Compd, 1992, 183,24 M Abed and H K Muller-Buschbaum, 2 Anorg Allg Chem, 1991,592,73 J Boje and H K Muller-Buschbaum, Z Anorg Allg Chem ,1991, 592,73 N Nguyen, L Er-Rakho, C Michel, J Choisnet and B Raveau, Mater Res Bull, 1980,15,891 R L Needs and M T Weller, J Chem SOC Chem Commun, 1994,353 17 R L Needs and M T Weller, J Chem SOC Dalton Trans, 1995, 3015 18 J Akimitsu, S Suzuki, M Wantanabe and H Sawa, Jpn J Appl Phys, 1988,27, L1859 19 A C Larson and R B Von Dreele, GSAS General Structure Analysis System, MS-H805, Los Alamos, NM, 1990 20 L Koster and W B Yelon, Neutron Diffraction Newsletter, 1982 21 S E Dann, D B Curne and M T Weller, J Solid State Chem, 1991,92,237 22 I D Brown and D Altermatt, Acta Crystallogr Sect B, 1985, 41,244 23 K Mader and H K Muller-Buschbaum, Z Anorg Allg Chem, 1988,559,89 24 G W Melson and R W Stotz, Coord Chem Rev, 1971,7,133 25 I Kubach, Gmelins Handbuch der anorganischen Chemie Scandium 39, Verlag Chemie GMBH Weinheim/Bergstrasse, 1973 26 G Wilkinson, R D Gillard and J A McCleverty, Comprehensiue Coordination Chemistry, Pergamon Press, Oxford, 1987, vol 3, p 1060 27 J D Forrester, A Zalkin and D H Templeton, Inorg Chem ,1964, 3,63 28 D Brown, Halides of the Lanthanides and Actinides, Wiley-Interscience, New York, 1968, p 218 Paper 6/00211K, Recewed 10th January, 1996 1224 J Muter Chem, 1996, 6(7), 1219-1224