J. MATER. CHEM., 1991, l(1) 17-21 Synthesis and Structure of Ba,CaCu,.,,O,.,,(Co~)~.~, a Perovskite containing Carbonate Anions, and Related Phases Colin Greaves* and Peter R. Slater Superconductivity Research Group, School of Chemistrx University of Birmingham, Birmingham B15 2TT, UK Calcination of appropriate mixtures of BaCO,, CaCO, and CuO has been shown to result in a tetragonal [P4lmmm; a= 5.799(2)A, c= 7.992(3)A] perovskite-related phase containing C0:- anions. The unit-cell compo- sition is Ba,CaCu,,,O, 96(C0,), 5, and powder neutron diffraction has demonstrated that the octahedral sites are occupied by Ca, Cu and C in an ordered fashion. The structure is compared with the cubic phase Ba,CaCu,O,,,. Related phases in which the Ca ions are replaced by other cations are also reported.Keywords: Barium calcium copper oxide; Perovskite; Oxide carbonate; Neutron diffraction The superconducting properties of YBa,Cu307 have stimu- lated numerous studies of the Y,O,-BaO-CuO phase dia- gram. Of particular relevance to the present paper are the phases that have been observed with Ba:Y:Cu ratios of ca. 3:1:2. Preliminary studies'.2 of such materials suggested the existence of two closely related structures, one tetragonal with unit-cell dimensions a=5.81 A, c=8.02 A, and the other cubic. Dysprosium analogues of these phases were sub-sequently found to play a significant role in the formation of epitaxial thin films of DyBa2Cu307 using molecular-beam epitaxy,, and similar phases have been found recently in the Yb,O,-BaO-CuO ~ystem.~The cubic phase containing Y has been shown to have the composition Ba4YC~308,5+x, and a unit cell (a=8.12 A) comprising eight perovskite sub- cek5 Studies of barium calcium copper oxides have recently revealed an analogous phase, Ba4CaCu308.6 1, with a related structure.6 A tetragonal modification was also synthesised, and since no reliable structural information has been reported on such tetragonal phases, a full structural analysis has been performed using neutron diffraction, and the results are pre- sented here.The possibility of replacing Ca by other cations has also been examined. Experimental Samples were synthesised from intimate mixtures of high- purity BaCO,, CaCO, and CuO. The mixtures were heated in air or oxygen for 14 h at 950 "C, re-ground and heated again in air or oxygen at 1000 "C for 14 h.Powder X-ray diffraction (XRD) examination of the products revealed that a single-phase tetragonal material was obtained for Ba:Ca:Cu ratios of ca. 4:1:2. Similar synthetic conditions were employed to examine the possibility of substituting Mg, Y, Sr or Eu for Ca. Ambient temperature time-of-flight powder neutron diffrac- tion data were collected on the diffractometer POLARIS at ISIS, Rutherford Appleton Laboratory, using a vanadium container. Structure refinements used the least-squares program GDELSQ, which is based on the Rietveld method and uses the Cambridge Crystallography Subroutine Librar~.~.~ Scat-tering lengths (/lo-', cm) of 0.525, 0.490, 0.7718 and 0.5805 were assigned to Ba, Ca, Cu and 0,respectively. Results and Discussion The cation ratio Ba,CaCu, produced a black phase with a powder XRD pattern, indicative of a tetragonal unit cell [a =5.800(2)A, c =7.990(3)A] related to a perovskite subcell (size up) with a=J2ap, c=2uP.This contrasts to the body- centred cubic superstructure (a=2up) observed previously for Ba4CaCU308.61.6 For this cubic phase, the ordering of Ca and Cu on the octahedral sites was similar to that exhibited by Li/Na and Sb in Ba4LiSb3OI2 and Ba4NaSb3OI2,' and shown in Fig. l(a). A similar tetragonal supercell to that observed in the present study was previously described for Ba4LiBi301,-x, where a different cation order exists." This structure [space group P4/mmrn, Fig.l(b)] with Ca substituted for Li and Cu for Bi, was therefore adopted for the initial stages of structure refinement using powder neutron diffraction data. During the early stages of refinement it became clear that the central Cu site [Cu(3) at (9,4,+)I was incompletely occupied by Cu, with an apparent site occupancy of 0.65. Vacancies on this site were consistent with the Cu content of the initial reaction mixture being less than for the nominal Ba,CaCu3 ratio required for this sample model. The O(3) and O(4) positions [see Fig. l(b)] around this site were also only partially occupied (around 0.5 and 0.25, respectively), which was consistent with incomplete occupancy of Cu(3). However, they were too close to Cu(3) (1.67 and 1.32 A) for conventional Cu-0 bonds, and the temperature factor for O(3) was very high (7 A2).The introduction of anisotropic temperature factors for this atom suggested that the primary problem related to its position along the [llo] direction, and conse- quently this site was split to give O(3) [ca. (0.25, 0.25, OS)] and O(5) [ca. (0.33, 0.33, OS)] either side of the original position. After the refinement, which resulted in reduced 'R factors, the Cu(3)-0(3) distance (1.9 A) was consistent with a Cu-0 bond, but Cu(3)-0(5) was very short and similar to Cu(3)-0(4) (1.4 and 1.3 A, respectively). Since these distances are very similar to C-0 bonds in metal carbonates (ca. 1.32 A), it was assumed that residual C from the BaCO, and CaCO, starting materials was located on this site, and was covalently linked to three oxygens in adjacent O(4) and O(5) positions.Scattering length considerations (C, 0.6648 x lo-', cm) indicated that the central Cu(3) site was still only partially occupied: ca. 50% by C, 25% by Cu. Two constraints were applied to aid convergence during subsequent refinements, owing to the complex nature of the defect structure in the region of the Cu(3) position. First, it was assumed that Cu at the Cu(3) site was linked only to O(3) to give 4-co-ordinate planar stereochemistry (relaxation of this restriction in a later refinement confirmed its validity). The three 0 atoms of the COS- ions were also positioned statistically on the six octahedral bonds around the (4, +,+) site, i.e.one in O(4) and two in O(5)sites. Although this J. MATER. CHEM., 1991, VOL. 1 ?= I I I7: I I L: oSb eLi \L, 001and0200 Q 003 (a) 804 (b) Fig. 1 The cubic and tetragonal perovskite-related structures of (a) Ba,LiSb,O,,, and (b) Ba,LiBi,O,,-,. For clarity the Ba ions, which occupy central positions in the perovskite subcells. are not marked. The subject of the present study has a structure related to (b) with Ca replacing Li and Cu replacing Bi constraint cannot be in accordance with the real structure since it requires 0-C-0 angles of 90 and 180°, the implied random orientation of the carbonate groups was supported by the previous refinement which indicated an O(5):O(4) ratio of 2:l.The refinement converged to give a central site occu- pancy of 0.52 C and 0.25 Cu, and an overall composition of Ba4CaC~2,2507(C03)o,52.The model provided good agree- ment between observed and calculated neutron diffraction profiles (Fig.2) except for small peaks (marked by arrows), which were attributed to CaO impurity. The regions corre- sponding to the most intense of these peaks were excluded in the final refinement. A refinement in which the Cog- ions were constrained to the geometry of free ions was slightly inferior (higher R factors), and in view of the unusual environment provided by the perovskite structure for such ions, and the possibility of consequential distortions, the 0-C-0 angle constraints were not maintained.Owing to the complex stoichiometry of this phase, two additional samples were synthesised from slightly different mixtures of starting materials (Ba:Ca:Cu ratios 4:1.06:2 and 4:1:1.87), and both appeared single phase from XRD analysis. Since virtually identical results were obtained from all three neutron diffraction data sets, averaged values of the refined parameters are presented in Table 1. The profile R factors, R,, R,, and RE,are indicative of a reliable structural model; the slightly higher values of RI are thought to be due to the influence of the CaO impurity on the weak reflections at low d-spacings. Bond distances and bond strengths around the Ca and Cu ions were calculated using the parameters of Brown and Altermatt," and are given in Table 2.Owing to the resulting high bond-strength sum at Cu( l), alternative calculations using an appropriate Cu3 + parameter' were also performed for this site as indicated in the table. Thermogravimetric reduction (in a 10% H2 in N2 mixture at 930°C to give BaO, CaO and Cu) produced a mass loss of 7.6%, which is in reasonable agreement with that expected (7.0%) for the phase composition suggested by the structure refinements. However, the crystallographic structure implies a higher Cu content than expected from the cation ratios in the pre-fired mixtures. Although this is supported by the presence of small CaO peaks in the neutron diffraction profiles, no Ba-containing impurities were detectable.It there- fore seems likely that some volatilisation of Ba may occur during the synthesis, as was also observed during the prep- aration of Ba4CaCU308.6,.6 The most striking feature of the structure is the presence of C, Cu, and vacancies in a 2:l:l ratio on the central site at (4, 1, 5). Although the presence of car-bonate anions in perovskite structures is uncommon, the complete replacement of layers of octahedral cations by C03 groups has been reported in the perovskite-related phase Sr2C~02(C03).'3The presence of lattice carbonate anions in Ba,CaCu2.2406,96(C03)0.5 is supported by three additional observations: (1) dissolution in dilute mineral acids resulted in the evolution of larger amounts of C02 gas than expected for possible traces of BaC03 contamination.(2) Attempts to synthesise the material from BaO, CaO, and CuO in C02- free oxygen environments were unsuccessful. Subsequent heat- ing of the products in air, however, resulted in the formation of the required phase. (3) Treatment at 350 "C in 200 bar oxygen, which formed part of a wider programme to investi- gate the effects of such conditions on mixed copper oxides, resulted in phase decomposition, and XRD clearly indicated a significant quantity of BaCO, in the products; the carbonate could have originated only from the sample under investi- gat ion. Owing to the disordered nature of the C03 groups in the structure, the proposed model is necessarily a simplistic representation of the true structure since it imposes 0-C-0 bond angles of 90 and 180" instead of the expected 120". In reality, some of the 0 atoms linked to C [0(4) and 0(5)]are likely to be shifted from the positions indicated by the refinement (Table l), in order to achieve appropriate bond angles, and this is supported by the high temperature factors observed for the O(5)atoms.The lower value observed for O(4) suggests that the C03 groups may be oriented with one C-0 bond [to 0(4)] parallel to [OOl], and the two O(5) atoms being displaced from the ideal (x,x,3)sites, mainly along [OOl]. Given this problem of locating the carbonate oxygen atoms accurately, the observed C-0 distances, Table 2, are not dissimilar to those found in other inorganic carbonates (around 1.32 A).It is interesting to note that despite the substantial oxygen deficiency in this perovskite-related phase, the 0(1) and O(2) sites, which are co-ordinated to Ca, are fully occupied. Octahedral stereochemistry is therefore maintained around the Ca ions, which was also observed in the cubic phase Ba4CaCU308.61.6 The Cu(3) site is co-ordinated to four coplanar O(3) atoms, and bond-strength calculations, Table 2, support Cu2+ at this position. The Cu(1) and Cu(2) sites, however, are co-ordinated to some sites that are only partially J. MATER. CHEM., 1991, VOL. 1 -Ill I Ill Ii IIIIII Ill ll ll Ill Ill ll ll I ll I t i ll I i I I I I I1 II I 'I-1.4--I 1.2 --1.0--4 L 0.8-I h P dspacinglA I I 1 I 1 1 1I 1 0.0. 4 L Q,a g 0. 2 c 3 Q) 0. t r n -1 0.4 ' 015 ' 016 ' 017 018 0'.9 ' l'.O l'.l dspacing/A Fig. 2 Observed (dots), calculated (continuous line) and difference neutron diffraction profiles 7atom position Y Y BjA' site occupancy 4i 0 ~21 0.2391(4) 0.64(4) 1 la 0 0 0 0.87( 13) 1 lc 21 21 0 0.15(5) 1 lb 0 0 -1 2.06( 13) 1 1 ~Id -1 21 2 0.86(18) 0.24(3) 1Id 21 + 2 0.86(18) 0.50(4) 2g 0 0 0.2799(8) 2.12( 11) 1 4J 0.2745(4) 0.2745(4) 0 0.77(5) 1 4k 0.267(4) 0.267(4) + 2.8(6) 0.24(3) 2h -1 -1 0.333(3) 2.0(4) 0.25(2) 4k 0.329(4) 0.329(4) 1 3.2(4) 0.232) aTetragonal, P4/mmm, a =5.799(2) A, c =7.992(3) A. Sample I: R, =6.9%, R, =2.8%, R,, =3.2%, RE 2.0%. Sample 2: R, =8.6%, R, =3.0%, RWP=3.4%, RE=2.3%.Sample 3: R,=9.2%, R,= 3.2%, Rwp= 3.5'/0, R,=2.1%. 20 J. MATER. CHEM., 1991, VOL. 1 Table 2 Selected bond lengths and bond-strength summations for Ba,CUCU,,,406,96(C03)0,~ bond bond length/A bond bond length/A bond bond length/A Ca-O( 1)-W)Cu(2)-O( 1) -0(3) -0(5) 2.237(6) 2.251(3)1.759(6) 2.19( 3) 2.69(3) (x 2) (x 4) (x 2) (x 4) (x 4) C~(l)-0(2) 1.85q3) (X 4)-0(4) 2.67(2) (x 2)CU(3)-O( 3) 1.9 l(3) (x 4) bond strength summations" C-0(4) -0(5) 1.33(2) 1.41(3) (x 2) (x 4) Ca 2.56 [for ro(Cu2f)= 1.6793 2.94 [for ro(Cu3+)= 1.7312] occupied, and bond-strength summations were appropriately weighted to account for this feature. Cu(1) has four short coplanar bonds to 0(2),.and in the average unit cell a more distant interaction with a partly occupied O(4) of a carbonate group.Cu(2) has two short bonds at 180" to 0(1), an average of one long bond to O(3) and an even longer bond to an O(5) of a carbonate group. The Cu( 1) and Cu(2) sites appear to be Cu3+ and Cu2+, respectively, on the basis of bond-strength considerations (Table 2). This assignment gives an average Cu oxidation state of 2.44 +, which is in reasonable agreement with the value of 2.20(11) determined by the phase compo- sition. The co-ordination around Cu(2) appears unusual, and ordering of the Cu, C and vacancies in the (3, 3, f-)sites may occur to provide a more favourable co-ordination for Cu(2). Such an ordering process, which must occur over relatively short distances since superstructure reflections are not ob- served, is consistent with the relatively high thermal par- ameters observed for Cu(2), and 0(1) and O(3) to which it is bonded.It is interesting to note that the bond-strength sum at Ca is higher than expected, and this overcompensation of charge also occurs at the octahedral Ca/Y sites in Ba4CaCU308.61 (Ca bond strength sums 2.67) and Ba4YCu308., +x (average Y bond strength sums 3.41 for x =0.5, and 3.76 for x =0 using the data of de Leeuw et a/.'). This feature may, in fact, be fairly common for the occupancy of octahedral sites by Ca or Y in perovskite phases, since a high value for Ca (3.33) is also calculated for the hexagonal perovskite Ba3CaRu2O9.I4 The distribution of Ca ions on the octahedral sites is clearly different in Ba4CaCU2.2406.96(C03)0.~from Ba4CaCU308.61, and the two types of order are closely related to that ob- served in Ba4LiBi30 -and Ba,LiSb,O t,9 respectively.Madelung energy calculations" suggested that the two structures have similar electrostatic stabilities, and subtle effects therefore determine the actual order adopted. For the cubic and tetragonal forms of the copper-containing phases, it is thought that the presence of cation vacancies and Cog- anions within the structure is primarily responsible for the adoption of the tetragonal structure determined here. This possibility can be rationalised by recognising that a common structural feature of both cubic Ba4CaCu308.61 and tetra- gonal Ba4CaC~2.2406.96(C03)0.5is the complete occupancy of the 0 sites co-ordinating to Ca, such that 0 vacancies form links only between two Cu positions.If some of the Cu sites are vacant or occupied by C atoms of Cog -groups, the co-ordination sphere around these sites will be modified such that the co-ordination preference of Ca can be satisfied only with the tetragonal structure. Comparison of the two struc- tures (Fig. 1)shows that only the (+,& f-)site in the tetragonal unit cell can readily accommodate such defects, since it is connected via anions to six Cu sites and allows the Ca co- ordination to remain unchanged. In contrast, the presence of defects on other Cu sites would have a direct influence on 1.91 2.14 Table 3 Unit cell dimensions of related phases ~~ cation ratio in initial reagents alA c/A Ba4MgCu2.25 5.6542) 8.033(3) Ba4Mg1.0Scu2.16 5.661(2) 8.002(3) Ba4YCu, 5.779( 2) 8.O19( 3) Ba4EuCu2.2 5 5.820(2) 8.067( 3) Ba4SrCu,., 5.707(2) 7.903(3) the stereochemistry of adjacent Ca ions and result in incom- plete co-ordination or elongated bonds. Accordingly, it appears relevant that vacancies and Cog- ions are found only at the central site in the tetragonal material. Attempts to substitute Ca for other cations resulted in the formation of several phases with closely related tetragonal structures, and the materials listed in Table 3 were all obtained as single-phase or nearly single-phase products (estimated purity >95%) according to XRD analysis. All these materials are believed to contain lattice carbonate anions, since prepara- tive routes involving pure oxide starting materials and C02- free conditions were always unsuccessful.The Mg, Y and Eu containing samples are thought to be essentially isostructural with Ba4CaCU2.2406.96(C03)0.5, and this is supported for the Y analogue by neutron diffraction data which will be reported elsewhere. The unit cell dimensions (Table 3) are in accordance with ionic radii considerations, except for the sample containing Ba, Sr and Cu (4:1:2.4), which appears anomalous. In particular, the unit cell is smaller than that of Ba4CaCU2.2406.96(C03)0.~,and is consistent with the partial substitution of Ba by Sr on the large, nominally 12- co-ordinate perovskite sites.It is therefore thought to be more closely related to Sr2Cu02C03, in which a complete layer of Cu cations has been replaced by Cog- ions.I3 Our studies of Sr2Cu02C03 have indicated a unit cell with a=5.524(1)A and c =7.496 (2) A, which, as expected, is smaller than that observed for the mixed Ba/Sr phase. A more detailed structural examination of layered phases of this type is planned. We are grateful to SERC for providing financial support, neutron diffraction facilities and a studentship to P.R.S.; we also thank ICI for additional funding. We are grateful for the advice and assistance of S. Hull during the collection of neutron diffraction data. References 1 R. S. Roth, K. L. Davis and J. R. Dennis, Adv. Ceram. Muter., 1987, 2, 303.2 K. G. Frase and D. R. Clarke, Adv. Cerum. Muter., 1987, 2,295. 3 E. S. Hellman, D. G. Schlom, A. F. Marshall, S. K. Streiffer, J. S. Harris Jr., M. R. Beasley, J. C. Bravman, T. H. Geballe, J. N. Eckstein and C. Webb, J. Muter. Res., 1989, 4, 476. J. MATER. CHEM., 1991, VOL. 1 21 4 J. Liang, X. Chen, S. Wu, J. Zhao, Y.Zhang and S. Xie, Solid State Commun., 1990, 74, 509. 10 C. Greaves and S. M. A. Katib, J. Solid State Chem., 1990, 84, 82. 5 6 7 8 9 D. M. de Leeuw, C. A. H. A. Mutsaers, R. A. Steeman, E. Frikkee and J. W. Zandbergen, Physica C, 1989, 158, 391. C. Greaves and P. R. Slater, Solid State Commun., 1990, 73, 629. J. C. Matthewman, P. Thompson and P. J. Brown, J. Appl. Crystallogr., 1982, 15, 167. P. J. Brown and J. C. Matthewman, Rutherford Appleton Lab- oratory Report RAL-87-010, 1987. J. A. Alonso, E. Mzayek and I. Rasines, Muter. Res. Bull., 1987, 22, 69. 11 12 13 14 I. D. Brown and D. Altermatt, Acta Crystallogr., Sect. B, 1985, 41, 244. I. D. Brown, J. Solid State Chem., 1989, 82, 122. Hk. Miiller-Buschbaum, Angew. Chem. Znt. Ed. Engl., 1989, 28, 1472. J. Darriet, M. Drillon, G. Villeneuve and P. Hagenmuller, J. Solid State Chem., 1976, 19, 213. Paper 0102565H; Received 8th June, 1990