J. MATER. CHEM., 1994, 4( l),5-8 Phase Formation and Electrical Properties in the System BaO-Li,O-TiO, Leticia M. Torres-Martinez," Claudia Suckut,b Ricardo Jirnenezbgc and Anthony R. WesP a UANL, Facultad de Quimica, Nuewo Leon, Mexico University of Aberdeen, Department of Chemistry, Meston Walk, Aberdeen, UK A69 2UE CSIC, C. Serrano 7 75dpdo, 28006 Madrid, Spain A survey of compound formation and phase equilibrium in the system BaO-Li,O-TiO, for compositions containing >50% TiO, has been made at subsolidus temperatures, ca. 11 50-1 200 "C. Four ternary phases were encountered, all of which have variable composition. These are phase A, centred on Ba3Li2Ti8020, phase B centred on BaLiPTi6014, phase C based on the line Ba2Til,,Li4x0,, and phase E, a hollandite-like phase Ba~Li~~+4~~Ti~~-~-~0,6.Phase A is new; phase B is the phase responsible for high Li+ ion conductivity reported by Zheng et a/. (Solid State lonics, 1989, 35, 235); phase C is related to one reported by Tillmanns and Wendt (Z.Kristallogr., 1976, 144 16) whose formula coincided with x=O.75; phase E was reported by Sucket et a/. (J. Mater. Chem., 1992, 2, 993). In addition, an extensive area of ternary solid solution based on Ba4Ti13030 is reported. Mechanisms of solid-solution formation are discussed; the two principal mechanisms appear to be Ti4+ +4Li+ and 2Ti4+ +2Li' +3Ba2+. Electrical conductivity data are reported; phases A, C and D appear to be very modest electronic conductors. Few studies on the formation of barium lithium titanate phases have been reported, in spite of the great importance of various barium titanates as dielectrics.The synthesis and crystal structure of Ba,Ti,~,,Li,O,, has been reported;' its structure is somewhat unusual in containing partial occupancy of both Ti and Li positions. A new Li' ion conductor has been reported,, its approximate composition was given but nothing is known of its structure. A new solid-solution phase with a hollandite structure has been rep~rted.~ Its formula is Ba3,Li(,, +4y)Ti8 -2x it appears to be a modest conductor of Li+ ions. We have investigated phase formation and equilibria in the system BaO-Li,0-Ti02 for compositions containing >50% TiO,. These results, with preliminary X-ray characterisation of the phases and electrical property data, are reported here.Experimental Starting materials were Li,CO,, BaCO, and TiO,, all reagent grade. The carbonates were dried at 300°C and the TiO, at 900"C, after which they were stored in a desiccator. Samples were weighed out in 5-20 g batches, mixed into a paste with acetone using an agate mortar and pestle, dried and fired in Pt crucibles, initially at 700-900°C for a few hours to drive off CO,, followed by 1150-1200°C for 2-10 days, with intermediate regrindings, to complete reaction. Reaction con- ditions were found by trial and error to be those necessary to achieve equilibrium, i.e. in which no change in the nature of the reaction product(s) occurred on heating at either higher temperatures or for longer times.Loss of lithia by volatilis- ation was not a problem under the conditions used, as shown by (a) weight-loss checks, (b)the self-consistency of the phase diagram results and (c) constancy in the nature of the final products after appropriate heating schedules. Lithia loss did become significant, however, if samples were melted. Reaction products were analysed and identified by X-ray powder diffraction using either a Philips Hagg Guinier camera for routine identification or a Stoe StadiP, psd-based diffractometer, both with Cu-Ka, radiation. For accurate d-spacings, KC1 was added as internal standard. Melting temperatures were determined approximately from visual observation of pelleted samples, heated isothermally at various temperatures for 5-10 min.Conductivities were measured on sintered, pelleted samples, using electrodes made from fired gold paste. Measurements were made over the range 30 mHz to 64 kHz with a Solartron 1250/1286 Impedance System and over the range 100Hz to 10MHz with a Hewlett Packard 4192 Impedance Analyser. Results and Discussion The results of a selection of heating experiments on ~180 compositions have been summarised in tabular form.? These were used to construct the phase diagram shown in Fig. 1. Four ternary phases were encountered, labelled A, B, C and E, together with an extensive area of solid solutions based on Ba4Til,0,,, labelled D and a small area of solid soiutions based on BaTi,O,,. The diagram in Fig.1 refers to tempera- tures of ca. 1150-1200 "C; these are subsolidus temperatures TiOs 10'Ago Fig. 1 Phase diagram BaTiO,-Li,TiO,-TiO, at 1150-1200"C. Single-phase regions are labelled A-E and BT,. 7 Supplementary data available (SUP 56976, 5 pages); details from Editorial Office but nevertheless, are close to melting for most regions of the phase diagram. Results are now summarised for each of the phases encountered. Phase A This is a new phase which forms over an area of compositions (Fig. 1). We know very little about this phase; its 'ideal' composition may be Ba3Li2Ti,020, since this is a simple formula that lies within the solid solution area at the composi- tion marked in Fig. 1. The phase melts at 1240+20"; melting is probably incongruent, to Ba4Ti13030 and liquid.Unindexed X-ray powder diffraction data are given in Table 1. Conductivity data for phase A are summarised in Fig. 2. A single semicircle was seen in the complex impedance data, without any evidence of low-frequency electrode polarisation phenomena. The conductivity appears to be electronic, there- fore. Actual conductivities are low, e.g. 1.6 x lop7R-' cm-' at 400 "C with an activation energy of 0.84 eV. Phase C Tillmanns and Wendt' reported the synthesis and crystal structure of the phase Ba2Tig.25Li3022. This appears to be the same phase as the one we have labelled C, since its X-ray powder diffraction dat? can be indexedo on an orthorhopbic unit cell [a=5.8040(4) A, b=9.9281(6) A, c= 14.0141(9)A for composition 13.5% BaO, 18.5% Li,O, 68% TiO,] which is very simitar to that rFported for Ba2Ti9.25Li3022[a= 5.8081(5)A, b =9.931( 1)A, c = 14.025( 1) A].Our results indicate that phase C occupies an area of compositions on the phase diagram which appears not, how- ever, to include the composition Ba,Ti,,,,Li,022. This is shown on expanded scale in Fig. 3. One edge of the solid solution area may be described by the general formula Ba2Ti,o-xLi4x022: 0.95 <x < 1.15. The composition of Ba,Ti,~,,Li,O,, may also be represented by this formula, with x=O.75 and is indicated by Tf, Fig. 3; it lies significantly outside our experimental range of values. Table 1 X-Ray powder diffraction data for phase A: Ba,Li,Ti,O,, dobs/A I dobslA I 9.02 14 14 1.8866 13 5.3764 47 1.8756 16 4.5022 57 1.7620 19 4.1416 35 1.7184 15 3.6606 61 1.6938 18 3.6252 27 1.6746 11 3.4300 13 1.6237 26 3.2886 28 1.6005 12 3.1 193 21 1.5767 13 3.0712 93 1.5688 14 2.9931 93 1.5438 13 2.9437 100 1.5365 20 2.8371 89 1.5148 12 2.8127 42 1.4993 15 2.6868 37 1.4903 22 2.6524 44 1.4182 28 2.5054 31 1.4134 23 2.3666 29 1.3993 13 2.3033 15 1.3927 14 2.2495 24 1.3555 14 2.23 16 29 1.2554 13 2.0708 52 1.2505 11 2.0562 52 1.2051 14 1.9800 30 1.1917 13 1.9644 22 1.1877 15 1.9465 67 1.1833 15 1.9004 16 J.MATER. CHEM., 1994, VOL. 4 ~O~WT Fig. 2 Conductivity data 12.5:12575 \\ / X/ Y "4 22.5:12.5:65 15 17.5 20 12.5:22.5:65 mol% Li20 Fig.3 Location of phase C solid solutions: 0, single phase; (3, mixture The crystal structure determination of Ba2Ti,.,,Li,O2, showed a partial occupancy of two of the Ti sites, together with partial occupancy of Li sites;' the existence of a solid- solution range with variable Li :Ti ratio, Fig. 3, may be readily understood, in terms of varying occupancies of Li and Ti sites with the replacement mechanism 4Li+ eTi4+(mechanism I). While such a mechanism may appear unusual at first sight, typic$ Ti-0 and ki-0 bond distances are comparable, ca. 1.96 A and ca. 2.10 A, respectively, for octahedral coordination of the metal, and therefore, there are no size constraints to prevent such a solid-solution mechanism. A similar solid solution occurs in the phase Li,TiO, which, in its high- temperature form with a cation disordered rock-salt structure, can accommodate both an excess and a deficiency of Li+ ion^.^,^ The structure determination' of Ba,Tig~2,Li302, showed J.MATER. CHEM., 1994, VOL. 4 four Ti sites, with occupancy factors as follows: Ti( l),4c, n = 1; Ti(2), 8d, n=0.86; Ti(3), 4c, n=0.88; Ti(4), 4c, n= 1. The three Li sites [Li(l), 4c; Li(2), 4c; Li(3), 4a] were all given occupancy n = %. The solid solution Ba2Ti,,-,Li4,O2, can be accounted for by varying the occupancy of Ti(2), Ti(3) sites and the Li sites. Thus, at the high x limit, 1.15, and assuming the Ti vacancies were equally distributed over Ti(2) and Ti(3) sites, the occu- pancy factor would be n =0.81 with an average Li occupancy factor of n=0.77. Conversely, at the lower x limit, x=O.95 these occupancies would be n=0.84 for Ti and n=0.63 for Li.A full structure analysis would, of course, be necessary to confirm these details. The composition of Ba2Ti,,,,Li,022 reported by Tillmanns and Wendt' was, in fact, obtained by refinement of X-ray diffraction data on a single crystal extracted from a melt whose starting composition, Ti, was rather different from the quoted final composition, Tf, Fig. 3. Given the difficulty in determining accurate Li' site occupancies due to the low X-ray scattering power of Li, and the additional possibility of some partial occupancy of Li on the Ti sites, it is conceivable that the crystal studied in ref.1 could have a somewhat higher Li:Ti ratio to that reported. There is not necessarily an inconsistency therefore between the literature formula and the experimental compositions for phase C reported here. In order to account for the existence of a solid-solution area off the join Ba2Til,~xLi4x022, Fig. 3, an additional solid- solution mechanism is required. One possibility (mechanism 11) is that the same mechanism operates as in the hollandite phase, E, namely: 3Ba+2Li e2Ti. This would have the effect of creating barium vacancies at the same time as the Li, Ti site occupancies vary. A second possibility (mechanism 111)is to simply leave out BaO from the structure, in which case both Ba vacancies and 0 vacancies would be generated.At present, we have no evidence in favour of either mechanism. The directions taken by these two mechanisms are indicated in Fig. 3; as an illustrative example, both are shown starting from the com- position with x = 1, i.e. Ba2Ti9Li402,. Perhaps the Ba vacancy mechanism is more likely since this would not involve disrup- tion of the lattice of TiO, octahedra. Conductivity data are shown in Fig. 2. Conductivities are low, as for phase A, and appear also to be electronic. Phase B This phase appears to be the one responsible for the high lithium ion conductivity.2 It also forms an area on the phase diagram, Fig. 1, and its 'ideal' composition may correspond to the formula BaLi,Ti,014 shown; the phase melts at 1270 20 "C, probably incongruently. One limiting edge of the solid-solution area corresponds to the replacement mech- anism 4Li +Ti, giving the general formula BaLi, +4xTi6-xo14, 0.08~~50.16,Fig.4. In order to account for the area of B solid solutions, similar considerations apply to those used for phase C, above. Unindexed X-ray powder diffraction data are given in Table 2; these agree with the data shown as a 'stick diagram' in ref. 2. Conductivity data for phase B are shown in Fig. 2 for one composition. These are comparable to the literature values; a change in slope occurs at ca. 300°C with activation energies in the low- and high-temperature regions of ca. 0.52 eV and ca.0.43 eV, respectively. Conductivities were measured for several compositions within the phase B solid- solution area. All gave comparable conductivities within a factor of 4 to 5 and no obvious trends were apparent. 5590 25:5:70 9 13 17v 21 525:70 mol% Li20 Fig. 4 Location of phase B solid solutions: 0, mixture; 0,single phase Table 2 X-Ray powder diffraction data for phase B: BaLi,ll'i,O,, dabs/A I 1 ~~~ ~ 7.2856 41 2.9748 19 5.6467 21 2.8947 19 5.0739 22 2.7883 78 4.6656 66 2.7347 51 4.5677 16 2.6675 35 4.3259 36 2.2850 14 3.7702 100 2.0722 54 3.5672 14 2.0182 99 3.4950 23 1.9624 23 3.2076 24 1.7735 18 3.1856 15 1.6920 11 3.0970 23 1.5875 11 3.0698 17 1.4456 50 3.0020 18 1.4380 14 Phase D This phase was originally thought to be a new phase but is, in fact, an extensive solid solution based on the stoichiometric phase Ba4Ti13030.6 It occupies an area on the phase diagram, Fig.1,shown on expanded scale in Fig. 5. The crystal structure Ti02 Ti,+ 4 Li v V V V BaTiO3 Ii2Ti03 Fig. 5 Location of Ba,TiI3O3,,solid solutions of Ba,Ti,O,, has been reported. It contains an array of TiO, octahedra in which oxide ions are in cubic close packing. All atom sites appeared to be fully occupied. In order to account for an extensive area of Ba,Ti,,O,, solid solutions, two solid solution mechanisms are necessary. One possibility is shown as direction A in Fig. 4 and involves the mechanism: 2Ti4+ g 2Li' + 3Ba2+. This presupposes a partial substitution of Li onto Ti sites, together with creation of interstitial Ba2+ ions.The second is shown by direction B and involves the mechanism Ti4+ $4Li+ discussed above in connection with phase C. A grid is superposed on the solid solution area, Fig. 5 which allows the Ba4Ti,03, solid solu- tions to be described in terms of the compositional variables x and y. Conductivity data are shown for one phase D composition in Fig. 2; conductivities are again very low. Phase E This phase, a modest conductor of Li+ ions with the hollandite structure, has been described previously.2 It is a solid-solution phase of general formula, Ba,,Li,,, + 4y)Ti(8 in which -,,-,,)016 two principal mechanisms operate, as described above, i.e.2Ti g2Li + 3Ba and Ti 4Li. Its location is shown in Fig. 1. Subsolidus Compatibility Relations The subsolidus phase relations in the system BaTiO,-Li,TiO,-TiO, are shown in Fig. 1. These refer to temperatures of 1150-12OO0C, which are within ca. 50°C of J. MATER. CHEM., 1994, VOL. 4 melting temperatures; at lower temperatures, samples did not usually reach equilibrium within timescales of a few days. The phase diagram contains the four ternary phases, A, B, C, E described above, together with a significant area of Ba,Til,030 ternary solid solutions, D; a small area of BaTi,O,, ternary solid solutions appears to form also. The remaining areas of the phase diagram are divided into a number of two- and three-phase regions; some are marked by 2 and 3, respectively, in Fig.1. Coexisting phases include the ternary solid solutions indicated above and various binary phases, notably BaTiO,, BaTi,O,, Li,Ti,O, and Li,TiO,; the latter forms a very extensive solid solution range at 1150-1200 "C in the Li,O-TiO, diagram (ref.4) and these solid solutions coexist with various of the ternary barium lithium titanate phases. We thank the British Council for supporting the Aberdeen-Mexico exchange programme, the SERC for finan- cial support (to A.R.W.) and DGICSA, SEP C91-03-19-001-365 and C91-01-001-856 (LMTM). References 1 E. Tillmanns and I. Wendt, 2.Kristallogr., 1976, 144, 16. 2 W. J. Zheng, R. Okuyama, T. Esaka and H. Iwahara, Solid State ionics, 1989,35,235. 3 C. Suckut, R. A. Howie, A. R. West and L. M. Torres-Martinez, J. Muter. Chem., 1992,2,993. 4 G. Izquierdoand A. R. West, Muter. Rex Bull., 1980, 15, 1655. 5 A. R. West, J. Muter. Sci. Lett., 1981, 16, 2023. 6 E. Tillmanns, Inorg. Nucl. Chem. Lett., 1971,7, 1169. Paper 3/04518H; Received 28th July, 1993