Structural Investigation of Ba6-xLn8+2/3xTii8054 Isostructural Ce Compound Strukturne raziskave izostrukturne Ce spojine s trdnimi raztopinami tipa Ba6-xLn8+2/3xTiis054 C. J. Ravvn1, D. Makovec, Z. Samardžija, D. Kolar, IJS Ljubljana Prejem rokopisa - received: 1996-10-04; sprejem za objavo - accepted for publication: 1996-11-22 Single crystals were synthesized and single crvsjal x-ray diffraction data revealed an orthorhombic unit celi with lattice parameters a = 22.377(6), b = 7.629(2) and c = 12.233(3) A. The majority of the systematic extinctions were indicative of a primitive celi, with the intensity data best fitting the centrosymmetric statistics, indicating the correct space group was Pnma (number 62). The structure is isostructural with crystals obtained from the Ba6-xLn8+2/3xTii80s4 (Ln = La, Pr, Nd, Sm, Eu, and Gd) solid solution series. Refinements in space group Pnma, resulted in R = 11.03%. The formula, calculated from refined site occupancies, results in 8.0+ 0.1 Ba atoms and 19.3+ 0.3 Ce atoms requiring 109.0± 0.6 O atoms for charge neutrality. Taking into account that the single crystal synthesis was in flovving Ar at high temperatures, some of the Ti4+ could be reduced to Ti3+. A minimum Ti3+/Ti4+ ratio of 1.9/34.1 would result in 108 O atoms for a formula of Ba8.oCeig.3Ti3+i.gTi4+34.iOw8 corresponding to 4.7 ± 0.1 at.% Ba, 11.3 + 0.2 at.% Ce, 21.01 at.% Ti, and 63.03 at.% O. Electron probe wavelength dispersive spectroscopic (WDS) microanalysis results in 4.9 ± 0.1 at. % Ba, 10.9 ± 0.2 at. % Ce, 20.4 ± 0.4 at. % Ti, and 63.8 ±0.6 at. % O. Key VVords: barium cerium titanate, single crystal x-ray diffraction, waveiength dispersive spectroscopic microanaiysis, structure Monokristaie smo pripravili z žganjem zmesi BaTiC>3 : CeC>2 : HO2 v pretoku Ar pri 1350°C. Monokristalna rentgenska difrakcij$ je pokazala primitivno ortorombsko osnovno celico s celičnimi parametri a = 22,377(6) A, b = 12,233(3) A in c = 7,629(2) A in prostorsko grupo Pnma (številka 62). Strukturo, kije izostrukturna s strukturo trdnih raztopin tipa Ba6-xLn8+2/3xTiisOs4 (Ln = La, Pr, Nd, Sm, Eu in Gd), lahko opišemo kot mrežo TiOf' oktaedrov, ki tvorijo pentagonalne in rombine (perovskitne) kanale, v katerih se nahajajo Ba in Ce atomi. V formuli spojine, izračunani na osnovi predhodno določene zasedenosti posameznih mest v strukturi, je 8,0 + 0,1 atomov Ba in 19,3 + 0,3 atomov Ce, kar zahteva 109,0 ± 0,6 atomov O za izenačitev naboja. Ker so bili monokristaii sintetizirani pri visoki temperaturi v atmosferi Ar, lahko pričakujemo, da je del Ti v reducirani obliki Ti3+. Razmerje Tp+/Ti4+ = 1,9/34,1 odgovarja formuli Ba8.oCei9.3Ti4+34.iTi3+i.gOWe s sestavo 4,7 ± 0,1 at.% Ba, 11,3± 0,2 at.% Ce, 21,01 at.% Ti in 63,03 at.% O. Mikroanaliza WDS v 19 točkah na štirih različnih monokristalih je pokazala sestavo: 4,9 +0,1 at.% Ba, 10,9 ± 0,2 at.% Ce, 20,4 ±0,4 at. % Ti in 63.8 +0,6 at. % O. Ključne besede: barijev cerijev tianat, monokristalna rentgenska difrakcija, mikroanaliza, strukture 1 Introduction Ba6-xLn8+2/3xTiisC>54 based microwave dielectrics are of commercial interest since they display: 1) High di-electric constants (k') that enable the miniaturization of devices. 2) Modest Q factors that transport, filter, and/or store electromagnetic energy with minimal losses (Q = l/tan5). 3) Low temperature coefficients of resonant fre-quency (tf) that minimizes frequency drift due to temperature fluctuations attributed to the environment and/or circuit heating. Past investigations have focused on the properties, phase equilibria, and crystal structures for analogues vvhere Ln = La, Pr, Nd, Sm, or Gd. This pre-sent investigation focuses on the isostructural single crystals that result when Ce substitutes for the Ln atom. In the single crystal x-ray diffraction investigation of the Pr-analogue, Matveeva et al.1 obtained a formula of Ba4Pr9.3Tii8054 (4.69 at.% Ba, 10.90 at.% Pr, 21.10 at.% Ti, and 63.31 at.% O) by refining on the site occupancies and a formula of Ba3.75Pr9.5Tii8054 (4.40 at.% Ba, 11.14 at.% Pr, 21.11 at.% Ti, and 63.34 at.% O) from x-ray spectral microanalysis. In the single crystal x-ray diffraction refinements of the Nd-2 and Sm-analogues3 the re- Dr. Claudia J. RAWN Inštitut J()žcf Štefan 1000 Ljubljana. Jamova 39 sulting compositions of the crystals were determined only by the site occupancies. An analytical method of de-termining the composition is preferred over the statistical composition that results from refining on the site occupancies, especially with Ba6-xLn8+2/3xTiis054 type crys-tals where the two heavy cations have a similar number of electrons (in the čase of La3+ and Ba2+ the number of electrons is equal) creating difficulty in resolving be-tween the two atoms using x-ray diffraction. Due to the strong overlapping of the most intense Ba and Ti spectral lines, compositional analysis using Energy Dispersive X-ray (EDX) microanalysis is also difficult. In this present study, WDS microanalysis was employed as the tech-nique to determine the composition of the crystals. WDS microanalysis, of samples containing both Ce and Ba, is also difficult due to the strong overlapping of the CeLai,2 and BaLpi,4 lines. To overcome this difficulty the calibration curve method was employed4. 2 Experimental Single crystals were prepared by firing an oxide mix-ture of 8.5 mol% BaO, 16.5 mol% Ce02, and 75 mol% Ti02 (BaTiCb - Transelco 219-8, Ce02 - Koch - Light Lab., 99.9%, and TiCh - anatase modification, Fluka AG, > 99%). The mixed powders were pressed into a pellet, with a 8 mm diameter and approximately 8 mm high, in flovving Ar at 1350°C for 12 h. The pellet vvas partially melted and on the surface needle-like crystals, approxi-mately 0.20 x 0.04 x 0.04 mm, were present. 2. / Single crystal x-ray diffraction data collection, proc- essing, and refmement Single crystal x-ray diffraction data were collected on an Enraf-Nonius CAD-4 diffractometer using MoKa ra-diation (K = 0.71073 A). By examining the systematic extinctions it was found that the majority of the system-atic absences were indicative of a primitive celi, space group Pna2i (number 33) or Pnma (number 62). Data were processed using MULTAN 1 1/825 and the structure vvas refined using SHELXL-936 against the structure fac-tor, F. The reflections were corrected for Lorentz and po-larization effects. The crystal faces were well defined making an analytical absorption correction a suitable choice. Once the composition was determined the data were reprocessed using the corrected composition to be applied to the absorption correction. Electron probe WDS microanalysis Polished surfaces were prepared by conventional metallographic techniques. The WDS microanalysis vvas performed on a JEOL JXA 840A electron probe mi-croanalyser using TRACOR softvvare for quantitative analysis. Measurements for Ba, Ce, and Ti vvere carried out on a LiF crystal at 20 kV and 15 nA. Counting tirne vvas set to assure a less than 1 % of the standard counting deviation. Oxygen content vvas calculated by difference. 3 Results and discussion MULTAN shovved the intensity data best fit the cen-trosymmetric statistics, indicating the correct space group vvas Pnma. Refinement of the data in space group Pnma, using 131 parameters, resulted in R1 = 11.03%, wR2 = 36.19%, and a goodness of fit = 0.991, with the thermal parameters of the Ce and Ba atoms refined iso-tropically and the thermal parameters of the Ti and O atoms refined anisotropically. The basic framevvork of the crystal structure, shovvn in Figure 1 (the ORTEP7, of the xz projection, generated from the fractional coordinates), is made up of corner sharing Ti062" octahedra linking to produce pentagonal and rhombic channels, identical to the framevvork for Ba6-xLns+2/3xTii8054 type analogues. Refinement on the Ba and Ce site occupancies re-vealed the Cel site to be 95.8% occupied (4.2% vacan-cies), the Ce2 site to be 95.6% occupied (4.4% vacan-cies), the Ce4 site to be 86.2% occupied (13.8% vacancies), the Ce5 site to be 98.8% occupied (1.2% va-cancies), the Ce3 site is shared vvith Ba (Ce/Ba 97.2%/4.88%, resulting in a site occupancy greater than 100%), the Bal site is shared vvith Ce (Ba/Ce 96.6%/4.92%, resulting in a site occupancy greater than 100%), and the Ba2 site is shared vvith Ce (Ba/Ce 99.0%/4.80%, resulting in a site occupancy greater than Figure 1: The ORTEP7 of the xz projection, generated from the atomic coordinates. shovving the basic framevvork of the crystal structure, made up of corner sharing TiOfi"2 octahedra linking to produce pentagonal and rhombic channels Slika 1: Slika strukture spojine v xz projekciji. Slika je bila narejena s pomočjo računalniškega programa ORTEP7. Strukturo sestavljajo oktaedri TiO«"2, med seboj povezani preko ogljišč, tako da tvorijo pentagonalne in romboederske kanale Table 1: Summary of the refmements on site occupancies of the Ce and Ba atoms Site % Ce Ce atoms % Ba Ba atoms % vacancy atom total Cel 95.8 3.832+0.054 0 0 4.2 3.832+0.054 Ce2 95.6 3.824±0.054 0 0 4.4 3.824+0.054 Ce3 97.2 3.888±0.062 4.88 0.195±0.003 0 4.08310.065 Ce4 86.2 3.448±0.059 0 0 13.8 3.44810.059 Ce5 98.8 3.952±0.055 0 0 1.2 3.95210.055 Bal 4.92 0.197±0.003 96.6 3.86410.066 0 4.06110.069 Ba2 4.80 0.192±0.003 99.0 3.960±0.063 0 4.152+0.066 total_19.3±0.3_8.0+0,1 100%). The sites with occupancies totaling greater than 100% could result from the difficulty in distinguishing between atoms, differing by only one e" (Ba2+ = 54e~ and Ce3+ = 55e~), using x-ray diffraction. The above site occupancies are summarized in Table 1. For 8.0 ± 0.1 Ba2+ atoms, 19.3 ± 0.3 Ce3+ (ali of the Ce4+ reduced to Ce3+) atoms, and 36 Ti4+ atoms, 109.0 ± 0.6 O2" atoms would be required to charge balance the cations, however, only 108 O atoms were found to be present in the refinement. Taking into account the single crystal synthesis in flow-ing Ar at high temperatures, some of the Ti4+ could be reduced to Ti3+, requiring a minimum Ti3+/Ti4+ ratio of 1.9/34.1 to account for 108 O atoms and result in the formula Ba8.oCei9.3Ti3+i.9Ti4+34.iOio8. The formula Ba8.oCei9.3Ti3+i.9Ti4+34.iOio8 represented in at.% is 4.7 ± 0.1 at.% Ba, 11.3 ± 0.2 at.% Ce, 21.01 at.% Ti, and 63.03 at.%0. This agrees well with the WDS microanalysis, averaged from nineteen measure-ments on four crystals, resulting in 4.9 ± 0.1 at.% Ba, 10.9 ± 0.2 at.% Ce, 20.4 ± 0.4 at.% Ti, and 63.8 ± 0.6 at.% O. WDS measurements, taken from nine points on one crystal, supports the claim that the crystals are com-positionally homogeneous. As a check on the reliability of the structure the bond-valence sums, as outlined by Brese and 0'Keeffe8, were determined. Table 2 summarizes the average cat-ion-oxygen bond lengths, bond-valence sums, the num-ber of oxygens bonded to each cation, and the average cation-oxygen bond length, for positionally similar cations, of the isostructural Sm-analogue3. The bond-va-lence sum should be approximately equal to the valence of the cation and for chemically similar sites the bond-valence sum should be nearly equal. Refinement on the site occupancies suggest that the Cel, Ce2, and Ce5 sites are chemically similar, however, this is not reflected in the bond-valence sums for these cations. The bond-valence sum for Cel is exceptionally high, the bond-valence sums for Ce2 and Ce4 are similar although refinement on the Ce4 site occupancy suggests a much higher vacancy content than the Ce2 site, and the bond-valence sums for Ce3 and Ce5 are similar although refinement on the Ce3 site occupancy suggests that the site is shared by both Ce and Ba while refinement on the Ce5 site occu-pancy suggest a small number of sites vacant. The results from the refinements on site occupancies suggest that the two pentagonal Ba sites are similar, hovvever, this is not indicated by the bond-valence sums. The bond-valence sum for Ba2 is exceptionally low due to the scarcity of short Ba-0 bond lengths, this is also reflected in the high average bond length. Only five oxygens are found to be bonded to Ti3 while Ti2 and Ti5 both have one bond length that is unusually long (Ti2-018 = 2.25 A and Ti5-04 = 2.37 A). The bond-valence sums for these three Ti atoms are lower than for Til and Ti4, suggesting a greater deviation from the ideal octahedral symmetry. A wider range of average bond lengths is observed for the structure reported here compared to the isostructural Sm-analogue. Table 2: Summary of average cation-anion bond lengths, bond-valence sums, and the number of anions bonded to each cation cation anion average bond- average bond bonds bond valence length (A) for the length sum isostructural Sm- (A) analogue3 Cel 11 2.63 3.44 Sml 2.69 Ce2 11 2.68 2.97 Sm2 2.70 Ce3 12 2.74 3.10 Sm3 2.71 Ce4 10 2.65 2.99 Sm4 2.70 Ce5 12 2.73 3.11 Sm5 2.69 Bal 12 2.98 2.15 Bal 3.07 Ba2 13 3.14 1.76 Ba2 3.09 Til 6 1.94 4.35 Til 1.95 Ti2 6 2.02 3.72 Ti 2 1.95 Ti 3 5 1.93 3.64 Ti 3 1.95 Ti4 6 1.97 4.05 Ti4 1.96 Ti5 6 2.01 3.90 Ti5 2.00 The high R factor (11.03%), for a structure domi-nated by heavy atoms, and the low correlation betvveen bond-valence sums, for chemically similar Ce and Ba sites, indicates that not ali the structural details are fully understood. The high R factor, shared Ce/Ba sites with occupancies greater than 100%, and bond-valence sums that show no correlation for chemically similar sites most likely occur due to the inability of x-ray diffraction techniques to distinguish between atoms that differ by only one electron (Ba2+ = 54e" and Ce3+ = 55e"). 4 Conclusions Single crystals, isostructural with Ba6-xLns+2/3xTiis O54 (Ln = La, Pr, Nd, Sm, Eu, and Gd) type crystals, were grown with Ce in the Ln position. The single crystal x-ray diffraction refinement was conducted in space group Pnma and resulted in R = 11.03%. The uncertainties of the at% Ba and Ce. determined from WDS microanalysis and by refining on the site occupancies of the Ce and Ba positions, overlap. The resulting composition supports the idea that some of the Ti4+ has been reduced to Ti3+ (Ba8.oCei9.3Ti4+34.iTi3+i.9Oios). WDS microanalysis also indicates that the crystals are compositionally homoge-neous. 5 References 1 R. G. Matveeva, M. B. Varfolomeev, and L. S. Il'yushchenko, Russ. J. Inorg. Chem. (Engl. Trans.), 29, 1984. 17-19 2 D. Kolar, S. Gaberšček, and D. Suvorov, Third Euro-Ceratnics, V2, eds. P. Duran and J. F. Femandez, pp. 229-34, 1993 3 C. J. Rawn, Ph. D. Dissertation, University of Arizona, 1995 4 Z. Samardžija, M. Čeh, D. Makovec, and D. Kolar, Mikrochimica Acta., in press 1996 5 P. Main, MULTAN 11-82, A Svstem of Computer Programs for the Automatic Solution of Crystal Structures from X ray Diffraction Data, 1982 6 G. M. Sheldrick. SHELXL 93, Program for Crvstal Structure Refinement, University of Gottingen, Germany, 1993 7 M. N. Burnett and C. K. Johnson, ORTEP-III, Report ORNL-5138, Oak Ridge National Laboratory, Tennessee, USA, 1996 8N. E. Brese and M. 0'Keeffe, Acta Cryst„ B47, 1991, 192-97 Acknowledgments The authors would like to thank L. Golič and A. Meden for the single crystal x-ray diffraction data collec-tion and discussions pertaining to the structure.