Structural phase transition near 825 K in titanite: Evidence from infrared spectroscopic observations

Transcription

1 Amerian Mineralogist, Volume 82, pages 3-35, 1997 Strutural phase transition near 825 K in titanite: videne from infrared spetrosopi observations MING ZHANG, 1 KHARD SALJ,1,2 AND ULRICH BISMA YR3,4 'Department of arth Sienes, University of Cambridge, Downing Street, Cambridge CB2 3Q, U.K. 'Interdisiplinary Researh Centre in Superondutivity, Madingley Road, Cambridge, CB3 OH, U.K. 3Mineralogish-Petrographishes Institut, Universitat Hamburg, Grindelallee 48, D-2146 Hamburg, Germany 4SFB 173, Universitat Hannover, D-36 Hanover, Germany ABSTRACT We report the diret experimental observation of a strutural anomaly near 825 K in syntheti and natural titanite samples by high-temperature, hard-mode infrared spetrosopy. The anomaly in titanite, CaTiSiOs, is haraterized by a break of the temperature dependene of the 562 m-l Si-O bending mode, the 675 m-i Ti-O band, and the 9 m-i Si-O strething modes, and the rapid derease of the IR signal at 873 m-i. The order parameter, as determined from the temperature evolution of the frequenies of the absorption bands in the middle infrared (MIR) region follows a seond-order Landau behavior with an order-parameter exponent f3 = 1,6.At T > 825 K, the Ti-O band shows further softening, whereas the Si-O bands at 562 and 9 m-i show hardening with inreasing temperature. In natural samples, the effets of impurities on the high-temperature transition are weak. For pure titanite, the transition temperature, T" is about 825 K and inreases with inreasing impurity onentration. INTRODUCTION The room-temperature struture of titanite was first studied in 193 (Zahariasen 193) and refined by Speer and Gibbs (1976). Speer and Gibbs (1976) desribed the room-temperature phase as being anti ferroeletri, in whih the Ti atoms were displaed from the enters of otahedra. A high-temperature X-ray study of pure titanite by Taylor and Brown (1976) showed a reversible, displaive phase transition at approximately 5 K. Above that temperature the spae group was reported to be A entered (A2Ia), whereas below 5 K the spae group is P2/a. The transition near 5 K was seen as anomalies in the temperature evolution of the optial birefringene and X-ray diffration intensities (Bismayer et al. 1992), some rather weak hanges in the lattie parameters (Ghose et al. 1991; Bismayer et al. 1992), and very strong hanges in the frequenies and sattering intensities of Raman-ative phonons (Salje et al. 1993) and in the IR-absorbane intensities. The marked softening of infrared-ative Ti-O phonons with amplitudes along the rystallographi a axis orrelates well with the observation of dieletri losses and the measurements of the exess speifi heat, Cp, at high temperatures (Zhang et al. 1995). By fitting the temperature evolution of the order parameter Q(T) (T < TJ to Q ex IT - TI~,Bismayer et al. (1992), Salje et al. (1993), and Zhang et al. (1995) noted that the effetive exponent, f3 = V8,is ompatible with that of a two-dimensional Ising model. The question, however, is whether the phase at T :2:5 K is the orret 3-4X/97/12-3$5. 3 paraphase of titanite or, alternatively, whether this phase is an intermediate phase and the true high-symmetry phase exists only at muh higher temperatures. The previous experimental evidene for this senario relies on the following observations. First, Bismayer et al. (1992) and Meyer et a!. (1996) reorded nonzero exess birefringene, I1n, at temperatures up to 85 K, whih is ompatible with the assumption that an intermediate phase exists. Seond, Salje et a!. (1993) observed that the exess intensities of the Raman signals derease gradually at T > 5 K with inreasing temperature while additional, very weak lines our near 4 m-i. This observation rules out the idea that the phase transition at 5 K ours between a paraeletri high-symmetry phase and its antiferroeletri low-symmetry phase by a simple shift of the Ti positions in the Ti6 otahedra. This effet has nothing to do with the additional effet of mobile antiphase boundaries (Van Heurk et al. 1991) simply beause only very few antiphase boundaries exist in pure, syntheti titanite rystals (Chrosh et a!., in preparation). The essential question we wish to answer in this paper is whether a seond phase transition ours at high temperatures. Hard-mode spetrosopy was employed in this study beause it has been shown that small strutural hanges orrespond to strong hanges in the phonon spetra (Zhang et al. 1995). It was expeted that IR spetrosopy would provide a suffiiently sensitive tool to detet the additional high-temperature phase transition in titanite. This expetation proved to be orret, and we report the first lear evidene that a seond phase transition ours near 825 K.

2 ZHANG T AL.: STRUCTURAL PHAS TRANSITION IN TITANIT 31 SAMPLS AND XPRIMNTAL MTHODS The syntheti sample used in our experiments had been extensively studied by optial birefringene and X-ray diffration tehniques (Bismayer et al. 1992), Raman (Salje et al. 1993) and infrared spetrosopy, and dieletri and DSC measurements (Zhang et al. 1995). In addition, the thermal behavior of natural titanite was also investigated. One natural sample is from Rauris, Austria (Fe 1.8%, Al 3.8%). Two other samples were provided by the Smithsonian Institution in Washington, D.C. (B2323 with Fe 2.8% and Al.5%, and R497 with Fe 8.8% and Al 13.9%). The natural samples were reently studied by IR spetrosopy, Raman spetrosopy, optial birefringene, and X-ray diffration (Meyer et al. 1996). The powder samples were obtained by ball milling the rystals in a Spex miromill for 3 min. The even-sized, fine-grained powder was kept in a drying oven at 12 C to prevent absorption of water. KBr and CsI were used as the matrix materials. The absorption spetra were reorded under vauum with the use of a Bruker 113V FT-IR spetrometer. The sample pellets were heated in a ylindrial platinumwound furnae oupled with a urothem type-815 temperature ontroller with a temperature stability of 1 K. xperiments were performed between 2 and 5 m-t at room temperature and in the region of ~5-5 m-i on heating and ooling between 3 and 94 K. ah single-beam spetrum was integrated over either 25 or 15 sans with instrumental resolution of 2 m-i and zero filling fator of 4. The software RAZOR (Spetrum Square Assoiates) and OPUSIIR (Bruker Analytishe Messtehnik GMBH) were used for data analysis. RSULTS AND DISCUSSION The room-temperature spetra of Rauris, B2323, R497, and the syntheti titanite are plotted in Figure I. The spetra are dominated at high frequenies by the IR band near 9 m-i, whih is attributed mainly to Si4 strething modes. Several spetral features are obvious without any analysis. First, the spetra of two natural samples (Rauris and B2323) are similar to that of the syntheti sample. Seond, R497 has muh broader bands, whih may be mainly due to loal struture disorder as well as impurities and defets in the sample. The sample is probably metamit. Third, the varying height of the Ti-O feature near 685 m-i results from the oupled substitution of Al and Fe for Ti and F for in the natural samples (Meyer et al. 1996) and is also related to the rystallinity of titanite (Chen et al. 1993). The temperature evolution of MIR spetra of natural and syntheti titanite samples is shown in Figures 2 and 3. The effet of temperature is most learly seen as an inrease in band width, a derease in band intensity, and a softening of phonon bands. With inreasing temperature, the Si-O bending at 563 m-i and the Si-O strething band at about 9 m-i exhibit softening in frequeny below 825 K. On further heating they harden again at temperatures above Syntheti Rauris R497 Titanite 3 K (m-1) FIGUR 1. Absorption spetra of natural (Rauris, B2323, and R497) and syntheti titanite samples. 825 K. Figures 4 and 5 show the break of the temperature dependenies of these phonon modes. Heating to 825 K, the IR band near 875 m-i broadens and further dereases in intensity. This feature is seen most learly in sample R497 (Fig. 3B). The broad and intense band near 675 m-i is assoiated with Ti6 otahedral strething modes. This signal is polarized stritly along the rystallographi a axis (Zhang et al. 1995). Its preise peak position may be influened by the high dieletri ontrast between titanite and the embedding medium and the substitution of Al and Fe for Ti and F for in the natural samples. The main effet of temperature on this band is the derease of the band frequeny. The integrated intensity also dereases below 825 K and then dereases with muh stronger temperature dependenies for well-rystallized samples (Figs. 6 and 7). The break of the temperature dependene of the phonon frequenies at 563, 685, and 9 m-i and the rapid derease in the IR signal at 875 m-i at about 825 K show that the anomaly is related to strutural hanges, i.e., the strutural state above 825 K is different from that below 825 K, whih onfirms the existene of the previously suspeted intermediate phase between 5 and 85 K (Bismayer et al. 1992; Salje et al. 1993; Zhang et al. 1995). In hard-mode spetrosopy (Bismayer 199; Salje 1992, 1994) the order parameter determined by the frequeny shifts and profile hanges of the infrared absorption is inherently loal in harater and reveals the mirosopi harater of the transition rather than its marosopi nature. The expeted frequeny hanges are orrelated with the order parameter. In lowest order theory, this orrelation is (Salje 1992) i1wf(t) = (w? - wj(w? + w,) = O;Q2(T) (1) where Q is the order parameter as "seen" by the phonons. With (w? + w,) approximately onstant, the frequeny shift beomes

3 32 ZHANG T AL.: STRUCTURAL PHAS TRANSITION IN TITANIT A Syntheti Titanite A ~~C :J C 9 K :J.::: ;.::: ~; ~ '-' '-' C C/) C/) ««3 K (m-1) (m-1) Natural Titanite : " ' :J.::: ; ~~~ '-' C C/) ««: " ' :J > C'-' C/) 88 K 32 K (m-1) FIGUR 2. evolution of the IR spetra of titanite samples (A) Rauris and (B) B2323 between 5 and 12 em-i. The temperature interval is 2 K (m-1) FIGUR 3. Stak plot of the IR spetra of (A) syntheti titanite and (B) sample R497 between 72 and 1 em-i, showing details of the spetral hanges during the strutural phase transition. The temperature interval is 2 K. t:.w,(t) = (w? - w,) < Q2(T). As shown in Figures 4-6 for IR bands near 563 and 9 em-i, there is a strong hange in aw/at at about T, = K, whih is assoiated with a strutural phase transition. Saling the data at T > 5 K to the redued tem- (2) perature with a ritial exponent, the temperature dependenies of t:.w(t) and Q2(T) an rewritten as t:.w;(t)< Q2(T) <(T, - T)2~T < T, (3) where, 13 = 12- This orresponds to lassial seond-order

4 ZHANG T AL.: STRUCTURAL PHAS TRANSITION IN TITANIT A 582 ;=- 56 ~ o 558 '';::; ' o 7,j ~ 558 o :;:;.~ 556 -" o o ~ 556 ' o 7,j ~56 u ~558 ~556..:::t: D I FIGUR 4. The temperature dependene of the IR band at 563 em-i. The slope hange near 825 K indiates a seond-order phase transition in the syntheti sample (A), the sample from Rauris (B), B2323 (C), and R497 (D). The average errors for the peak positions are.25 em-i near room temperature and 1. em-i for high temperatures. The lines are visual guides. behavior, whih has been predited (Salje et al. 1993; Salje 1993). Comparison of the temperature dependene of the Ti-O phonon frequeny at 675 m-1 with that of other Si-O bands reveals a systemati differene. Although the Ti-O band softens at T > 825 K, the Si-O modes at 563 and 9 m-1 harden. This observation has physial signifiane: It shows that the Si4 and Ti6 polyhedra reat differently to the strutural phase transition. Although the Si4 tetrahedra behave as a rigid body both above and below T, they beome softer near Surprisingly, this is T" the ase not only for the O-Si-O bending motion but also for the Si-O strething modes. This softening may reflet a general weakening of the Si-O band beause of the polarization effets of Ca, i.e., one expets the Ca position to be diretly involved in the transition mehanism. The behavior of the Ti6 otahedron is very different from that of the Si4 tetrahedra. The strething mode of the otahedron shifts to lower frequenies on heating at temperatures below T, and almost ollapses in the hightemperature phase. It is diffiult to imagine that the near ollapse stems simply from a deformation of the otahedron. It appears more likely that the Ti position inside the otahedron hanges during the transition. A probable senario is that the Ti atoms are distributed over split positions at T < T, (i.e., the struture remains loally antiferroeletri on a long time sale in omparison with that of phonons), whereas at T > T, the Ti atoms move in one global, shallow potential well. In summary, we may envisage the phase transition to be related to hanges in the Ca positions, a weak distortion of the Si4 tetrahedra, and a hange in the loal potential of the Ti positions (but with little effet on the volume-averaged Ti position). We now fous on the effets of the impurities in natural titanite on the phase transitions. With substitution of AI and Fe for Ti, the low-temperature phase transition at about 5 K is smeared in natural samples (Figs. 4 and 7), so the transition point ould not be determined. We suspet that the weak break in the temperature dependene of the phonon at 563 m-1 in the sample R497 at about 58 K orresponds to the phase transition at 5 K in pure titanite. The impurities in the natural titanite exhibit muh less effet on the seond phase transition near 825 K, however. The three natural samples show a thermal behavior that is similar, within our experimental res-

5 34 ZHANG T AL.: STRUCTURAL PHAS TRANSITION IN TITANIT 895 ~: :;::; '" ro C]) ~892 : :;::; '" roc]) 9 A FIGUR 5. The temperature evolution of the Si-O strething band near 899 m-'. (A) Syntheti sample; (B) B2323. olution, to that of the pure syntheti sample (Fig. 4). The Rauris and B2323 samples, whih have lower onentrations of Al and Fe, show nearly the same transition temperature as pure titanite. The transition point for the : ~ g 675 :;::; - o ~ 67 roc]) FIGUR 6. The IR band near 685 m-' of sample B2323 shows strong phonon softening at temperatures above 825 K. B :J.i ~ 22 ~ - a5 2 - C]) 18 ;, FIGUR 7. The integrated intensity of the 687 m-] band in sample B2323 shown as a funtion of temperature. For integration, a linear baseline between the spetral minima on either side of the peaks was used. The hange in temperature dependene between 5 and 6 K orresponds to the phase transition at 5 K for the syntheti titanite sample (Bismayer et al. 1992; Salje et al. 1993; Zhang et al. 1995). The lines are visual guides. sample, R497, with the most impurities is modified by the effet of the impurities and shifts to higher temperature. The different dependenies of the two transition mehanisms on doping of the Ti positions allow us to haraterize the driving fores of the phase transitions even further. The phase transition near 5 K is antiferroeletri on a marosopi sale. The transition shows a strong dependene on the hemial oupany of the position that is filled by Ti in the pure material. This result is orrelated with the observation that the antiferroeletri displaement of Ti in the Ti6 otahedron is the essential driving mehanism of the low-temperature phase transition. The high-temperature phase transition shows a ompletely different behavior. It does not depend muh on any dopant replaing Ti, nor does the Ti-O strething mode show hardening in the high-temperature phase as disussed before. These observations indiate that the Ti6 otahedron plays only a minor role in the mehanism of the high-temperature phase transition. If we assume, furthermore, that the weak hanges in the phonons related to the Si4 tetrahedra an be explained by depolarization effets by Ca (and some small geometri hanges in the tetrahedra), we expet that the Ca polyhedra play an essential part in the strutural phase transition near 825 K. RFRNCS CITD Bismayer, D. (199) Hard mode Raman spetrosopy and its appliation to ferroelasti and ferroeletri phase transitions. Phase Transitions, 27, Bismayer, D., Shmahl, W, Shmidt,., and Groat, L.A. (1992) Linear birefringene and X-ray diffration studies of the strutural phase transition in titanite, CaTiSiO,. Physis and Chemistry of Minerals, 19, Chen, S.K., Liu, H.S., and Wu, C.S. (1993) Quantitative analysis of sphene

6 ZHANG T AL.: STRUCTURAL PHAS TRANSITION IN TITANIT 35 (CaTiOSiO,) and wollastonite (CaSiO,) rystallizations in annealed erami frits by FT-IR absorption spetrosopy. Applied Spetrosopy, 47, Ghose, S., Ito, Y., and Hath, D.M. (1991) Paraeletri-antiferroeletri phase transition in titanite, CaTiSiO,: I. A high temperature X-ray diffration study of the order parameter and transition mehanism. Physis and Chemistry of Minerals, 17, Meyer, H.W., Zhang, M., Bismayer, U., Salje,.K.H., Shmidt, C, Kek, S., Morgenroth, W., and Bleser, T (1996) Phase transformation of natural titanite: An infrared, Raman spetrosopi, optial birefringene and X-ray diffration study. Phase Transitions, in press. Salje,.K.H. (1992) Hard mode spetrosopy: xperimental studies of strutural phase transitions. Phase Transitions, 37, (1993) Phase transitions in ferroelasti and oelasti rystals: Student edition, 229 p. Cambridge University Press, U.K. - (1994) Phase transitions and vibrational spetrosopy in feldspars. In I. Parson, d., Feldspars and their reations, p Kluwer, Dordreht, The Netherlands. Salje,.K.H., Shmidt, C, and Bismayer, U. (1993) Strutural phase transition in titanite, CaTiSiO,: A Raman spetrosopi study. Physis and Chemistry of Minerals, 19, Speer, J.A., and Gibbs, G.Y. (1976) The rystal struture of syntheti titanite, CaTiOSiO" and the domain textures of natural titanites. Amerian Mineralogist, 61, Taylor, M., and Brown, G.. (1976) High-temperature strutural study of the P2/a A21a phase transition in syntheti titanite, CaTiSiO,. Amerian Mineralogist, '" 61, Van Heurk, C, Van Tendeloo, G., Ghose, S., and Amelinkx, S. (1991) Paraeletri-antiferroeletri phase transition in titanite, CaTiSiO,: II. letron diffration and eletron mirosopi studies of transition dynamis. Physis and Chemistry of Minerals, 17, Zahariasen, W.H. (193) The rystal struture of titanite. Zeitshrift fur Kristallographie, 73, Zhang, M., Salje,.K.H., Bismayer, U., Unruh, H.G., Wruk, B., and Shmidt, C (1995) Phase transition(s) in titanite CaTiSiO,: An infrared spetrosopi, dieletri response and heat apaity study. Physis and Chemistry of Minerals, 22, MANUSCRIPT RCIVD JANUARY 16, 1996 MANUSCRIPT ACCPTD OCTOBR I, 1996