J O U R N A L O F C H E M I S T R Y Materials High temperature phase segregation of a new host for Er3+ upconversion: Cs3Tl2Cl9 Lukas Kamber, Philipp Egger, Bernhard Trusch, Rudolf Giovanoli and Ju�rg Hulliger* Department of Chemistry and Biochemistry, University of Berne, Freiestrasse 3, CH-3012 Berne, Switzerland Cs3Tl2Cl9 is one of the few air stable low-phonon host lattices of interest to Er3+ upconversion.Solution growth at ambient temperature demonstrated that water and a number of other molecular liquids do not yield Er3+ doped crystals. Optical and diVerential scanning calorimetry measurements revealed a phase segregation at 310 °C which is reversible in the presence of 1 atm Cl2. Growth from molten salt solutions or gas phase deposition techniques is hence restricted to a deposition temperature of less than 300 °C.Preliminary results show that molten ZnCl2 may be used as a flux to obtain lanthanide doped single crystals or epitaxial layers. the site symmetry for the dopant is 3m, a non-centrosymmetric Introduction site being a requirement for a high absorption cross section. High optical storage density devices and displaying appli- Single crystals of Cs3Tl2Cl9 have been obtained from aquecations demand compact and eYcient laser sources in the ous solution.7 However, lanthanide ions strongly coordinate visible spectral range.One implementation may be resonant to the oxygen of water molecules. It was therefore not possible conversion of infrared to visible light by solid state devices, to grow Er3+ doped Cs3Tl2Cl9 crystals from H2O solutions so-called upconversion (UC).1 For UC to take place, appro- near 300 K.The ultimate alternative is high temperature priate host lattices are doped with diVerent lanthanide ions, growth either from a flux or from the vapour. Both routes mainly Er, Tm, Pr or Ho. Recent investigations mainly used need knowledge on the thermal stability field of Cs3Tl2Cl9.fluoride and oxide crystals as well as heavy metal glasses as Solid state UC devices demonstrated so far are single crystals host materials.2 As the reduction of the phonon energies of the or waveguiding glass fibers grown at high temperatures.2 In host significantly slows down non-radiative multiphonon relax- particular, single crystalline waveguides allow the combination ation3,4 and thus increases UC eYciencies, there is considerable of the advantages of bulk crystals (narrow bands leading to interest in low-phonon UC materials.Because lowering phonon high gain coeYcients) with those of the fiber geometry (high energies can be achieved by introducing heavier ligands, chlor- pump intensities along the whole absorption length). High ide and bromide compounds have been investigated, showing quality waveguiding layers are most eVectively grown by liquid pronounced UC eYciencies.5,6 Due to the high sensitivity of phase epitaxy (LPE),8 pulsed laser deposition9 or high energy most of these materials to air and moisture, applications such ion implantation.10 A first fluoride LPE-grown UC waveguide as those mentioned before are chemically demanding.We have has been reported recently.11 A waveguide of Cs3Tl2Cl9 may found Cs3Tl2Cl9 to be one of the only low-phonon UC hosts be achieved by growing a lanthanide doped epitaxial layer on which is really stable in air and can be doped with, for example, a Cs3 Tl2Cl9 substrate. Here, too, knowledge of the thermal Er3+.6 Cs3Tl2Cl9 crystallizes in the rhombohedral space group behaviour of this potential host or substrate material is basic R39c featuring two face-sharing distorted octahedra (Fig. 1).7 to any further improvement. Tl3+ ions are situated in the centre of the octahedra, leading The thermal behaviour of powdered Cs3Tl2Cl9 samples in to short Tl3+–Tl3+ distances within a dimer (d#4 A° ) and open and closed systems has been investigated by Richter.12 facilitating energy transfer in the case of erbium.Furthermore, Following this analysis, Cs3Tl2Cl9 undergoes phase segregation at 301 °C, i.e. it decomposes into Cs2TlITlIIICl6+CsCl+Cl2 with an intermediate product identified as Cs11Tl6Cl29 (see ref. 12, p. 59). A weight loss of 6.2% has been found by thermogravimetry on unsealed samples. Compared to Cs3Tl2Cl9, a significantly simpler powder diVraction pattern was measured at temperatures above 301 °C.The Fm3m symmetry was attributed to Cs2TlITlIIICl6. Raman spectroscopy demonstrated a reversible phase segregation. In this work we present a thermal analysis relying on single crystals of Cs3Tl2Cl9 being kept under Cl2 (ca. 1 atm) in closed glass capillaries. Experimental conditions for the phase segregation study have been chosen as close as possible to parameters needed for high temperature single crystal and epitaxial growth of Cs3Tl2Cl9 to follow.Experimental Due to the toxicity of thallium compounds, care had to be Fig. 1 Dimeric structure of [Tl2Cl9]3- (ref. 7) taken. When handling powders or samples at elevated temp- J. Mater. Chem., 1998, 8(5), 1259–1262 1259eratures, they were kept in a special dry box (used for Tl The ampoules were glued to a thin glass plate with silver paste in order to improve the thermal contact.compounds only) or in sealed ampoules, respectively. Synthesis of Cs3Tl2Cl9 (ref. 13) Results and Discussion Wet TlCl3 (4 g; hygroscopic) and 5 g CsCl were dissolved While heating up to 330 °C DSC showed a strong endothermic separately in 10 ml distilled H2O, poured together and stirred.peak around 310 °C (10.6±0.65 kJ mol-1) accompanied by A white, finely spread precipitate was immediately obtained. several weaker endothermic signals in the range of 180–250 °C CsCl was added in excess to bring all TlCl3 to reaction. The (Fig. 3). Crystals tempered above 200 °C under N2 (48 h) did precipitate was recrystallised by addition of 70 ml distilled not show such spurious signals. In any case, there were no H2O, heating to 60 °C and slow cooling in a previously heated endothermic peaks around 100 °C, indicating that the crystals dewar. Transparent, needle shaped crystals were finally do not represent a hydrate or show water bubble inclusions.obtained which could be filtered oV and dried at 80 °C under Upon cooling an exothermic peak around 280 °C was observed.vacuum for several hours. Interesting thermal features were observed by orthoscopy: sealed crystals were positioned under crossed polars to yield Single crystal growth maximum extinction. Several bright spots indicated defects and strained areas [Fig. 4(a)]. On heating up to 180–250 °C Single crystals (Fig. 2) were grown using the DT method.14 An these areas dimmed and some of them came to complete excess of Cs3Tl2Cl9(s) and a saturated aqueous Cs3Tl2Cl9 extinction. Upon further heating, the entire crystal volume lit solution were filled into a specially designed tube. The tube up around 310 °C [Fig. 4(b) and (c)]. This phenomenon started was inclined at an angle of 30° to the horizontal.A DT of 1.5 °C was suYcient to induce constant convection, transporting saturated solution to the upper part of the tube where cooling led to supersaturation and subsequent nucleation. Nucleation could be promoted, i.e. controlled by use of a Peltier cooling element which was placed in the upper part of the tube.15 dT pulses allowed the supersaturation to be increased for a short time of 20–30 s.By X-ray diVraction we determined the growth direction to be the c-axis of the R39c system. The lattice constants are c=18.27 A ° and a=12.82 A ° .7 DiVerential scanning calorimetry DSC measurements were carried out using a Mettler Toledo DSC 25. The heating/cooling rates were 2 °C min-1. Crystals of ca. 10 mg were placed into sealed aluminium crucibles under N2.Fig. 3 DSC measurement of a Cs3Tl2Cl9 single crystal from room temperature to 330 °C Gandolfi X-ray photographs These were taken using an UB/71 OYcine Elettrotecnica Di Tenno instrument. A typical exposure time for single crystals was 4 h at 900W Cu-Ka1 radiation. Photographs at elevated temperatures were obtained by heating crystals sealed in Lindemann type capillaries (normal or quartzlass) under ca. 800 mbar of Cl2 by means of a heated air flow. The exposure time for rotated samples was typically 9 h at 900W Cu-Ka1. Additional photographs of single or polycrystalline samples were taken with the Gandolfi camera without rotating the sample, with sample exposure for 2 h. Orthoscopic investigation Orthoscopic investigations were performed using a Linkham THMS 600 heatable microscope table and a Leitz Orthoplan microscope with tenfold magnification.The samples were mounted in small Pyrex ampoules under N2 or Cl2 atmosphere. Fig. 4 Orthoscopy of the phase transition close to 310 °C (dashed lines Fig. 2 Single crystals of Cs3Tl2Cl9 grown from aqueous solution by indicate crystal edges). (a) T<310 °C; (b) start of the phase segregation around 310 °C; (c) end of the phase segregation.the DT method14 1260 J. Mater. Chem., 1998, 8(5), 1259–1262from crystal edges and defects, which did not come to complete present pattern could not be attributed to either Cs2TlITlIIICl6, Cs11Tl6Cl29 or CsCl, proposed by Richter (the products of extinction in the first phase of heating. From there, the illuminated sections spread over the entire crystal.No diVer- phase segregation of powder samples).12 The release of Cl2 at the phase change complicates the DSC ence was noticed whether working under N2 or Cl2 atmosphere. Upon cooling and about 30 °C below the transformation measurements on cooling, because Cl2 may react with the aluminium pan. However, at 280 °C an exothermic peak temperature (heating cycle), crystals showed some increase of local extinction, albeit not complete.supports reversibility of the transition under Cl2. Thermodynamical calculations16 applied to TlCl(s) and Optical inspection carried out in addition to orthoscopy monitored an increasing absorption (yellow colour) TlCl3(s) show the preferred formation of Tl+ at high temperatures, thus indicating the presence of at least one Tl+ containing above 200 °C.Around 300 °C, the crystal sharply turned yellow–brown. compound after the phase segregation. The sharp colour change to yellow around 310 °C may therefore be attributed Preliminary Gandolfi measurements were acquired without rotation of the crystal, showing a typical single crystalline to the formation of a mixed valence compound containing Tl3+ and Tl+.pattern at T<310 °C replaced by a powder pattern above 310 °C. DiVraction of rotated samples taken above the trans- Finally, strain as a source of defects mentioned in the paragraph on orthoscopy can be excluded, as relaxation of ition temperature produced photographs showing only a few lines (Table 1). Samples cooled to room temperature recovered strain would lead to exothermic peaks in the DSC.The observed relaxation temperatures varied with the crystals lines typical of Cs3Tl2Cl9. Taking into account all the observations mentioned above, examined. We assume that these areas act as nucleation centres for the phase segregation behaviour observed by optical means. the endothermic peak and the loss of birefringence around 310 °C can be attributed to a segregation reaction, transforming Compared to the transition temperature reported for powdered samples of Cs3Tl2Cl9,12 the single crystal transition single crystalline Cs3Tl2Cl9 into a composite of one or more new crystalline phases, at least one of them being yellow.On temperature is shifted by about 10 °C. This can be explained by the presence of fewer defects and a smaller surface.cooling, Cs3Tl2Cl9 is recovered in form of a polycrystalline, opaque material. However, this phase segregation was found to be reversible only if carried out in a closed system of either a small volume (under N2) or under Cl2 (ca. 1 atm). It may be Conclusions and preliminary results on solvent that Cl2 gas is released in the course of this reaction.Note systems that the recovery of the Cs3Tl2Cl9 phase required non-reactive vessels, such as Pyrex or quartz glass. In the case of normal Single crystals of Cs3Tl2Cl9 could be grown to optical quality and dimensions that allowed phase transition phenomena to Lindemann capillaries a completely diVerent and also unknown diVraction pattern was obtained. be tracked, indicating significant deviations from results gained on powder data.12 As the high temperature diVraction data revealed only a few lines, we conclude that predominantly one solid phase is Investigations on single crystals are indicative of phase segregation at 310 °C. At this temperature, an as yet unknown formed, other minor contributions being below the detection limit.The small number of lines in the high temperature new phase is formed.However, there is strong evidence that Cl2 and Tl+ are involved in the reaction. Within experimental diVraction pattern indicate a new phase of high symmetry. The Table 1 d-Values (A ° ) of the new and so far unknown high temperature phase, measured by a Gandolfi camera (further columns contain powder data of products reported by Richter;12 there is no agreement between the data of the first column with d-values reported by Richter) new high-temp. Cs3Tl2Cl9 (Ref. 18) Cs11Tl6Cl29 a-Cs2TlTlCl6 (Ref. 12) b-Cs2TlTlCl6 (Ref. 12) CsCl (Ref. 20) phase (300 K) (Ref. 19) (low-temp.)/(high-temp.) (300 K) 7.0562 (m) 6.4105 (m) 6.3551 (w) 4.4192 (m) 4.2322 (w) 4.120 (m) 4.2305 (m) 3.9893 (s) 3.9414 (w) 3.8951 (s) 3.9468 (s) 3.7944 (m) 3.8659 (m) 3.7024 (s) 3.7522 (m) 3.5308 (w) 3.5364 (w) 2.917 (s) 2.8934 (m) 2.8372 (s) 2.8158 (m) 2.7544 (s) 2.7925 (m) 2.7741 (m) 2.7914 (m) 2.67 (m) 2.7066 (m) 2.7331 (m) 2.59 (s) 2.5562 (w) 2.5695 (w) 2.5226 (m) 2.29 (m) 2.3564 (m) 2.380 (w) 2.24 (s) 2.2525 (w) 2.2181 (m) 2.2534 (w) 2.2793 (m) 2.2096 (s) 2.2092 (m) 2.2367 (m) 2.1375 (m) 2.1402 (w) 2.1489 (w) 2.1205 (m) 2.1045 (w) 1.9952 (m) 1.9465 (w) 1.9738 (m) 2.062 (w) 1.8972 (m) 1.8739 (m) 1.8510 (s) 1.8576 (m) 1.65 (w) 1.8442 (m) 1.844 (w) 1.59 (s) 1.7171 (w) 1.7145 (w) 1.7514 (w) 1.7655 (w) 1.683 (m) 1.6772 (w) 1.6643 (w) 1.7307 (w) 1.457 (w) 1.34 (w) 1.6469 (w) 1.6718 (w) 1.32 (s) 1.5358 (w) 1.5496 (m) 1.26 (w) 1.4190 (w) 1.5465 (w) 1.5412 (w) 1.4920 (w) 1.374 (w) (s), (m), (w) denote strong, medium and weak, respectively.J. Mater. Chem., 1998, 8(5), 1259–1262 1261error, the transition was reversible at ca. 1 atm Cl2 pressure. We thank Prof. T. Peters for the use of the Linkham microscope, Prof. T. Armbruster for the use of a Gandolfi camera, These important new results imply that bulk and epitaxial and Dr. C. Ba�rlocher for the use of the program TREOR.growth of Cs3Tl2Cl9 will have to be carried out at temperatures This work was supported by the Swiss National Science below T#300 °C and under Cl2 atmosphere. Foundation (project 20–43116.95) and the Priority Program The evident approach to bulk and epitaxial growth would ‘Optics II’ (project no. 232) of the Swiss Board of the Federal then be the use of a suitable low temperature solvent or high Institutes of Technology.temperature flux under Cl2 atmosphere. In the course of this work several solvents (DMSO, acetonitrile, liquid NH3, etc.) were tested. All of them converted the Cs3Tl2Cl9 single crystals References into a white mush, probably due to the soft Tl3+ ion binding 1 M. C. Brierley, J. F. Massicott, T. J. Whitley, C. A. Millar, soft Lewis bases.Recrystallisation from liquid NH3 yielded an R. Wyatt, S. T. Davey and D. Szebesta, BT T echnol. J., 1993, unknown powder pattern. Cs3Tl2Cl9 was insoluble in ethanol 11, 128. and methanol. Commonly used inorganic acids, HCl( l ) and 2 R. M. Macfarlane, J. Phys. IV (suppl.), 1994, 4, C4–289. some other solvents (Cl3PO) either yielded no significant 3 L. A. Riseberg and H.W. Moos, Phys. Rev., 1968, 174, 429. solubility or led to redox reactions. We therefore turned to an 4 J. M. F. van Dijk and M. F. H. Schuurmans, J. Chem. Phys., 1983, exploration of several relatively low melting fluxes: rhodanides 78, 5317. 5 M. P. Hehlen, K. Kra�mer, H. U. Gu� del, R. A. McFarlane and and nitrates were not suitable due to coordination to Tl3+. R. N. Schwartz, Phys.Rev. B, 1994, 49, 12475. SnCl2 systems were oxidised to SnCl4 under Cl2 here. 6 P. Egger, P. Rogin, T. Riedener, H. U. Gu� del, M. S. Wickleder and A mixture of Cs3Tl2Cl9 and ZnCl2 melting at 230 °C turned J. Hulliger, Adv. Mater., 1996, 8, 668. out to be a suitable system, showing suYcient solubility for 7 J. L. Hoard and L. Goldstein, J. Chem. Phys., 1935, 3, 199.both Cs3Tl2Cl9 and ErCl3. However, care has to be taken in 8 P. Rogin and J. Hulliger, J. Cryst. Growth, 1997, 179, 551. 9 D. S. Gill, A. A. Anderson, R. W. Eason, T. J. Warburton and order to avoid precipitation of ZnCl2·2CsCl.17 Further D. Shepherd, Appl. Phys. L ett., 1996, 69, 10. investigations on the use of the ZnCl2 flux in crystal growth 10 S. J. Field, D. C. Hanna, A. C.Large, D. B. Shepherd, experiments are in progress. A. C. Tropper, P. J. Chandler, P. D. Townsend and L. Zhang, An alternative approach would be physical vapour depos- Electron. L ett., 1991, 27, 2375. ition. In preliminary experiments, CsCl, ErCl3 and Cs3Tl2Cl9 11 P. Rogin and J. Hulliger, Opt. L ett., 1997, 22, 1701. 12 R. Richter, PhD thesis, Albert-Ludwigs-Universita�t, Freiburg, were evaporated separately in a nitrogen flow, carried over a 1993.substrate crystal of Cs3Tl2Cl9 held below 300 °C. Powder 13 J. H. Pratt, Z. Anorg. Chem., 1895, 9, 23. diVraction patterns of the materials deposited showed lines 14 J. Hulliger, Angew. Chem., Int. Ed. Engl., 1994, 33, 143. which were diVerent from those of Cs3Tl2Cl9, with the yellow 15 O. Ko� nig, P. Rechsteiner, B. Trusch, C. Andreoli and J. Hulliger, colour indicating the presence of Tl+. This could well be due J. Appl. Cryst., 1997, 30, 507. to Cl2 loss. In order to avoid reduction, the Cs3Tl2Cl9 source 16 I. Bahrin, T hermodynamical Data of Pure Substances, VCH Verlagsgesellschaft mbH, Weinheim, 1989. should therefore be kept under a Cl2 flow. Thermodynamical 17 B. F. Markov, I. D. Panchenko and T. G. Kostenko, Ukr. Khim. calculations using data given in ref. 16 and applied to TlCl3 Zh., 1956, 22, 290. and TlCl show that the reduction of TlCl3 to TlCl is exergonic 18 JCPDS database, card no. 44–716. at a sublimation temperature of about 400 °C in a N2 atmos- 19 G. Thiele and R. Richter, Z. Kristallogr., 1993, 207, 142. phere. In contrast, TlCl3 is thermodynamically favoured under 20 JCPDS database, card no. 5–607. a Cl2 atmosphere of about 5 atm to promote the formation of the Cs3Tl2Cl9 phase. Paper 7/09190G; Received 23rd December, 1997 1262 J. Mater. Chem., 1998, 8(5), 1259&ndas