J O U R N A L O F C H E M I S T R Y Materials Hydrothermal synthesis and characterisation of BaTiO3 fine powders: precursors, polymorphism and properties† Iain J. Clark,a Tomanari Takeuchi,b Noikazu Ohtoric and Derek C. Sinclaira aChemistry Department, University of Aberdeen, Meston Walk, Aberdeen, UK AB24 3UE bOsaka National Research Institute, 1–8-31, Midorigaoka, Ikeda, Osaka, Japan cGraduate School of Science and Technology, Niigatu University, Ikarashi 2-no cho, Niigata, Japan Received 19th June 1998, Accepted 22nd July 1998 The influence of two Ti-precursors, TiO2 (anatase) and H2TiO3 (b-titanic acid), on the purity and particle size of BaTiO3 powders prepared via hydrothermal synthesis is discussed.Amorphous H2TiO3 was found to be an excellent Ti-precursor material and oVers several advantages over crystalline anatase.Phase pure powders which have small particle sizes, ca. 40–80 nm and narrow particle size distributions can be prepared at 180 °C after 24 h using H2TiO3 as a precursor material. Although the initial reaction is very fast, ca. 90% yield after 8–10 h, extended reaction periods at 180 °C are required in order to drive the reaction to completion.Lowering the reaction temperature from 180 to 85 °C does produce powders with even smaller particle sizes, however, very long reaction periods are required, e.g.>72 h, to ensure complete reaction. Raman spectra of as-prepared and heat treated (1000 °C) powders with average particle sizes as small as ca. 20–40 nm indicate asymmetry within the TiO6 octahedra of the BaTiO3 lattice.These results contradict the widely cited ‘critical’ particle size theory for the stabilisation of the cubic polymorph, at least for particle sizes greater than ca. 20–40 nm. As-prepared powders contain many defects, primarily in the form of lattice OH- ions. Preliminary ac impedance spectroscopy data on samples heat treated to remove lattice hydoxyl ions demonstrate these materials to be modest proton conductors.size eVect but there is still no clear consensus as to its origin.‡ Introduction Two commonly cited models involve the generation of higher- BaTiO3 is one of the most widely used and studied ferroelectric than-usual stress in fine grained ceramics5,6 whereas a third materials in the electro-ceramics industry. It has various suggests a core-shell structure,7,8 whereby grains below a polymorphs, all of which are based on the perovskite struc- ‘critical’ size (<0.2 mm) have cubic symmetry, intermediate ture;1 however, the two most studied are the tetragonal, t, and grain sizes (ca. 1 mm) have a ferroelectric-tetragonal core with cubic, c, polymorphs.t-BaTiO3 forms between ca. 0 and 130 °C a cubic surface layer and large grains (5 mm) are essentially whereas c-BaTiO3 is stable above 130 °C.In the t-polymorph, tetragonal. It should be noted that there is not an accepted titanium ions are displaced from the centrosymmetric position ‘critical’ particle or grain size below which the cubic polymorph within TiO6 octahedra and give rise to spontaneous polaris- is stabilised, on the contrary, a wide range of values, ca. ation.Due to this asymmetry within the crystal structure, t- 25–190 nm have been reported.9–16 In addition, a variety of BaTiO3-based materials have high permittivities and are widely reasons have been proposed for the room-temperature stabilisemployed as dielectrics in ceramic capacitors.2 The transition ation of the cubic polymorph, including the presence of lattice from the polar (ferroelectric) t- to non-polar (paraelectric) c- ‘defects’ such as hydroxyl ions17 (associated with powders polymorph normally occurs at ca. 130 °C, the Curie tempera- formed via wet chemical methods), small deviations in Ba/Ti ture, Tc. Above Tc, the Ti ions occupy, on average, the stoichiometry7 and excess surface energy associated with centrosymmetric position within TiO6 octahedra and there is ultrafine particles.18,19 Clearly, much remains to be done in no net polarisation within the solid.order to establish the factors which influence and control the A typical room temperature permittivity value, e25, for crystal symmetry and electrical properties of sub-micron ceramic t-BaTiO3 with an average grain size of ca. 10 mm is BaTiO3 powders and ceramics. 1000–2000; however, this rises sharply to a maximum value, Despite these fundamental problems, BaTiO3 will continue emax, of ca. 10 000 at Tc. In such ‘large’ grained ceramics, a to be used in the manufacture of thermistors,20 multilayer variety of ferroelectric domain structures form to relieve capacitors,21 electro-optic devices22 and DRAM (dynamic internal stress associated with the cubic to tetragonal phase random access memories)23 into the next century.Improved transformation. Although a detailed discussion of domain performance and miniaturisation of BaTiO3-based devices, structures in BaTiO3 powders and ceramics is outwith the either in the form of thin ceramic layers, ca. 20mm, for scope of this paper, it is important to note that ferroelectric multilayer capacitors or as thin films for integrated electronic domain structures and dipole–dipole interactions control the circuits remains a priority.High permittivities and miniaturispermittivity characteristics of BaTiO3 and that these depend ation can be achieved by controlling the ceramic microstructure on particle or grain size. For example, Arlt et al.3 have shown which, in turn, depends on the homogeneity, composition, that e25 for BaTiO3 ceramics can be optimised to a value surface area and particle size of the starting BaTiO3 powder.approaching 5000 by controlling the grain size to be ca. In addition, it is well known that BaTiO3 undergoes exagger- 0.7–1.0 mm. ated grain growth during sintering at elevated temperatures Several models have been proposed to explain this grain and long sintering periods.This can produce ceramics †Basis of the presentation given at Materials Chemistry Discussion ‡During the preparation of this manuscript, Frey et al.4 published a paper which appears to have clarified many of the problems associated No. 1, 24–26 September 1998, ICMCB, University of Bordeaux, France. with the origin of the so-called grain size eVect in BaTiO3.J. Mater. Chem., 1999, 9, 83–91 83containing very large grain sizes, e.g. >50 mm or duplex and the surrounding BaTiO3. For this model, complete reaction may be diYcult to achieve due to increasing encapsulation microstructures of low density consisting of both large (>10 mm) and small (<1 mm) grains.For these reasons, many of TiO2 via product formation. Homogeneous and heterogeneous dissolution–precipitation synthetic methods and sintering profiles24–28 have been investigated in an attempt to produce sub-micron, deagglomerated models have also been proposed. In the homogeneous model, a low concentration of TiO2 dissolves in the form of soluble fine powders of BaTiO3 that can be sintered into dense, finegrained ceramics.hydroxytitanium complexes which then react with Ba2+ ions in solution to precipitate BaTiO3. In the heterogeneous model, Traditionally, BaTiO3 has been produced via the mixed oxide route. This method involves repeated calcination and BaTiO3 nuclei form on the surfaces of the dissolving TiO2 particles and, as for the in situ transformation mechanism, the regrinding of BaCO3 and TiO2 powders above 1100 °C.The reaction mechanism in air has been proposed to take place in reaction becomes limited by the increasing isolation of the reactants via BaTiO3 product layers on the surface of TiO2 at least three stages29–32 and relies on the diVusion of Ba2+ ions into TiO2. Firstly, BaCO3 reacts with the outer surface particles.Eckert et al. reviewed these models41 in more detail and regions of TiO2 to form a surface layer of BaTiO3 on individual TiO2 grains. Further diVusion of Ba2+ ions into TiO2 necessi- undertook a kinetic study of hydrothermal reactions involving barium hydroxide octahydrate and anatase precursors. Two tates the formation of Ba2TiO4 between unreacted BaCO3 and the previously formed BaTiO3.After prolonged sintering reaction regimes were clearly identified; during the early stages the reaction is controlled by a dissolution–precipitation pro- periods, the intermediate Ba-rich phase Ba2TiO4 reacts with the remaining TiO2 in the core-regions of the TiO2 grains to cess, whereas, for the second regime at longer reaction times the results were inconclusive.The authors suggest two plausible form BaTiO3. Although the nominal starting Ba/Ti ratio is 151, long reaction times are required to form homogeneous but diVerent explanations for their kinetic data, one model is based on heterogeneous dissolution–precipitation followed by powders free from secondary phases. For such a complex and slow reaction mechanism, it is desirable to have high surface in situ transformation, whereas, the second model suggests that dissolution–precipitation is the controlling mechanism, area TiO2 powders of fine particle size and narrow size distribution in order to control the morphology and grain size with nucleation and growth of BaTiO3 controlling the first regime and the dissolution rate of TiO2 controlling the of the resultant BaTiO3 powder.Nevertheless, the mixed oxide route tends to produce coarse, agglomerated powders that second regime. In the present paper we discuss some of our recent work on require high sintering temperatures to form dense ceramics. In order to produce powders that are suitable for sintering the hydrothermal synthesis of BaTiO3 powders and highlight some advantages of using amorphous H2TiO3 as opposed to into micron-grain-sized ceramics for MLC applications there has been much interest in developing wet and novel chemical crystalline anatase as a precursor material.In particular, H2TiO3 promotes faster and more-complete reactions and also routes. As these reactions take place in the liquid, as opposed to the solid state, more intimate mixing of the cations is produces powders with much smaller particle sizes.We discuss the influence of temperature on the rate of reaction and on achieved and consequently, much shorter diVusion pathways are created. Deagglomerated, sub-micron powders of BaTiO3 the particle size, distribution and water content of the BaTiO3 powders. The importance of using Raman spectroscopy to can then be obtained at low temperatures and short reaction periods, for example via hydrothermal synthesis at 240 °C detect the presence of trace amounts of unreacted Ti-containing precursors and to yield information on the polymorphism of for 24 h.33 Other common methods include, oxalate (Clabauch),34 citrate (Pechini),35 catecholate (Milne)36 and sub-micron powders is discussed.Finally, we present preliminary results on powders and ceramics heat treated above sol–gel.37 Although the reaction mechanisms and advantages and disadvantages of many of these routes have been dis- 800 °C that support the idea that hydrothermal BaTiO3 powders contain many defects which influence their physical cussed,38 for the present purposes it is worth discussing these issues for hydrothermal synthesis. properties.BaTiO3 powders have been prepared via hydrothermal processing since the 1940’s and this method is commonly employed Experimental to produce commercial powders. Although it is well established that BaTiO3 is the only thermodynamically stable binary Ba(OH)2·8H2O (98+%, Aldrich) and anatase (99.9%, Aldrich) or b-H2TiO3 (+99%,Mitsuwa Chemicals) were mixed compound produced under all conditions,39 i.e.pH, Ba/Ti ratio, reaction temperature, etc. products can contain unre- with a Ba5Ti ratio of 1.0551 in a 23 ml Teflon-lined pressure vessel (Model 4749, Parr Instruments) together with 10 ml of acted TiO2 and BaCO3. Normally, an excess of Ba2+ in the starting solution is employed (Ba/Ti=1.05–1.1051) in an de-ionised water. The vessel was sealed, shaken and placed in an oven at either 85, 120 or 180 °C for periods ranging from attempt to remove any unreacted TiO2 and therefore drive the reaction to completion.Any excess Ba, in the form of BaCO3 1 to 72 h. After cooling, the contents of the bomb were diluted in 30 ml of 0.1M formic acid in an attempt to dissolve any can then be removed via acid washing of the powders.Although this procedure is eVective in controlling the purity BaCO3 formed by the addition of excess Ba2+ to the starting solution. The mixture was vacuum filtered using a Buchner of the powders, relatively little is known about the influence of acid washing on powder stoichiometry as Ba2+ may be funnel, the residue thoroughly washed with distilled water and dried overnight in air at 120 °C.The percentage yield of leached from the surfaces of the particles. Although controlling the precise stoichiometry of hydrothermal BaTiO3 powders BaTiO3 was estimated from the weight of the dried powders. Phase purity and polymorphism of the BaTiO3 powders can be problematic, the low cost and easy handling of the reagents, and the fast reaction rate at low temperatures ensures were studied by X-ray diVraction (XRD) and Raman spectroscopy.XRD was carried out on a Stoe STADI P automated that deagglomerated powders consisting of small particles of narrow size distribution are readily obtained. powder diVractometer employing monochromatic Cu-Ka1 radiation and a linear position sensitive detector over the 2h Several mechanisms have been proposed for the hydrothermal synthesis of BaTiO3.Hertl40 has suggested an in situ range 20 to 80 °. Cubic and tetragonal polymorphs were distinguished by the splitting of the (200) peak at 2h#45 °. transformation whereby dissolved Ba2+ ions react with undissolved TiO2 to produce a continuous layer of BaTiO3 on the Unit cell parameters were obtained by refinement of XRD data. Raman spectroscopy was carried out on a Jobin Yvon surface of TiO2 particles. The reaction is then controlled either by diVusion of Ba2+ through the product BaTiO3 layer or T64000 using a 100 mW Ar laser with a wavelength of 514.5 nm.A 300 mm slit and integration time of 5 s were used further reaction between unreacted TiO2 within grain interiors 84 J. Mater. Chem., 1999, 9, 83–91giving a resolution of 8.4 cm-1.Spectra were measured over An important aspect of this work is to obtain homogeneous the range 58–1103 cm-1. Impurity phases such as BaCO3 and particles of controlled stoichiometry which can be sintered TiO2 (anatase) were detected from characteristic peaks at 1060 into dense ceramics of uniform microstructure. The Ba5Ti and 150 cm-1, respectively. The presence of a peak in the ratio of the particles is an important parameter but is not spectra of BaTiO3 at 305 cm-1 indicates asymmetry within readily obtained via XRD or Raman spectroscopy.We choose, TiO6 octahedra42 and was used to distinguish between tetra- therefore, to study the phase assemblage and Ba5Ti ratio of gonal and cubic polymorphs. the grains in sintered ceramics using EPMA and from this, Thermogravimetric analysis (TGA) was carried out using a deduce information about the Ba5Ti ratio of the powders.Stanton Redcroft TG-DTA simultaneous thermal analyser This is an indirect method for obtaining information on the (model STA 1000/1500) to determine the water content of as- stoichiometry of the powders, but since neither barium nor prepared powders.Samples were accurately weighed to five titanium are particularly volatile at the sintering temperature, decimal places in a platinum crucible and weight losses we perceive this method to be valid. recorded from 25–1000 °C against an alumina reference. All experiments were carried out in air at a heating rate of 10 °Cmin-1 with an amplifier setting of 50 mV. Results and discussion Powder morphology was determined via transmission electron microscopy (TEM) using a JEOL electron microscope The results are subdivided into three sections.The first section (model 2000EX TEMSCAN) operating at 200 kV. Particle discusses the influence of two Ti-precursors, TiO2 (anatase) size and distribution were obtained by measuring the cross and H2TiO3 (b-titanic acid), on the purity and particle size of diagonals of approximately 130 particles from scanned nega- the BaTiO3 powders.The second section assesses the role of tives of TEM images. Energy dispersive X-ray analysis (EDX) reaction temperature on the rate of reaction, purity, particle was used in an attempt to determine the Ba5Ti ratio of the size and water content of BaTiO3 powders prepared from BaTiO3 particles and also to detect the presence of any Ba(OH)2·8H2O and H2TiO3.The final section discusses preunreacted TiO2. liminary electrical property measurements on powders heat Electron probe micro-analysis (EPMA) was employed to treated above 800 °C. determine the homogeneity of sintered pellets fabricated from the various powders. Pellets were sintered at ca. 1350 °C and I Titanium precursors a Cameca SX51 electron microprobe employed using an accelerating voltage of 20 kV and a beam current of 40 nA. One of the major problems associated with the hydrothermal Ba-La and Ti-Ka lines were measured using benitoite, synthesis of BaTiO3 is the low solubility of TiO2 in the highly BaTiSi3O9, as a standard. Prior to analysis, small pieces of alkaline environments, ca.pH>12, required for synthesis. pellet were mounted in resin blocks, surface polished and Although it is well documented that diVerent Ti-precursors carbon coated. and reaction temperatures influence the rate of reaction there Electrical measurements were carried out using a Solartron have been few comprehensive studies on the influence of these frequency response analyser (model 1250 or 1260) combined parameters on the purity and particle size of hydrothermal with a Solartron dielectric interface (model 1296).A frequency powders. Kutty et al. suggested that amorphous Ti-precursors range of 10-3–106 Hz was employed with an applied voltage react fastest,17 then anatase with rutile giving the slowest of 100 mV. Cold pressed pellets of various powders were reaction.Various workers reported an increase in average sintered between 800 and 1350 °C and organopaste and gold particle size and a change (via XRD) from cubic to tetragonal foil electrodes applied to the surfaces of the pellets at 800 °C, symmetry with increasing reaction temperature.9–16We decided prior to mounting in a conductivity jig. Measurements were to reinvestigate some of these findings by comparing the recorded over the temperature range 25–500 °C.reaction rates, powder purity and particle size of hydrothermal Before discussing the results it is important to appreciate BaTiO3 powders prepared at various temperatures and reac- the need for a wide variety of techniques to determine the tion periods using two Ti-precursors: amorphous H2TiO3 phase purity and crystal structure of fine-grain BaTiO3 pow- (BET using N2, surface area 193 m2 g-1) and crystalline ders.XRD is an excellent technique for obtaining information anatase (BET, surface area 8.9 m2 g-1). on the average or long range crystal structure of a material; In general, for a set reaction temperature and period, the however, peak broadening is significant for diVraction from reaction rate was faster and the average BaTiO3 particle size sub-micron sized particles and XRD cannot reveal the very smaller for reactions using the amorphous H2TiO3.Raman subtle unit cell distortions which occur in BaTiO3 for microdospectra of powders prepared under the same conditions for mains of dimensions of ca. 100 A° . In addition, routine XRD the two precursors are shown in Fig. 1. Spectra of anatase and is relatively insensitive to small quantities of impurity phases, the tetragonal polymorph of BaTiO3 prepared via the tra- especially for secondary phases which are amorphous/poorlyditional mixed oxide route are included for comparison. The crystalline and/or have weak X-ray scattering power. In our Raman spectrum of H2TiO3 is not shown but is similar to experience, XRD is sensitive to small quantities of BaCO3 in that of anatase and is also dominated by a large peak at ca.hydrothermally processed BaTiO3 but is much less sensitive in 150 cm-1. There are two noticeable features in the Raman the detection of unreacted Ti-containing precursors, either spectra of the products which are not readily discernible from crystalline anatase or amorphous H2TiO3.the XRD patterns (not shown) and which provide important Uncertainties over crystal symmetry can largely be resolved information. using alternative techniques such as Raman spectroscopy to First, the powder products from the H2TiO3 reaction appear probe short range order or local symmetry. There are reliable to be phase pure BaTiO3, whereas unreacted TiO2, as shown Raman spectra in the literature42 for various BaTiO3 polyby the presence of the peak at ca. 150 cm-1, is clearly present morphs; however, it must be stated that the diVerences in in the products of the anatase reaction. Small amounts of spectra are rather subtle, especially for the orthorhombic, unreacted anatase or amorphous H2TiO3 are diYcult to detect tetragonal and cubic (at temperatures slightly greater than Tc) via routine XRD but are readily detected (by the peak at ca.polymorphs. In contrast to XRD, Raman spectroscopy is 150 cm-1) in the Raman spectra. This result demonstrates the sensitive to the presence of both BaCO3 and unreacted Tiimportance of using Raman spectroscopy to monitor the phase containing precursors.For these reasons we choose to present purity of BaTiO3 powders prepared via hydrothermal syn- Raman spectra, as opposed to XRD patterns to illustrate the thesis. We should stress that single phase powders can, in fact, phase purity and crystal symmetry of as-prepared BaTiO3 powders. be prepared from crystalline anatase at 180 °C, however the J. Mater. Chem., 1999, 9, 83–91 85Fig. 2 TEM micrographs of BaTiO3 powders prepared via hydrothermal synthesis at 180 °C for 24 h using anatase (a) and H2TiO3 (b), as Ti-precursors. Fig. 1 Raman spectra of BaTiO3 prepared via a mixed oxide route (top) and via hydrothermal synthesis at 180 °C for 24 h using H2TiO3 and anatase as Ti-precursors. The bottom spectrum is of crystalline II Reaction temperature anatase.Raman spectra of powder products from two reaction temperatures, 85 and 180 °C after various reaction periods are shown in Fig. 3(a) and (b), respectively. With increasing time, both times required are far in excess of those required using sets of spectra illustrate a decrease in peak intensity at amorphous H2TiO3, typically in excess of 72 h. 150 cm-1 associated with the titanium precursor as BaTiO3 is Second, the presence of a peak at 305 cm-1 in both spectra formed.In fact, the spectrum after reaction at 180 °C for 24 h indicates asymmetry within the TiO6 octahedra of BaTiO3 and shows no evidence of this peak, suggesting that the reaction demonstrates clearly, on a local scale, that as-prepared powders has gone to completion. A small peak at 150 cm-1 is still do not have cubic symmetry.In contrast, XRD patterns apparent in the corresponding spectrum for powders prepared showed appreciable peak broadening, presumably associated at 85 °C. A higher reaction temperature of ca. 180 °C, is clearly with the small particle size and poor crystallinity of the required to ensure a more complete reaction for short reaction powders, however, there was little evidence of peak splitting periods, e.g. 24 h. The high surface area (193 m2 g-1) and the and the XRD data ‘appeared’ consistent with that for the fact that hydroxylation of many Ti–O–Ti bridging bonds has cubic polymorph, as reported previously.9–16 The higher sensi- already been achieved in H2TiO3, ensure rapid formation of tivity of Raman spectroscopy to probe the local rather than BaTiO3, even at 85 °C.It should be noted that many spectra long range structure clearly demonstrates that sub-micron contained a small peak at 1060 cm-1 (not shown) associated powders prepared via hydrothermal processing are tetragonal with a small amount of BaCO3 which was not completely rather than cubic. removed by the mild acid wash. The particle size of powders prepared from H2TiO3 were, The weight of the dried powders was used to calculate the on average, 3–5 times smaller than those prepared from TiO2 percentage yield of BaTiO3 and is shown in Fig. 4. We stress and, to our knowledge, are amongst the smallest reported for that these values represent upper estimates of the yield, especi- BaTiO3 powders prepared hydrothermally. Representative ally for short reaction periods where powders contain unremicrographs of powders prepared at 180 °C for 24 h from acted precursor and are also heavily hydrated.Nevertheless, H2TiO3 and TiO2 are shown in Fig. 2 and demonstrate the these estimates show that reaction is rapid and yields in excess much smaller particle size, ca. 80 nm compared with ca. of 90% are obtained within 8 h of reaction.From the Raman 250 nm. The faster rate of reaction and smaller particle size spectra in Fig. 3, extended periods, e.g. 24 h at elevated of powders prepared from amorphous H2TiO3 demonstrate temperatures (180 °C) are required in order to drive the two clear advantages of using this precursor instead of crystal- reaction close to completion, even for a reactive precursor line anatase.The remaining sections of the paper concentrate such as H2TiO3. on powders and ceramics prepared using H2TiO3 as the Ti- As-prepared powders are hydrated and lose water both adsorbed and structural on heating to ca. 1000 °C. In general, precursor. 86 J. Mater. Chem., 1999, 9, 83–91Fig. 3 Raman spectra of powder products from two reaction temperatures, 85 (a) and 180 °C (b), after various reaction periods.dramatic decrease in reaction rate after ca. 8–10 h, as shown by the percentage yield of BaTiO3. Nucleation of BaTiO3 crystals presumably takes place via a structural rearrangement of the amorphous H2TiO3 associated with the incorporation of Ba2+ or via direct reaction in solution of dissolved hydroxytitanium complexes with Ba2+ ions. Either way, the reaction involves dehydration, for example, for the rearrangement mechanism (1).H2TiO3+Ba2++2OH-�BaTiO3+H2O (1) Dehydration is slow in superheated fluids and leads to partial retention of H2O and OH- in BaTiO3, especially for powders prepared at low temperatures. The variation in particle size as a function of temperature and time are shown in Fig. 5.All powders have narrow particle size distribution ranges and for each temperature, the mean particle size approximately doubles over a period of ca. 20 h, Fig. 4 Percentage yield (weight%) of powder products from reactions e.g. 23±5 to 43±12 nm, between 2 and 24 h at 85 °C and at 85 and 180 °C and their associated water content (mol%) as a from 53±14 to 80±15 nm at 180 °C. Particle size values after function of reaction period.reaction at 120 °C were intermediate between those at 85 and 180 °C. On comparing the particle size values with the variations in product yield, it appears that after an initial ‘burst’ weight loss commenced at ca. 100 °C and was complete at ca. 600 °C. The powders prepared at 85 °C clearly contained of nucleation, growth of BaTiO3 particles is a rather slow process. substantially more water, especially for short reaction periods where up to 4 weight% loss was recorded.The water content Micrographs of powders obtained at 85 and 180 °C for 2 h are shown in Fig. 6. At 85 °C, they have poor crystallinity and of powders for each reaction temperature remained reasonably constant after ca. 8 h, Fig. 4. This coincides with the ill-defined morphology, Fig. 6(a), whereas at 180 °C they are J. Mater. Chem., 1999, 9, 83–91 87Fig. 5 Particle size histograms for powders prepared at 85 °C after 2 (a) and 24 h (b) and at 180 °C after 2 (c) and 24 h (d). well-defined, regular particles, Fig. 6(b). The powders prepared at 85 °C have a higher water content and presumably, a higher concentration of lattice defects, such as OH- ions.Despite the poor crystallinity of such powders, the presence of the peak at 305 cm-1 in all Raman spectra, Fig. 3(a), demonstrates that these small particles, ca. 20 nm, are tetragonal rather than cubic. This result suggests that the widely-cited ‘critical’ particle/grain size model is incorrect and that distortion of Ti–O bonds within the BaTiO3 lattice is possible, even for particles as small as ca. 20 nm and which contain substantial amounts of lattice defects, such as OH- ions. Although our results clarify some of the existing problems in the literature we are unable, at this stage, to identify the reaction mechanism(s) involved in hydrothermal synthesis of BaTiO3. In the final section we discuss some of the properties of heat treated, hydrothermal BaTiO3 powders.III Heat treated powders Hydroxyl ions play an important role in the synthesis of BaTiO3 via hydrothermal processing, as high pH environments, >12, are required to obtain single phase materials. Infrared spectroscopy17,43,44 and TGA45 have both been employed to demonstrate that as-prepared hydrothermal powders contain weakly-bound water molecules adsorbed onto particle surfaces and more strongly bonded structural water in the form of lattice OH- ions.In general, only total water contents are reported as deconvolution into the two distinct types is diYcult. In order to maintain electro-neutrality, it is generally accepted that barium vacancies (VBa) are created on the surfaces of individual particles46 to compensate for the incorporation of lattice OH- ions on O2- sites (OHOV), according to eqn.(2). 2[OHOV]=VBa (2) On heating above ca. 300 °C, lattice OH- ions are removed as follows, Fig. 6 TEM micrographs of powders prepared at 85 (a) and 180 °C (b), after a reaction period of 2 h. 2OHOV�Oo x+VOVV+H2O (g) (3) 88 J. Mater. Chem., 1999, 9, 83–91resulting in powders which contain significant concentrations of VBa and VOVV.High concentrations of lattice hydroxyl ions in as-prepared hydrothermal powders therefore influence the stoichiometry (Ba/Ti ratio), defect chemistry and grain growth of sintered BaTiO3 ceramics. Elimination of water in our powders was complete by ca. 600 °C, but there were no significant changes in the Raman spectra of the dehydrated powders until samples were heated >1000 °C.These subtle changes in spectra will be discussed elsewhere,47 however, the peak at 305 cm-1 was present in all processed powders, indicating the distorted crystal symmetry of the particles, irrespective of the presence of adsorbed water or lattice hydroxyl ions. This result contradicts the suggestion that lattice hydroxyl ions play a crucial role in controlling the crystal symmetry of hydrothermal BaTiO3 powders17,44,45 and is in agreement with the work of Frey and Payne who came to the same conclusion for BaTiO3 powders prepared via a Fig. 8 Lattice parameters (&) and c/a ratio (#) of BaTiO3 powders sol–gel route.48 prepared at 85 °C for 72 h as a function of processing temperature. A selection of XRD patterns over a limited 2h range for powders prepared at 85 °C and heated up to 1100 °C are shown As mentioned previously, cation and anion defects influence in Fig. 7. At 800 °C, the broad peak at ca. 45.2 ° (in as-made the stoichiometry, grain growth and electrical properties of hydrated powders) sharpens but clear splitting is not apparent BaTiO3 powders. The diVusion coeYcients for oxygen and until heating at ca. 1100 °C. In general, XRD reflections barium vacancies in BaTiO3 become appreciable above 800 become sharper and more intense after heating at higher and 1100 °C, respectively. Thus, on heating hydrothermal temperatures. XRD patterns for powders heated above 1100 °C powders at elevated temperatures, any barium deficiency, were indexed on a tetragonal unit cell, whereas, all others were either as compensating cation defects for lattice hydroxyl ions indexed on a cubic unit cell.Lattice parameters and c/a ratios or as a consequence of acid washing, or excess titanium, in are shown in Fig. 8 and demonstrate that powders heated at the form of unreacted Ti-containing precursor, should result temperatures &100 °C have c/a ratios of ca. 1.01, which is in in the formation of Ti-rich phases.EPMA and SEM on good agreement with those reported in the literature.7 ceramic pellets sintered at 1350 °C from the powders produced at 85 and 180 °C after 24 h, both demonstrated the existence of Ti-rich secondary phases. For powders produced at 85 °C, where unreacted TiO2 was detected via Raman spectroscopy, Fig. 3(a), secondary phases were clearly visible as dark intergranular regions in back scattered electron images (BSE), Fig. 9(a).EPMA results identified the presence of both BaTi2O5 and Ba6Ti17O40 in inter-granular regions but their volume fractions were too small to be detected via XRD. For powders produced at 180 °C, where the products appeared phase-pure by Raman spectroscopy, sintered pellets contained only a very small Fig. 7 XRD diVractograms over the 2h range 44–46 ° of BaTiO3 Fig. 9 Back scattered electron images of ceramic pellets sintered at prepared at 85 °C (top) and after heat treatment at 800, 1000 and 1100 °C. 1350 °C from powders produced at 85 (a) and 180 °C (b). J. Mater. Chem., 1999, 9, 83–91 89volume fraction of Ba6Ti17O40, as shown by the dark, isolated regions in the BSE image, Fig. 9(b).The Ti-rich secondary phases in powders prepared at 85 °C are clearly associated with unreacted Ti-precursor, which reacts with surrounding BaTiO3 particles at elevated temperatures to form Ti-rich binary phases. It is unclear, however, whether the secondary phase produced from powders prepared at 180 °C is associated with unreacted Ti-precursor which has not been detected via Raman spectroscopy (or analytical TEM) or is associated with a compositional Ba/Ti gradient within individual particles of the as-prepared powders.Direct evidence of VBa on the surfaces of individual particles, either from incorporation of lattice hydroxyl ions during hydrothermal synthesis or via acid/aqueous media wet milling of powders has been limited.Abicht et al.49 used HREM and EELS to demonstrate that the outer surfaces of individual BaTiO3 particles which have been wet milled in aqueous media Fig. 11 Log resistivity versus 1000 K/T for pellets of powder prepared show evidence of Ba2+ leaching. They propose that a Ba/Ti at 85 °C and sintered at 1000 °C (filled squares) and 1350 °C (open concentration gradient exists within individual particles which symbols).Calculated activation energies are in eV. consists of a 3–5 nm thick TiOx-rich outer layer followed by a intermediate layer, ca. 10 nm thick, with a molar Ba/Ti ratio omena at the electrodes and supports the idea that conduction increasing from 0 to 1. These core-shell structures aVect the is mainly by means of ions at these temperatures. sintering of BaTiO3 powders due to the reactivity of the Ti- Bulk resistivity values were extracted from the low frequency rich, outer layers.Our observation of small quantities of Ti- intercept of the semi-circular arc with the Z¾ axis on the Z* rich intergranular regions in sintered pellets of hydrothermal plots. Initially, the bulk resistivity increases on heating, as powders produced at 180 °C is consistent with this model but shown by the increase in diameter of the bulk semi-circular in-depth studies using HREM and EELS are required to arc in the Z* plots, Fig. 10, and in the log resistance, Arrheniusestablish the presence or absence of any Ba/Ti compositional type plot shown in Fig. 11. The bulk resistivity rises by several gradient within individual particles of BaTiO3 (produced via orders of magnitude on heating to ca. 150 °C, then remains hydrothermal synthesis). temperature independent before decreasing in accordance with Heat treatment of hydrothermal powders at 1000 °C is the Arrhenius law at temperatures above ca. 350 °C. In this suYcient to remove any adsorbed or structural water and create high temperature region, there was no evidence of the low oxygen vacancies within the lattice, eqn. (3).It is insuYcient, frequency, inclined spike in Z*; instead, the plots consisted of however, to cause substantial migration of barium vacancies or a single, bulk semi-circular arc, indicating that conduction was densification and grain growth via liquid phase sintering involv- predominantly electronic. The resistivity behaviour was reversing Ti-rich outer particle surfaces and/or any excess unreacted ible on thermal cycling.For comparison, bulk and grain Ti-precursor material. Consequently, ceramic pellets formed at boundary resistivities extracted from high and low frequency 1000 °C are poorly sintered, have low density, ca. 65% of the semicircular arcs in Z* plots for a pellet sintered at 1350 °C theoretical density, and consist of small grains with very poor are also shown in Fig. 11. These values are consistent with inter-granular contact. Despite this the pellets exhibit exception- those reported in the literature for dense BaTiO3 ceramics.50 ally low room temperature resistivities, ca. 10–50MV. It should be noted that the room temperature resistivity of Complex impedance plane, Z*, plots at 32 and 44 °C for a these ceramics is in excess of 1 TV and that no low frequency, pellet sintered at 1000 °C are shown in Fig. 10.The plots consist inclined spike was observed in any Z* plots. of a high frequency semicircular arc with an associated capaci- The exceptionally low bulk resistivity value of 20 MV at tance of 32 pF and a low frequency-spike with an associated 25 °C, the presence of a low frequency spike in Z* plots below capacitance in the order of 1 mF.The capacitance value for the ca. 150 °C and the fact that resistivity initially rises with bulk or intra-granular component is consistent with a poorly temperature, all indicate that BaTiO3 prepared via hydrothermal sintered BaTiO3 ceramic and the low frequency, inclined spike synthesis at 85 °C and sintered at 1000 °C is a modest proton is attributable to ionic polarisation and diVusion-limited phen- conductor, especially at temperatures close to room temperature.Presumably water vapour is adsorbed from the ambient on cooling (from the sintering temperature of 1000 °C) and partially reverses the reaction given in eqn. (3), thus, converting doubly ionised oxygen vacancies into lattice hydroxyl ions.Subsequent reheating removes water rather easily and the bulk resistivity increases rapidly; however, temperatures in excess of 350 °C are required before the predominant charge carriers are electronic. More detailed studies of this unusual bulk resistivity behaviour will be reported elsewhere,51 however, such materials are clearly poor dielectrics. To our knowledge, this is the first time that proton conduction has been demonstrated in oxygen-deficient BaTiO3 materials.Many other oxygen-deficient perovskites such as doped, alkaline earth zirconates52 and cerates53 are well-known proton conductors. Several diVerent mechanisms of incorporation of protonic defects in such oxides have been suggested,54 involving interstitial protons and lattice hydroxyl ions, H2O (g)+VOVV=2HiV+OO x (4) Fig. 10 Complex impedance plane plots at 32 and 44 °C for a pellet H2O (g)+OO x+VOVV=2OHOV (5) of powder prepared at 85 °C and sintered at 1000 °C. Selected frequencies (in Hz) are shown for filled symbols. however, the detailed mechanism remains unclear. 90 J. Mater. Chem., 1999, 9, 83–9111 M. Kataoka, K.Suda, N. Ishizawa, F. Marumo, Y. Shimizugawa Conclusions and K. Ohsumi, J. Ceram. Soc. Jpn., 1994, 102, 213. 12 K. Uchino, E. Sadanaga and T. Hirose, J. Am. Ceram. Soc., 1989, Amorphous H2TiO3 is an excellent Ti-precursor to use in the 72, 1555. hydrothermal synthesis of BaTiO3. Phase pure powders with 13 F. J. Gotor, C. Real, M. J. Dianez and J. M. Criado, J. Solid State small particle sizes, ca. 40–80 nm and narrow particle size Chem., 1996, 123, 301. distributions can be prepared at 180 °C after 24 h. Although 14 S. Schlag and H. F. Eicke, Solid State Commun., 1994, 91, 883. initial reaction is very fast, ca. 90% yield after 8–10 h, extended 15 B. D. Begg, E. R. Vance and J. Nowotny, J. Am. Ceram. Soc., 1994, 77, 3186. heating at 180 °C is required to drive reactions to completion. 16 H. I. Hsiang and F. S Yen, J. Am. Ceram. Soc., 1996, 79, 1053. Lowering the reaction temperature produces powders with 17 R. Vivekanandan and T. R. N. Kutty, Powder Technol., 1989, even smaller particle sizes but very long reaction periods are 57, 181. required, >72 h, to ensure complete reaction. In addition, the 18 K. Uchino, N. Lee, T. Toba, N. Usuki, H.Aburatani and Y. Ito, powders are poorly crystalline and have high water content. J. Ceram. Soc. Jpn., 1992, 100, 1091. In order to obtain phase pure BaTiO3 ceramics it is import- 19 F. Yen, C. T. Chang and Y. Chang, J. Am. Ceram. Soc., 1970, 73, 3422. ant to control the Ba/Ti ratio. As the rate limiting step in the 20 J. Nowotny and M. Rekas, Solid State Ionics, 1991, 49, 135.synthesis involves reaction of the Ti-precursor, it is important 21 D. Hennings, M. Klee and R. Waser, Adv. Mater., 1991, 3, 334. to use a technique which can detect very small quantities of 22 M. Mori and T. Kineri, J. Am. Ceram. Soc., 1995, 78, 2391. unreacted amorphous or crystalline Ti-precursor material. We 23 A. I. Kingon, S. K. StreiVer, C. Basceri and S. R. Summerfelt, have clearly demonstrated that Raman spectroscopy, rather MRS Bull., 1996, 6, 46. 24 K. Wa. Gachigi, U. Kumar and J. P. Dougherty, Ferroelectrics, than XRD, is a simple and eVective technique for this purpose. 1993, 143, 229. Raman spectra of powders with an average particle size as 25 F. Chaput and J. P. Boilt, J. Am. Ceram. Soc., 1990, 73, 942. small as ca. 20–40 nm indicate asymmetry within the TiO6 26 M.Demartin, C. Herard, C. Carry and J. Lemaitre, J. Am. Ceram. octahedra of the BaTiO3 lattice. This contradicts the widely Soc., 1997, 80, 1079. cited ‘critical’ particle size theory for the stabilisation of the 27 W. Zhu, C. C. Wang, S. A. Akbar and R. Asiaie, J. Mater. Sci., cubic polymorph, at least for particle sizes in excess of ca. 1997, 32, 4303. 28 W.Zhu, C. C. Wang, S. A. Akbar, R. Asiaie, P. K. Dutta and 20–40 nm. Raman spectra of powders heat treated at ca. M. A. Alim, Jpn. J. Appl. Phys., 1996, 35, 6145. 1000 °C to remove adsorbed and structural water are very 29 A. Beauger, J. C. Mutin and J. C. Niepce, J. Mater. Sci., 1983, similar to that of as-prepared powders. This contradicts the 18, 3041. suggestion that lattice hydroxyl ions stabilise the cubic poly- 30 A.Beauger, J. C. Mutin and J. C. Niepce, J. Mater. Sci., 1983, morph of BaTiO3 prepared via a wet chemical technique, such 18, 3543. 31 J. C. Mutin and J. C. Niepce, J. Mater. Sci. Lett., 1984, 3, 591. as hydrothermal synthesis. Finally, as-prepared powders con- 32 A. Beauger, J. C. Mutin and J. C. Niepce, J. Mater. Sci., 1984, tain many defects, primarily lattice OH- ions.Preliminary 19, 195. conductivity results on pellets of powders which have been 33 R. Asiaie, W. Zhu, S. A. Akbar and P. K. Dutta, Chem. Mater., heated treated to remove lattice hydroxyl ions reveal these 1996, 8, 226. materials to be modest proton conductors at room 34 W. S. Clabaugh, E. M. Swiggard and R. Gilchrist, J. Res. Natl. temperature. Bur.Stand., 1956, 56, 289. 35 M. P. Pechini, US Pat., 3 330 697, 1967. 36 N. J. Ali and S. J. Milne, Trans. J. Br. Ceram. Soc., 1987, 86, 113. 37 M. H. Frey and D. A. Payne, Chem.Mater., 1995, 7, 123. Acknowledgements 38 A. D. Hilton and R. Frost, Key Eng. Mat., 1992, 66/67, 145. 39 M. M. Lencka and R. E. Riman, Chem. Mater., 1993, 5, 61. The authors would like to thank Dr Eric Lachowski for 40 W.Hertl, J. Am. Ceram. Soc., 1988, 71, 879. assistance with electron microscopy, Dr Alison Coats for 41 J. O. Eckert Jr., C. C. Hung-Houston, B. L. Gersten, EPMA, Dr Susan Blake for XRD, Mr James Marr for BET M. M. Lencka and R. E. Riman, J. Am. Ceram. Soc., 1996, 79, analysis and Professor Tony West for useful discussions. The 2929. EPSRC and AIST (Japan) are gratefully acknowledged for 42 C.H. Perry and D. B. Hall, Phys. Rev. Lett., 1965, 15, 700. financial support. 43 G. Busca, V. Buscaglia, M. Leoni and P. Nanni, Chem. Mater., 1994, 6, 955. 44 D. Hennings and S. Schreinemacher, J. Eur. Ceram. Soc., 1992, 9, 4. References 45 S. Wada, T. Suzuki and T. Noma, J. Ceram. Soc. Jpn., 1996, 104, 383. 1 B. JaVe, W. R. Cook and H. JaVe, Piezoelectric Ceramics, 46 R.Waser, J. Am. Ceram. Soc., 1988, 71, 58. Academic Press, London, 1971. 47 I. J. Clark, T. Takeuchi, N. Ohtori and D. C. Sinclair, unpub- 2 D. Hennings, Int. J. High Technology Ceramics, 1987, 3, 91. lished work. 3 G. Arlt, D. Hennings and G. de With, J. Appl. Phys., 1985, 58, 48 M. H. Frey and D. A. Payne, Phys. Rev. B, 1996, 54, 3158. 1619. 49 H. P. Abicht, D.Voltzke, A. Roder, R. Schneider and 4 M. H. Frey, Z. Xu, P. Han and D. A. Payne, Ferroelectrics, 1998, J. Woltersdorf, J. Mater. Chem., 1997, 7, 487. 206, 337. 50 N. Hirose and A. R.West, J. Am. Ceram. Soc., 1996, 79, 1633. 5 W. R. Bussen, L. E. Cross and A. K. Goswami, J. Am. Ceram. 51 I. J. Clark and D. C. Sinclair, unpublished work. Soc., 1966, 49, 33. 52 H. Iwahara, T. Esaka, H.Uchida and N. Maeda, Solid State 6 G. Arlt, Ferroelectrics, 1990, 104, 217. Ionics, 1981, 3/4, 359. 7 A. Morell and J. C. Niepce, J. Mater. Educ., 1991, 13, 173. 53 H. Uchida, H. Yoshikawa and H. Iwahara, Solid State Ionics, 8 T. Takeuchi, K. Ado, T. Asai, H. Kageyama, Y. Saito, 1989, 34, 103. C. Masquelier and O. Nakamura, J. Am. Ceram. Soc., 1994, 77, 54 K. D. Kruer, Chem.Mater., 1996, 8, 610. 1665. 9 W. Kanzig, Phys. Rev., 1955, 98, 549. 10 K. Kinoshita and A. Yamaji, J. Appl. Phys., 1976, 47, 371. Paper 8/05756G J. Mater. Chem., 1999, 9, 83–91 91 J O U R N A L O F C H E M I S T R Y Materials Hydrothermal synthesis and characterisation of BaTiO3 fine powders: precursors, polymorphism and properties† Iain J. Clark,a Tomanari Takeuchi,b Noikazu Ohtoric and Derek C.Sinclaira aChemistry Department, University of Aberdeen, Meston Walk, Aberdeen, UK AB24 3UE bOsaka National Research Institute, 1–8-31, Midorigaoka, Ikeda, Osaka, Japan cGraduate School of Science and Technology, Niigatu University, Ikarashi 2-no cho, Niigata, Japan Received 19th June 1998, Accepted 22nd July 1998 The influence of two Ti-precursors, TiO2 (anatase) and H2TiO3 (b-titanic acid), on the purity and particle size of BaTiO3 powders prepared via hydrothermal synthesis is discussed. Amorphous H2TiO3 was found to be an excellent Ti-precursor material and oVers several advantages over crystalline anatase.Phase pure powders which have small particle sizes, ca. 40–80 nm and narrow particle size distributions can be prepared at 180 °C after 24 h using H2TiO3 as a precursor material.Although the initial reaction is very fast, ca. 90% yield after 8–10 h, extended reaction periods at 180 °C are required in order to drive the reaction to completion. Lowering the reaction temperature from 180 to 85 °C does produce powders with even smaller particle sizes, however, very long reaction periods are required, e.g.>72 h, to ensure complete reaction.Raman spectra of as-prepared and heat treated (1000 °C) powders with average particle sizes as small as ca. 20–40 nm indicate asymmetry within the TiO6 octahedra of the BaTiO3 lattice. These results contradict the widely cited ‘critical’ particle size theory for the stabilisation of the cubic polymorph, at least for particle sizes greater than ca. 20–40 nm. As-prepared powders contain many defects, primarily in the form of lattice OH- ions. Preliminary ac impedance spectroscopy data on samples heat treated to remove lattice hydoxyl ions demonstrate these materials to be modest proton conductors. size eVect but there is still no clear consensus as to its origin.‡ Introduction Two commonly cited models involve the generation of higher- BaTiO3 is one of the most widely used and studied ferroelectric than-usual stress in fine grained ceramics5,6 whereas a third materials in the electro-ceramics industry.It has various suggests a core-shell structure,7,8 whereby grains below a polymorphs, all of which are based on the perovskite struc- ‘critical’ size (<0.2 mm) have cubic symmetry, intermediate ture;1 however, the two most studied are the tetragonal, t, and grain sizes (ca. 1 mm) have a ferroelectric-tetragonal core with cubic, c, polymorphs. t-BaTiO3 forms between ca. 0 and 130 °C a cubic surface layer and large grains (5 mm) are essentially whereas c-BaTiO3 is stable above 130 °C. In the t-polymorph, tetragonal. It should be noted that there is not an accepted titanium ions are displaced from the centrosymmetric position ‘critical’ particle or grain size below which the cubic polymorph within TiO6 octahedra and give rise to spontaneous polaris- is stabilised, on the contrary, a wide range of values, ca.ation. Due to this asymmetry within the crystal structure, t- 25–190 nm have been reported.9–16 In addition, a variety of BaTiO3-based materials have high permittivities and are widely reasons have been proposed for the room-temperature stabilisemployed as dielectrics in ceramic capacitors.2 The transition ation of the cubic polymorph, including the presence of lattice from the polar (ferroelectric) t- to non-polar (paraelectric) c- ‘defects’ such as hydroxyl ions17 (associated with powders polymorph normally occurs at ca. 130 °C, the Curie tempera- formed via wet chemical methods), small deviations in Ba/Ti ture, Tc.Above Tc, the Ti ions occupy, on average, the stoichiometry7 and excess surface energy associated with centrosymmetric position within TiO6 octahedra and there is ultrafine particles.18,19 Clearly, much remains to be done in no net polarisation within the solid. order to establish the factors which influence and control the A typical room temperature permittivity value, e25, for crystal symmetry and electrical properties of sub-micron ceramic t-BaTiO3 with an average grain size of ca. 10 mm is BaTiO3 powders and ceramics. 1000–2000; however, this rises sharply to a maximum value, Despite these fundamental problems, BaTiO3 will continue emax, of ca. 10 000 at Tc. In such ‘large’ grained ceramics, a to be used in the manufacture of thermistors,20 multilayer variety of ferroelectric domain structures form to relieve capacitors,21 electro-optic devices22 and DRAM (dynamic internal stress associated with the cubic to tetragonal phase random access memories)23 into the next century.Improved transformation. Although a detailed discussion of domain performance and miniaturisation of BaTiO3-based devices, structures in BaTiO3 powders and ceramics is outwith the either in the form of thin ceramic layers, ca. 20mm, for scope of this paper, it is important to note that ferroelectric multilayer capacitors or as thin films for integrated electronic domain structures and dipole–dipole interactions control the circuits remains a priority. High permittivities and miniaturispermittivity characteristics of BaTiO3 and that these depend ation can be achieved by controlling the ceramic microstructure on particle or grain size.For example, Arlt et al.3 have shown which, in turn, depends on the homogeneity, composition, that e25 for BaTiO3 ceramics can be optimised to a value surface area and particle size of the starting BaTiO3 powder. approaching 5000 by controlling the grain size to be ca.In addition, it is well known that BaTiO3 undergoes exagger- 0.7–1.0 mm. ated grain growth during sintering at elevated temperatures Several models have been proposed to explain this grain and long sintering periods. This can produce ceramics †Basis of the presentation given at Materials Chemistry Discussion ‡During the preparation of this manuscript, Frey et al.4 published a paper which appears to have clarified many of the problems associated No. 1, 24–26 September 1998, ICMCB, University of Bordeaux, France. with the origin of the so-called grain size eVect in BaTiO3. J. Mater. Chem., 1999, 9, 83–91 83containing very large grain sizes, e.g. >50 mm or duplex and the surrounding BaTiO3. For this model, complete reaction may be diYcult to achieve due to increasing encapsulation microstructures of low density consisting of both large (>10 mm) and small (<1 mm) grains.For these reasons, many of TiO2 via product formation. Homogeneous and heterogeneous dissolution–precipitation synthetic methods and sintering profiles24–28 have been investigated in an attempt to produce sub-micron, deagglomerated models have also been proposed.In the homogeneous model, a low concentration of TiO2 dissolves in the form of soluble fine powders of BaTiO3 that can be sintered into dense, finegrained ceramics. hydroxytitanium complexes which then react with Ba2+ ions in solution to precipitate BaTiO3. In the heterogeneous model, Traditionally, BaTiO3 has been produced via the mixed oxide route.This method involves repeated calcination and BaTiO3 nuclei form on the surfaces of the dissolving TiO2 particles and, as for the in situ transformation mechanism, the regrinding of BaCO3 and TiO2 powders above 1100 °C. The reaction mechanism in air has been proposed to take place in reaction becomes limited by the increasing isolation of the reactants via BaTiO3 product layers on the surface of TiO2 at least three stages29–32 and relies on the diVusion of Ba2+ ions into TiO2.Firstly, BaCO3 reacts with the outer surface particles. Eckert et al. reviewed these models41 in more detail and regions of TiO2 to form a surface layer of BaTiO3 on individual TiO2 grains. Further diVusion of Ba2+ ions into TiO2 necessi- undertook a kinetic study of hydrothermal reactions involving barium hydroxide octahydrate and anatase precursors.Two tates the formation of Ba2TiO4 between unreacted BaCO3 and the previously formed BaTiO3. After prolonged sintering reaction regimes were clearly identified; during the early stages the reaction is controlled by a dissolution–precipitation pro- periods, the intermediate Ba-rich phase Ba2TiO4 reacts with the remaining TiO2 in the core-regions of the TiO2 grains to cess, whereas, for the second regime at longer reaction times the results were inconclusive.The authors suggest two plausible form BaTiO3. Although the nominal starting Ba/Ti ratio is 151, long reaction times are required to form homogeneous but diVerent explanations for their kinetic data, one model is based on heterogeneous dissolution–precipitation followed by powders free from secondary phases.For such a complex and slow reaction mechanism, it is desirable to have high surface in situ transformation, whereas, the second model suggests that dissolution–precipitation is the controlling mechanism, area TiO2 powders of fine particle size and narrow size distribution in order to control the morphology and grain size with nucleation and growth of BaTiO3 controlling the first regime and the dissolution rate of TiO2 controlling the of the resultant BaTiO3 powder.Nevertheless, the mixed oxide route tends to produce coarse, agglomerated powders that second regime. In the present paper we discuss some of our recent work on require high sintering temperatures to form dense ceramics.In order to produce powders that are suitable for sintering the hydrothermal synthesis of BaTiO3 powders and highlight some advantages of using amorphous H2TiO3 as opposed to into micron-grain-sized ceramics for MLC applications there has been much interest in developing wet and novel chemical crystalline anatase as a precursor material. In particular, H2TiO3 promotes faster and more-complete reactions and also routes.As these reactions take place in the liquid, as opposed to the solid state, more intimate mixing of the cations is produces powders with much smaller particle sizes. We discuss the influence of temperature on the rate of reaction and on achieved and consequently, much shorter diVusion pathways are created. Deagglomerated, sub-micron powders of BaTiO3 the particle size, distribution and water content of the BaTiO3 powders. The importance of using Raman spectroscopy to can then be obtained at low temperatures and short reaction periods, for example via hydrothermal synthesis at 240 °C detect the presence of trace amounts of unreacted Ti-containing precursors and to yield information on the polymorphism of for 24 h.33 Other common methods include, oxalate (Clabauch),34 citrate (Pechini),35 catecholate (Milne)36 and sub-micron powders is discussed.Finally, we present preliminary results on powders and ceramics heat treated above sol–gel.37 Although the reaction mechanisms and advantages and disadvantages of many of these routes have been dis- 800 °C that support the idea that hydrothermal BaTiO3 powders contain many defects which influence their physical cussed,38 for the present purposes it is worth discussing these issues for hydrothermal synthesis.properties. BaTiO3 powders have been prepared via hydrothermal processing since the 1940’s and this method is commonly employed Experimental to produce commercial powders. Although it is well established that BaTiO3 is the only thermodynamically stable binary Ba(OH)2·8H2O (98+%, Aldrich) and anatase (99.9%, Aldrich) or b-H2TiO3 (+99%,Mitsuwa Chemicals) were mixed compound produced under all conditions,39 i.e.pH, Ba/Ti ratio, reaction temperature, etc. products can contain unre- with a Ba5Ti ratio of 1.0551 in a 23 ml Teflon-lined pressure vessel (Model 4749, Parr Instruments) together with 10 ml of acted TiO2 and BaCO3.Normally, an excess of Ba2+ in the starting solution is employed (Ba/Ti=1.05–1.1051) in an de-ionised water. The vessel was sealed, shaken and placed in an oven at either 85, 120 or 180 °C for periods ranging from attempt to remove any unreacted TiO2 and therefore drive the reaction to completion. Any excess Ba, in the form of BaCO3 1 to 72 h.After cooling, the contents of the bomb were diluted in 30 ml of 0.1M formic acid in an attempt to dissolve any can then be removed via acid washing of the powders. Although this procedure is eVective in controlling the purity BaCO3 formed by the addition of excess Ba2+ to the starting solution. The mixture was vacuum filtered using a Buchner of the powders, relatively little is known about the influence of acid washing on powder stoichiometry as Ba2+ may be funnel, the residue thoroughly washed with distilled water and dried overnight in air at 120 °C.The percentage yield of leached from the surfaces of the particles. Although controlling the precise stoichiometry of hydrothermal BaTiO3 powders BaTiO3 was estimated from the weight of the dried powders.Phase purity and polymorphism of the BaTiO3 powders can be problematic, the low cost and easy handling of the reagents, and the fast reaction rate at low temperatures ensures were studied by X-ray diVraction (XRD) and Raman spectroscopy. XRD was carried out on a Stoe STADI P automated that deagglomerated powders consisting of small particles of narrow size distribution are readily obtained.powder diVractometer employing monochromatic Cu-Ka1 radiation and a linear position sensitive detector over the 2h Several mechanisms have been proposed for the hydrothermal synthesis of BaTiO3. Hertl40 has suggested an in situ range 20 to 80 °. Cubic and tetragonal polymorphs were distinguished by the splitting of the (200) peak at 2h#45 °. transformation whereby dissolved Ba2+ ions react with undissolved TiO2 to produce a continuous layer of BaTiO3 on the Unit cell parameters were obtained by refinement of XRD data.Raman spectroscopy was carried out on a Jobin Yvon surface of TiO2 particles. The reaction is then controlled either by diVusion of Ba2+ through the product BaTiO3 layer or T64000 using a 100 mW Ar laser with a wavelength of 514.5 nm. A 300 mm slit and integration time of 5 s were used further reaction between unreacted TiO2 within grain interiors 84 J.Mater. Chem., 1999, 9, 83–91giving a resolution of 8.4 cm-1. Spectra were measured over An important aspect of this work is to obtain homogeneous the range 58–1103 cm-1. Impurity phases such as BaCO3 and particles of controlled stoichiometry which can be sintered TiO2 (anatase) were detected from characteristic peaks at 1060 into dense ceramics of uniform microstructure.The Ba5Ti and 150 cm-1, respectively. The presence of a peak in the ratio of the particles is an important parameter but is not spectra of BaTiO3 at 305 cm-1 indicates asymmetry within readily obtained via XRD or Raman spectroscopy. We choose, TiO6 octahedra42 and was used to distinguish between tetra- therefore, to study the phase assemblage and Ba5Ti ratio of gonal and cubic polymorphs.the grains in sintered ceramics using EPMA and from this, Thermogravimetric analysis (TGA) was carried out using a deduce information about the Ba5Ti ratio of the powders. Stanton Redcroft TG-DTA simultaneous thermal analyser This is an indirect method for obtaining information on the (model STA 1000/1500) to determine the water content of as- stoichiometry of the powders, but since neither barium nor prepared powders.Samples were accurately weighed to five titanium are particularly volatile at the sintering temperature, decimal places in a platinum crucible and weight losses we perceive this method to be valid. recorded from 25–1000 °C against an alumina reference. All experiments were carried out in air at a heating rate of 10 °Cmin-1 with an amplifier setting of 50 mV.Results and discussion Powder morphology was determined via transmission electron microscopy (TEM) using a JEOL electron microscope The results are subdivided into three sections. The first section (model 2000EX TEMSCAN) operating at 200 kV.Particle discusses the influence of two Ti-precursors, TiO2 (anatase) size and distribution were obtained by measuring the cross and H2TiO3 (b-titanic acid), on the purity and particle size of diagonals of approximately 130 particles from scanned nega- the BaTiO3 powders. The second section assesses the role of tives of TEM images. Energy dispersive X-ray analysis (EDX) reaction temperature on the rate of reaction, purity, particle was used in an attempt to determine the Ba5Ti ratio of the size and water content of BaTiO3 powders prepared from BaTiO3 particles and also to detect the presence of any Ba(OH)2·8H2O and H2TiO3.The final section discusses preunreacted TiO2. liminary electrical property measurements on powders heat Electron probe micro-analysis (EPMA) was employed to treated above 800 °C. determine the homogeneity of sintered pellets fabricated from the various powders. Pellets were sintered at ca. 1350 °C and I Titanium precursors a Cameca SX51 electron microprobe employed using an accelerating voltage of 20 kV and a beam current of 40 nA. One of the major problems associated with the hydrothermal Ba-La and Ti-Ka lines were measured using benitoite, synthesis of BaTiO3 is the low solubility of TiO2 in the highly BaTiSi3O9, as a standard.Prior to analysis, small pieces of alkaline environments, ca. pH>12, required for synthesis. pellet were mounted in resin blocks, surface polished and Although it is well documented that diVerent Ti-precursors carbon coated. and reaction temperatures influence the rate of reaction there Electrical measurements were carried out using a Solartron have been few comprehensive studies on the influence of these frequency response analyser (model 1250 or 1260) combined parameters on the purity and particle size of hydrothermal with a Solartron dielectric interface (model 1296).A frequency powders. Kutty et al. suggested that amorphous Ti-precursors range of 10-3–106 Hz was employed with an applied voltage react fastest,17 then anatase with rutile giving the slowest of 100 mV.Cold pressed pellets of various powders were reaction. Various workers reported an increase in average sintered between 800 and 1350 °C and organopaste and gold particle size and a change (via XRD) from cubic to tetragonal foil electrodes applied to the surfaces of the pellets at 800 °C, symmetry with increasing reaction temperature.9–16We decided prior to mounting in a conductivity jig.Measurements were to reinvestigate some of these findings by comparing the recorded over the temperature range 25–500 °C. reaction rates, powder purity and particle size of hydrothermal Before discussing the results it is important to appreciate BaTiO3 powders prepared at various temperatures and reac- the need for a wide variety of techniques to determine the tion periods using two Ti-precursors: amorphous H2TiO3 phase purity and crystal structure of fine-grain BaTiO3 pow- (BET using N2, surface area 193 m2 g-1) and crystalline ders.XRD is an excellent technique for obtaining information anatase (BET, surface area 8.9 m2 g-1).on the average or long range crystal structure of a material; In general, for a set reaction temperature and period, the however, peak broadening is significant for diVraction from reaction rate was faster and the average BaTiO3 particle size sub-micron sized particles and XRD cannot reveal the very smaller for reactions using the amorphous H2TiO3. Raman subtle unit cell distortions which occur in BaTiO3 for microdospectra of powders prepared under the same conditions for mains of dimensions of ca. 100 A° . In addition, routine XRD the two precursors are shown in Fig. 1. Spectra of anatase and is relatively insensitive to small quantities of impurity phases, the tetragonal polymorph of BaTiO3 prepared via the tra- especially for secondary phases which are amorphous/poorlyditional mixed oxide route are included for comparison.The crystalline and/or have weak X-ray scattering power. In our Raman spectrum of H2TiO3 is not shown but is similar to experience, XRD is sensitive to small quantities of BaCO3 in that of anatase and is also dominated by a large peak at ca. hydrothermally processed BaTiO3 but is much less sensitive in 150 cm-1.There are two noticeable features in the Raman the detection of unreacted Ti-containing precursors, either spectra of the products which are not readily discernible from crystalline anatase or amorphous H2TiO3. the XRD patterns (not shown) and which provide important Uncertainties over crystal symmetry can largely be resolved information. using alternative techniques such as Raman spectroscopy to First, the powder products from the H2TiO3 reaction appear probe short range order or local symmetry.There are reliable to be phase pure BaTiO3, whereas unreacted TiO2, as shown Raman spectra in the literature42 for various BaTiO3 polyby the presence of the peak at ca. 150 cm-1, is clearly present morphs; however, it must be stated that the diVerences in in the products of the anatase reaction.Small amounts of spectra are rather subtle, especially for the orthorhombic, unreacted anatase or amorphous H2TiO3 are diYcult to detect tetragonal and cubic (at temperatures slightly greater than Tc) via routine XRD but are readily detected (by the peak at ca. polymorphs. In contrast to XRD, Raman spectroscopy is 150 cm-1) in the Raman spectra.This result demonstrates the sensitive to the presence of both BaCO3 and unreacted Tiimportance of using Raman spectroscopy to monitor the phase containing precursors. For these reasons we choose to present purity of BaTiO3 powders prepared via hydrothermal syn- Raman spectra, as opposed to XRD patterns to illustrate the thesis. We should stress that single phase powders can, in fact, phase purity and crystal symmetry of as-prepared BaTiO3 powders.be prepared from crystalline anatase at 180 °C, however the J. Mater. Chem., 1999, 9, 83–91 85Fig. 2 TEM micrographs of BaTiO3 powders prepared via hydrothermal synthesis at 180 °C for 24 h using anatase (a) and H2TiO3 (b), as Ti-precursors. Fig. 1 Raman spectra of BaTiO3 prepared via a mixed oxide route (top) and via hydrothermal synthesis at 180 °C for 24 h using H2TiO3 and anatase as Ti-precursors.The bottom spectrum is of crystalline II Reaction temperature anatase. Raman spectra of powder products from two reaction temperatures, 85 and 180 °C after various reaction periods are shown in Fig. 3(a) and (b), respectively. With increasing time, both times required are far in excess of those required using sets of spectra illustrate a decrease in peak intensity at amorphous H2TiO3, typically in excess of 72 h. 150 cm-1 associated with the titanium precursor as BaTiO3 is Second, the presence of a peak at 305 cm-1 in both spectra formed.In fact, the spectrum after reaction at 180 °C for 24 h indicates asymmetry within the TiO6 octahedra of BaTiO3 and shows no evidence of this peak, suggesting that the reaction demonstrates clearly, on a local scale, that as-prepared powders has gone to completion. A small peak at 150 cm-1 is still do not have cubic symmetry.In contrast, XRD patterns apparent in the corresponding spectrum for powders prepared showed appreciable peak broadening, presumably associated at 85 °C. A higher reaction temperature of ca. 180 °C, is clearly with the small particle size and poor crystallinity of the required to ensure a more complete reaction for short reaction powders, however, there was little evidence of peak splitting periods, e.g. 24 h. The high surface area (193 m2 g-1) and the and the XRD data ‘appeared’ consistent with that for the fact that hydroxylation of many Ti–O–Ti bridging bonds has cubic polymorph, as reported previously.9–16 The higher sensi- already been achieved in H2TiO3, ensure rapid formation of tivity of Raman spectroscopy to probe the local rather than BaTiO3, even at 85 °C.It should be noted that many spectra long range structure clearly demonstrates that sub-micron contained a small peak at 1060 cm-1 (not shown) associated powders prepared via hydrothermal processing are tetragonal with a small amount of BaCO3 which was not completely rather than cubic.removed by the mild acid wash. The particle size of powders prepared from H2TiO3 were, The weight of the dried powders was used to calculate the on average, 3–5 times smaller than those prepared from TiO2 percentage yield of BaTiO3 and is shown in Fig. 4. We stress and, to our knowledge, are amongst the smallest reported for that these values represent upper estimates of the yield, especi- BaTiO3 powders prepared hydrothermally. Representative ally for short reaction periods where powders contain unremicrographs of powders prepared at 180 °C for 24 h from acted precursor and are also heavily hydrated. Nevertheless, H2TiO3 and TiO2 are shown in Fig. 2 and demonstrate the these estimates show that reaction is rapid and yields in excess much smaller particle size, ca. 80 nm compared with ca. of 90% are obtained within 8 h of reaction. From the Raman 250 nm. The faster rate of reaction and smaller particle size spectra in Fig. 3, extended periods, e.g. 24 h at elevated of powders prepared from amorphous H2TiO3 demonstrate temperatures (180 °C) are required in order to drive the two clear advantages of using this precursor instead of crystal- reaction close to completion, even for a reactive precursor line anatase.The remaining sections of the paper concentrate such as H2TiO3. on powders and ceramics prepared using H2TiO3 as the Ti- As-prepared powders are hydrated and lose water both adsorbed and structural on heating to ca. 1000 °C. In general, precursor. 86 J. Mater. Chem., 1999, 9, 83–91Fig. 3 Raman spectra of powder products from two reaction temperatures, 85 (a) and 180 °C (b), after various reaction periods. dramatic decrease in reaction rate after ca. 8–10 h, as shown by the percentage yield of BaTiO3. Nucleation of BaTiO3 crystals presumably takes place via a structural rearrangement of the amorphous H2TiO3 associated with the incorporation of Ba2+ or via direct reaction in solution of dissolved hydroxytitanium complexes with Ba2+ ions.Either way, the reaction involves dehydration, for example, for the rearrangement mechanism (1). H2TiO3+Ba2++2OH-�BaTiO3+H2O (1) Dehydration is slow in superheated fluids and leads to partial retention of H2O and OH- in BaTiO3, especially for powders prepared at low temperatures.The variation in particle size as a function of temperature and time are shown in Fig. 5. All powders have narrow particle size distribution ranges and for each temperature, the mean particle size approximately doubles over a period of ca. 20 h, Fig. 4 Percentage yield (weight%) of powder products from reactns e.g. 23±5 to 43±12 nm, between 2 and 24 h at 85 °C and at 85 and 180 °C and their associated water content (mol%) as a from 53±14 to 80±15 nm at 180 °C. Particle size values after function of reaction period. reaction at 120 °C were intermediate between those at 85 and 180 °C. On comparing the particle size values with the variations in product yield, it appears that after an initial ‘burst’ weight loss commenced at ca. 100 °C and was complete at ca. 600 °C. The powders prepared at 85 °C clearly contained of nucleation, growth of BaTiO3 particles is a rather slow process. substantially more water, especially for short reaction periods where up to 4 weight% loss was recorded. The water content Micrographs of powders obtained at 85 and 180 °C for 2 h are shown in Fig. 6. At 85 °C, they have poor crystallinity and of powders for each reaction temperature remained reasonably constant after ca. 8 h, Fig. 4. This coincides with the ill-defined morphology, Fig. 6(a), whereas at 180 °C they are J. Mater. Chem., 1999, 9, 83–91 87Fig. 5 Particle size histograms for powders prepared at 85 °C after 2 (a) and 24 h (b) and at 180 °C after 2 (c) and 24 h (d).well-defined, regular particles, Fig. 6(b). The powders prepared at 85 °C have a higher water content and presumably, a higher concentration of lattice defects, such as OH- ions. Despite the poor crystallinity of such powders, the presence of the peak at 305 cm-1 in all Raman spectra, Fig. 3(a), demonstrates that these small particles, ca. 20 nm, are tetragonal rather than cubic.This result suggests that the widely-cited ‘critical’ particle/grain size model is incorrect and that distortion of Ti–O bonds within the BaTiO3 lattice is possible, even for particles as small as ca. 20 nm and which contain substantial amounts of lattice defects, such as OH- ions. Although our results clarify some of the existing problems in the literature we are unable, at this stage, to identify the reaction mechanism(s) involved in hydrothermal synthesis of BaTiO3.In the final section we discuss some of the properties of heat treated, hydrothermal BaTiO3 powders. III Heat treated powders Hydroxyl ions play an important role in the synthesis of BaTiO3 via hydrothermal processing, as high pH environments, >12, are required to obtain single phase materials.Infrared spectroscopy17,43,44 and TGA45 have both been employed to demonstrate that as-prepared hydrothermal powders contain weakly-bound water molecules adsorbed onto particle surfaces and more strongly bonded structural water in the form of lattice OH- ions. In general, only total water contents are reported as deconvolution into the two distinct types is diYcult. In order to maintain electro-neutrality, it is generally accepted that barium vacancies (VBa) are created on the surfaces of individual particles46 to compensate for the incorporation of lattice OH- ions on O2- sites (OHOV), according to eqn.(2). 2[OHOV]=VBa (2) On heating above ca. 300 °C, lattice OH- ions are removed as follows, Fig. 6 TEM micrographs of powders prepared at 85 (a) and 180 °C (b), after a reaction period of 2 h. 2OHOV�Oo x+VOVV+H2O (g) (3) 88 J. Mater. Chem., 1999, 9, 83–91resulting in powders which contain significant concentrations of VBa and VOVV. High concentrations of lattice hydroxyl ions in as-prepared hydrothermal powders therefore influence the stoichiometry (Ba/Ti ratio), defect chemistry and grain growth of sintered BaTiO3 ceramics. Elimination of water in our powders was complete by ca. 600 °C, but there were no significant changes in the Raman spectra of the dehydrated powders until samples were heated >1000 °C. These subtle changes in spectra will be discussed elsewhere,47 however, the peak at 305 cm-1 was present in all processed powders, indicating the distorted crystal symmetry of the particles, irrespective of the presence of adsorbed water or lattice hydroxyl ions.This result contradicts the suggestion that lattice hydroxyl ions play a crucial role in controlling the crystal symmetry of hydrothermal BaTiO3 powders17,44,45 and is in agreement with the work of Frey and Payne who came to the same conclusion for BaTiO3 powders prepared via a Fig. 8 Lattice parameters (&) and c/a ratio (#) of BaTiO3 powders sol–gel route.48 prepared at 85 °C for 72 h as a function of processing temperature.A selection of XRD patterns over a limited 2h range for powders prepared at 85 °C and heated up to 1100 °C are shown As mentioned previously, cation and anion defects influence in Fig. 7. At 800 °C, the broad peak at ca. 45.2 ° (in as-made the stoichiometry, grain growth and electrical properties of hydrated powders) sharpens but clear splitting is not apparent BaTiO3 powders.The diVusion coeYcients for oxygen and until heating at ca. 1100 °C. In general, XRD reflections barium vacancies in BaTiO3 become appreciable above 800 become sharper and more intense after heating at higher and 1100 °C, respectively. Thus, on heating hydrothermal temperatures.XRD patterns for powders heated above 1100 °C powders at elevated temperatures, any barium deficiency, were indexed on a tetragonal unit cell, whereas, all others were either as compensating cation defects for lattice hydroxyl ions indexed on a cubic unit cell. Lattice parameters and c/a ratios or as a consequence of acid washing, or excess titanium, in are shown in Fig. 8 and demonstrate that powders heated at the form of unreacted Ti-containing precursor, should result temperatures >1100 °C have c/a ratios of ca. 1.01, which is in in the formation of Ti-rich phases. EPMA and SEM on good agreement with those reported in the literature.7 ceramic pellets sintered at 1350 °C from the powders produced at 85 and 180 °C after 24 h, both demonstrated the existence of Ti-rich secondary phases.For powders produced at 85 °C, where unreacted TiO2 was detected via Raman spectroscopy, Fig. 3(a), secondary phases were clearly visible as dark intergranular regions in back scattered electron images (BSE), Fig. 9(a). EPMA results identified the presence of both BaTi2O5 and Ba6Ti17O40 in inter-granular regions but their volume fractions were too small to be detected via XRD.For powders produced at 180 °C, where the products appeared phase-pure by Raman spectroscopy, sintered pellets contained only a very small Fig. 7 XRD diVractograms over the 2h range 44–46 ° of BaTiO3 Fig. 9 Back scattered electron images of ceramic pellets sintered at prepared at 85 °C (top) and after heat treatment at 800, 1000 and 1100 °C. 1350 °C from powders produced at 85 (a) and 180 °C (b). J. Mater. Chem., 1999, 9, 83–91 89volume fraction of Ba6Ti17O40, as shown by the dark, isolated regions in the BSE image, Fig. 9(b). The Ti-rich secondary phases in powders prepared at 85 °C are clearly associated with unreacted Ti-precursor, which reacts with surrounding BaTiO3 particles at elevated temperatures to form Ti-rich binary phases.It is unclear, however, whether the secondary phase produced from powders prepared at 180 °C is associated with unreacted Ti-precursor which has not been detected via Raman spectroscopy (or analytical TEM) or is associated with a compositional Ba/Ti gradient within individual particles of the as-prepared powders. Direct evidence of VBa on the surfaces of individual particles, either from incorporation of lattice hydroxyl ions during hydrothermal synthesis or via acid/aqueous media wet milling of powders has been limited.Abicht et al.49 used HREM and EELS to demonstrate that the outer surfaces of individual BaTiO3 particles which have been wet milled in aqueous media Fig. 11 Log resistivity versus 1000 K/T for pellets of powder prepared show evidence of Ba2+ leaching.They propose that a Ba/Ti at 85 °C and sintered at 1000 °C (filled squares) and 1350 °C (open concentration gradient exists within individual particles which symbols). Calculated activation energies are in eV. consists of a 3–5 nmhick TiOx-rich outer layer followed by a intermediate layer, ca. 10 nm thick, with a molar Ba/Ti ratio omena at the electrodes and supports the idea that conduction increasing from 0 to 1.These core-shell structures aVect the is mainly by means of ions at these temperatures. sintering of BaTiO3 powders due to the reactivity of the Ti- Bulk resistivity values were extracted from the low frequency rich, outer layers. Our observation of small quantities of Ti- intercept of the semi-circular arc with the Z¾ axis on the Z* rich intergranular regions in sintered pellets of hydrothermal plots.Initially, the bulk resistivity increases on heating, as powders produced at 180 °C is consistent with this model but shown by the increase in diameter of the bulk semi-circular in-depth studies using HREM and EELS are required to arc in the Z* plots, Fig. 10, and in the log resistance, Arrheniusestablish the presence or absence of any Ba/Ti compositional type plot shown in Fig. 11. The bulk resistivity rises by several gradient within individual particles of BaTiO3 (produced via orders of magnitude on heating to ca. 150 °C, then remains hydrothermal synthesis). temperature independent before decreasing in accordance with Heat treatment of hydrothermal powders at 1000 °C is the Arrhenius law at temperatures above ca. 350 °C. In this suYcient to remove any adsorbed or structural water and create high temperature region, there was no evidence of the low oxygen vacancies within the lattice, eqn. (3). It is insuYcient, frequency, inclined spike in Z*; instead, the plots consisted of however, to cause substantial migration of barium vacancies or a single, bulk semi-circular arc, indicating that conduction was densification and grain growth via liquid phase sintering involv- predominantly electronic.The resistivity behaviour was reversing Ti-rich outer particle surfaces and/or any excess unreacted ible on thermal cycling. For comparison, bulk and grain Ti-precursor material. Consequently, ceramic pellets formed at boundary resistivities extracted from high and low frequency 1000 °C are poorly sintered, have low density, ca. 65% of the semicircular arcs in Z* plots for a pellet sintered at 1350 °C theoretical density, and consist of small grains with very poor are also shown in Fig. 11. These values are consistent with inter-granular contact. Despite this the pellets exhibit exception- those reported in the literature for dense BaTiO3 ceramics.50 ally low room temperature resistivities, ca. 10–50MV. It should be noted that the room temperature resistivity of Complex impedance plane, Z*, plots at 32 and 44 °C for a these ceramics is in excess of 1 TV and that no low frequency, pellet sintered at 1000 °C are shown in Fig. 10. The plots consist inclined spike was observed in any Z* plots. of a high frequency semicircular arc with an associated capaci- The exceptionally low bulk resistivity value of 20 MV at tance of 32 pF and a low frequency-spike with an associated 25 °C, the presence of a low frequency spike in Z* plots below capacitance in the order of 1 mF.The capacitance value for the ca. 150 °C and the fact that resistivity initially rises with bulk or intra-granular component is consistent with a poorly temperature, all indicate that BaTiO3 prepared via hydrothermal sintered BaTiO3 ceramic and the low frequency, inclined spike synthesis at 85 °C and sintered at 1000 °C is a modest proton is attributable to ionic polarisation and diVusion-limited phen- conductor, especially at temperatures close to room temperature.Presumably water vapour is adsorbed from the ambient on cooling (from the sintering temperature of 1000 °C) and partially reverses the reaction given in eqn.(3), thus, converting doubly ionised oxygen vacancies into lattice hydroxyl ions. Subsequent reheating removes water rather easily and the bulk resistivity increases rapidly; however, temperatures in excess of 350 °C are required before the predominant charge carriers are electronic. More detailed studies of this unusual bulk resistivity behaviour will be reported elsewhere,51 however, such materials are clearly poor dielectrics.To our knowledge, this is the first time that proton conduction has been demonstrated in oxygen-deficient BaTiO3 materials. Many other oxygen-deficient perovskites such as doped, alkaline earth zirconates52 and cerates53 are well-known proton conductors.Several diVerent mechanisms of incorporation of protonic defects in such oxides have been suggested,54 involving interstitial protons and lattice hydroxyl ions, H2O (g)+VOVV=2HiV+OO x (4) Fig. 10 Complex impedance plane plots at 32 and 44 °C for a pellet H2O (g)+OO x+VOVV=2OHOV (5) of powder prepared at 85 °C and sintered at 1000 °C.Selected frequencies (in Hz) are shown for filled symbols. however, the detailed mechanism remains unclear. 90 J. Mater. Chem., 1999, 9, 83–9111 M. Kataoka, K. Suda, N. Ishizawa, F. Marumo, Y. Shimizugawa Conclusions and K. Ohsumi, J. Ceram. Soc. Jpn., 1994, 102, 213. 12 K. Uchino, E. Sadanaga and T. Hirose, J. Am. Ceram. Soc., 1989, Amorphous H2TiO3 is an excellent Ti-precursor to use in the 72, 1555.hydrothermal synthesis of BaTiO3. Phase pure powders with 13 F. J. Gotor, C. Real, M. J. Dianez and J. M. Criado, J. Solid State small particle sizes, ca. 40–80 nm and narrow particle size Chem., 1996, 123, 301. distributions can be prepared at 180 °C after 24 h. Although 14 S. Schlag and H. F. Eicke, Solid State Commun., 1994, 91, 883. initial reaction is very fast, ca. 90% yield after 8–10 h, extended 15 B. D. Begg, E. R. Vance and J. Nowotny, J. Am. Ceram. Soc., 1994, 77, 3186. heating at 180 °C is required to drive reactions to completion. 16 H. I. Hsiang and F. S Yen, J. Am. Ceram. Soc., 1996, 79, 1053. Lowering the reaction temperature produces powders with 17 R. Vivekanandan and T. R. N. Kutty, Powder Technol., 1989, even smaller particle sizes but very long reaction periods are 57, 181.required, >72 h, to ensure complete reaction. In addition, the 18 K. Uchino, N. Lee, T. Toba, N. Usuki, H. Aburatani and Y. Ito, powders are poorly crystalline and have high water content. J. Ceram. Soc. Jpn., 1992, 100, 1091. In order to obtain phase pure BaTiO3 ceramics it is import- 19 F. Yen, C. T. Chang and Y.Chang, J. Am. Ceram. Soc., 1970, 73, 3422. ant to control the Ba/Ti ratio. As the rate limiting step in the 20 J. Nowotny and M. Rekas, Solid State Ionics, 1991, 49, 135. synthesis involves reaction of the Ti-precursor, it is important 21 D. Hennings, M. Klee and R. Waser, Adv. Mater., 1991, 3, 334. to use a technique which can detect very small quantities of 22 M. Mori and T. Kineri, J.Am. Ceram. Soc., 1995, 78, 2391. unreacted amorphous or crystalline Ti-precursor material. We 23 A. I. Kingon, S. K. StreiVer, C. Basceri and S. R. Summerfelt, have clearly demonstrated that Raman spectroscopy, rather MRS Bull., 1996, 6, 46. 24 K. Wa. Gachigi, U. Kumar and J. P. Dougherty, Ferroelectrics, than XRD, is a simple and eVective technique for this purpose. 1993, 143, 229.Raman spectra of powders with an average particle size as 25 F. Chaput and J. P. Boilt, J. Am. Ceram. Soc., 1990, 73, 942. small as ca. 20–40 nm indicate asymmetry within the TiO6 26 M. Demartin, C. Herard, C. Carry and J. Lemaitre, J. Am. Ceram. octahedra of the BaTiO3 lattice. This contradicts the widely Soc., 1997, 80, 1079. cited ‘critical’ particle size theory for the stabilisation of the 27 W. Zhu, C. C. Wang, S. A. Akbar and R. Asiaie, J. Mater. Sci., cubic polymorph, at least for particle sizes in excess of ca. 1997, 32, 4303. 28 W. Zhu, C. C. Wang, S. A. Akbar, R. Asiaie, P. K. Dutta and 20–40 nm. Raman spectra of powders heat treated at ca. M. A. Alim, Jpn. J. Appl. Phys., 1996, 35, 6145. 1000 °C to remove adsorbed and structural water are very 29 A. Beauger, J. C. Mutin and J. C. Niepce, J. Mater. Sci., 1983, similar to that of as-prepared powders. This contradicts the 18, 3041. suggestion that lattice hydroxyl ions stabilise the cubic poly- 30 A. Beauger, J. C. Mutin and J. C. Niepce, J. Mater. Sci., 1983, morph of BaTiO3 prepared via a wet chemical technique, such 18, 3543. 31 J. C. Mutin and J. C. Niepce, J. Mater. Sci. Lett., 1984, 3, 591. as hydrothermal synthesis. Finally, as-prepared powders con- 32 A. Beauger, J. C. Mutin and J. C. Niepce, J. Mater. Sci., 1984, tain many defects, primarily lattice OH- ions. Preliminary 19, 195. conductivity results on pellets of powders which have been 33 R. Asiaie, W. Zhu, S. A. Akbar and P. K. Dutta, Chem. Mater., heated treated to remove lattice hydroxyl ions reveal these 1996, 8, 226. materials to be modest proton conductors at room 34 W. S. Clabaugh, E. M. Swiggard and R. Gilchrist, J. Res. Natl. temperature. Bur. Stand., 1956, 56, 289. 35 M. P. Pechini, US Pat., 3 330 697, 1967. 36 N. J. Ali and S. J. Milne, Trans. J. Br. Ceram. Soc., 1987, 86, 113. 37 M. H. Frey and D. A. Payne, Chem.Mater., 1995, 7, 123. Acknowledgements 38 A. D. Hilton and R. Frost, Key Eng. Mat., 1992, 66/67, 145. 39 M. M. Lencka and R. E. Riman, Chem. Mater., 1993, 5, 61. The authors would like to thank Dr Eric Lachowski for 40 W. Hertl, J. Am. Ceram. Soc., 1988, 71, 879. assistance with electron microscopy, Dr Alison Coats for 41 J. O. Eckert Jr., C. C. Hung-Houston, B. L. Gersten, EPMA, Dr Susan Blake for XRD, Mr James Marr for BET M. M. Lencka and R. E. Riman, J. Am. Ceram. Soc., 1996, 79, analysis and Professor Tony West for useful discussions. The 2929. EPSRC and AIST (Japan) are gratefully acknowledged for 42 C. H. Perry and D. B. Hall, Phys. Rev. Lett., 1965, 15, 700. financial support. 43 G. Busca, V. Buscaglia, M. Leoni and P. Nanni, Chem. Mater., 1994, 6, 955. 44 D. Hennings and S. Schreinemacher, J. Eur. Ceram. Soc., 1992, 9, 4. References 45 S. Wada, T. Suzuki and T. Noma, J. Ceram. Soc. Jpn., 1996, 104, 383. 1 B. JaVe, W. R. Cook and H. JaVe, Piezoelectric Ceramics, 46 R. Waser, J. Am. Ceram. Soc., 1988, 71, 58. Academic Press, London, 1971. 47 I. J. Clark, T. Takeuchi, N. Ohtori and D. C. Sinclair, unpub- 2 D. Hennings, Int. J. High Technology Ceramics, 1987, 3, 91. lished work. 3 G. Arlt, D. Hennings and G. de With, J. Appl. Phys., 1985, 58, 48 M. H. Frey and D. A. Payne, Phys. Rev. B, 1996, 54, 3158. 1619. 49 H. P. Abicht, D. Voltzke, A. Roder, R. Schneider and 4 M. H. Frey, Z. Xu, P. Han and D. A. Payne, Ferroelectrics, 1998, J. Woltersdorf, J. Mater. Chem., 1997, 7, 487. 206, 337. 50 N. Hirose and A. R.West, J. Am. Ceram. Soc., 1996, 79, 1633. 5 W. R. Bussen, L. E. Cross and A. K. Goswami, J. Am. Ceram. 51 I. J. Clark and D. C. Sinclair, unpublished work. Soc., 1966, 49, 33. 52 H. Iwahara, T. Esaka, H. Uchida and N. Maeda, Solid State 6 G. Arlt, Ferroelectrics, 1990, 104, 217. Ionics, 1981, 3/4, 359. 7 A. Morell and J. C. Niepce, J. Mater. Educ., 1991, 13, 173. 53 H. Uchida, H. Yoshikawa and H. Iwahara, Solid State Ionics, 8 T. Takeuchi, K. Ado, T. Asai, H. Kageyama, Y. Saito, 1989, 34, 103. C. Masquelier and O. Nakamura, J. Am. Ceram. Soc., 1994, 77, 54 K. D. Kruer, Chem. Mater., 1996, 8, 610. 1665. 9 W. Kanzig, Phys. Rev., 1955, 98, 549. 10 K. Kinoshita and A. Yamaji, J. Appl. Phys., 1976, 47, 371. Paper 8/05756G J. Mater. Chem., 1999, 9, 83–91 91