IN-SITU XRD TO OPTIMIZE POWDER SYNTHESIS OF AURIVILLIUS PHASES

Transcription

1 Copyright (c)jcpds-international Centre for Diffraction Data 2002, Advances in X-ray Analysis, Volume IN-SITU XRD TO OPTIMIZE POWDER SYNTHESIS OF AURIVILLIUS PHASES Michael S. Haluska, Scott Speakman and Scott T. Misture New York State College of Ceramics at Alfred University, Alfred, NY ABSTRACT In situ x-ray diffraction was used to optimize the synthesis of Aurivillius structure ceramic powders. Powders were prepared via conventional solid-state synthesis and the Pechini polymer precursor method. The phases Bi 4 Ti 3 O 12, and Bi 2 Sr 2 Nb 2 TiO 12 were studied. The Bi 4 Ti 3 O 12 phase formed easily, regardless of processing route, between o C and phase purity was achieved by 700 o C when using the Pechini method and by 900 o C with the solid-state method. The Bi 2 Sr 2 Nb 2 TiO 12 phase formation was strongly dependent on processing route. Achieving phase purity via solid-state synthesis required over 100 hours of heat treatment time at elevated temperatures. In contrast, phase purity was achieved after about 2 hours at 900 o C when using the Pechini method. INTRODUCTION Aurivillius phases were first discovered around 50 years ago. 1 These phases are of great interest today as ferroelectrics and are potentially valuable fast ion conductors when modified to incorporate oxygen vacancies. The Aurivillius structure consists of a layered perovskite sandwich with bismuth oxide layers capping off the perovskite blocks. Because of the perovskite blocks, the structure may have one or more layers with the general composition of (Bi 2 O 2 )(A n-1 B n O 3n+1 ) where A is generally a large cation such as Bi 3+, Sr 2+, La 3+, etc. The B site is generally filled with a Ti 4+, Nb 5+, Ta 5+, or W +6. Complete replacement of the perovskite blocks with Brownmillerite blocks introduces a large population of oxygen vacancies. The Brownmillerite blocks contain oxygen vacancies on one in six of the oxygen sites, but are otherwise indistinguishable from perovskites. Naturally, aliovalent doping can also create oxygen vacancies. The goal of this work was to use high temperature x-ray diffraction (HTXRD) to study the reaction sequences during the synthesis of Aurivillius ceramics in-situ, and then use this information to prepare bulk powders. With a knowledge of the appropriate synthesis conditions, the ionic conductivity of the material can be studied and improved. These materials are known to exhibit fast ion conduction after undergoing an order-disorder transition. The discontinuous jump to fast ion conduction at high temperature is a result of a jump in the mobility of the oxygen ions by disordering of the oxygen sublattice. Once synthesized, the structure of the material can be determined using Reitveld refinements and with this structural information, the nature of oxygen conduction can be obtained.

2 This document was presented at the Denver X-ray Conference (DXC) on Applications of X-ray Analysis. Sponsored by the International Centre for Diffraction Data (ICDD). This document is provided by ICDD in cooperation with the authors and presenters of the DXC for the express purpose of educating the scientific community. All copyrights for the document are retained by ICDD. Usage is restricted for the purposes of education and scientific research. DXC Website ICDD Website -

3 Copyright (c)jcpds-international Centre for Diffraction Data 2002, Advances in X-ray Analysis, Volume EXPERIMENTAL Samples were prepared via solid-state synthesis and the Pechini method. Solid state synthesis consists of mixing bulk powders in stoichiometric ratios and then heating to obtain the desired product. The main advantage of solid-state synthesis is its simplicity in preparation. However, longer reaction times are frequently necessary due to the disparity in diffusion paths, which are highly dependent on the mixing process. The Pechini method 2 consists of first chelating the constituent cations with a carboxylic acid such as citric acid and then reacting the citric acid with an alcohol such as ethylene glycol in an esterification reaction to form a polymer. The benefit of this synthesis method lies in the fact that the cations are mixed on the atomic level, significantly shortening the diffusion distances, leading to shorter reaction times at lower temperatures. Solid-state samples were prepared by intimately mixing stoichiometric ratios of precursor powders. Bi 2 O 3, SrO, TiO 2, and Nb 2 O 5 (Alfa-Aesar) powders were weighed and mixed via ball milling in isopropanol. The SrO was first heated above 1000 o C to eliminate water. The mixed batch was then heated to various temperatures as determined by HTXRD analysis. Bulk samples were first pressed into pellets and heated in covered MgO crucibles. In order to achieve phase purity, the samples were reground, repelletized, and reheated several times. Samples were also prepared by the Pechini method. The following chemicals were used: Bi 2 O 3, SrNO 3, titanium butoxide, and NbCl 5 using the following general procedure. Bi 2 O 3 (and) SrNO 3 were dissolved in an aqueous solution of HNO 3 or HCl until a clear solution was obtained. Once dissolved, the appropriate amount of citric acid (anhydrous) was added and dissolved. Then, titanium butoxide was separately dissolved in ethylene glycol. It was determined that the titanium butoxide had to be added in this fashion, otherwise a precipitation reaction would occur when placed in contact with water. Finally, the titanium butoxide/ethylene glycol mixture was added drop-wise to the Bi 2 O 3 / SrNO 3 /H 2 O/ HNO 3 (HCl) solution until a clear solution of approximately ml total volume was obtained. When using NbCl 5, the powder had to be dissolved separately in HCl in order to prevent reaction. NbCl 5 has been shown to form a gel and it did so on several occasions in this system. Nonetheless, once complete dissolution was obtained in all systems, the solution was placed in a 50 ml Pyrex beaker and heated under constant stirring on a hot plate to 80 o C for approximately 3 hours and then the temperature was increased to approximately 130 o C for the duration. The citric acid acted as the chelating agent for the Bi, Sr, Nb, and Ti ions. Above 130 o C, an esterification reaction occurs between the citric acid and the ethylene glycol, ultimately yielding a hard, transparent solid polymer. The length of time required for complete reaction depended on how much water was used in the initial solution. Therefore, minimization of water content will decrease the time required. Generally, once the water left the system, approximately 3-5 hours were necessary for complete formation of the polymer. Originally when HNO 3 was used to dissolve the Bi 2 O 3, a strong exothermic reaction would occur simultaneously with the exit of the water yielding an unknown precipitate. This reaction did not occur when HCl was used. Once the polymer was formed, the beaker was placed in a furnace and ashed at 350 o C for several hours. The resultant shiny black foam residual was easily ground into a powder. It should be noted that this powder was not completely free from the polymer at this point. Complete decomposition does not occur until approximately 500 o C. 3

4 Copyright (c)jcpds-international Centre for Diffraction Data 2002, Advances in X-ray Analysis, Volume In-situ diffraction was performed using a custom diffraction furnace designed and built at Alfred that provides a stable specimen plane and can operate up to 1600 o C under controlled atmosphere. Cr radiation was used in conjunction with an mbraun position-sensitive detector to rapidly collect data in-situ. All samples were powder placed on thin alumina strip sample holders. Data analysis was completed using Jade+ version 5. 4 RESULTS AND DISCUSSION The Bi 4 Ti 3 O 12 phase was studied simply because it was the parent three-layer Aurivillius structure and it was the simplest case scenario. In this case, Bi 3+ occupies the Bi 2 O 2 planes as well as the perovskite A-site. It is one of the most widely studied Aurivillius phases. In the solid-state case, the initial reactants were Bi 2 O 3 (bismite, PDF# ) 5 and TiO 2 (anatase PDF ). As seen in Figure 1, the Bi 2 O 3 undergoes a reaction towards a cubic Bi 2 O 3 (PDF# ) by 600 o C. This phase persists until approximately 900 o C. The TiO 2 decomposes by 700 o C and corresponds with the growth of Bi 4 Ti 3 O 12. The desired phase Bi 4 Ti 3 O 12 forms as early as 600 o C and becomes phase pure on decomposition of the Bi 2 O Figure 1. HTXRD pattern of Bi 4 Ti 3 O 12 solid-state sample on alumina sample holder. Each scan represents a different temperature and temperature increases going into the page. Bi 4 Ti 3 O 12 samples prepared from the Pechini method originally start out as nearly pure bismuth metal although its pattern is not shown. The process of ashing the original polymer creates a reducing atmosphere that reduces the Bi 3+ to bismuth metal even under an oxygen atmosphere. However, the bismuth metal ultimately forms a bismuth oxide Bi 2 O 3 (PDF# 27-50) as seen in Figure 2. Note that this bismuth oxide transforms before 300 o C into the tetragonal Bi 20 TiO 32 phase (PDF# ) and furthermore, these two phases are both tetragonal and share very

5 Copyright (c)jcpds-international Centre for Diffraction Data 2002, Advances in X-ray Analysis, Volume similar lattice parameters. A non-stoichiometric bismuth oxide phase Bi 2 O 2.33 (PDF# 27-52) forms between o C and persists until approximately 700 o C. The growth of this phase is parasitic as it takes material from the Bi 20 TiO 32 phase. The desired phase, Bi 4 Ti 3 O 12, forms between o C from the Bi 20 TiO 32 and Bi 2 O 2.33 and reaches phase purity by 700 o C. It is interesting to note that the peaks for titanium(oxide) are not visible in this case (figure 3). It is possible that the peaks at approximately 38 o 2θ are derived from titanium oxide (most likely TiO 2 ) but statistically it is not clear due to the low signal to noise ratio. Figure 2. HTXRD patterns of Bi 4 Ti 3 O 12 prepared using the Pechini sample. The powder sample was placed on alumina sample holder. Each scan represents a different temperature as indicated in the figure. The Bi 2 Sr 2 Nb 2 TiO 12 phase was also studied. In this case, Sr 2+ substitutes for the Bi 3+ on the perovskite A-site and Nb 5+ substitutes for some Ti 4+ on the perovskite B-site. 6 The HTXRD pattern for this analysis was quite complex. Due to the large number of secondary phases that formed, it was not possible to identify with exact certainty all of the phases present. However, two major phases did form; SrBi 2 Nb 2 O 9 (PDF# ), a two-layer Aurivillius phase 7, and Sr 5 Nb 4 O 15 (PDF# ). Since it was estimated that more time might be necessary for reaction completion, a separate high temperature measurement was performed on this same powder and it can be seen in Figure 3. The temperature was held at 1100 o C and scans were made every two hours. The final scan represents 8 hours at temperature. It is clear from Figure 3 that no progression was made towards reaction completion. A bulk sample was separately prepared and heated as shown in Figure 4 to investigate the effect of a much longer heat treatment. A pelletized sample was heated in a covered MgO crucible for the times and temperatures listed. After several intermittent regrindings, over 100 hours of heat treatment was required to achieve phase purity. Structural characterization of this phase is currently underway.

6 Copyright (c)jcpds-international Centre for Diffraction Data 2002, Advances in X-ray Analysis, Volume Figure 3. HTXRD pattern of Bi 2 Sr 2 Nb 2 TiO 12 solid state held at 1100 o C for 2-hour intervals. Powdered sample placed on alumina sample strip. Figure 4. X-ray diffraction patterns of Bi 2 Sr 2 Nb 2 TiO 12 bulk samples heated at different intervals and temperatures. Each pattern represents a bulk sample measured after cooling.

7 Copyright (c)jcpds-international Centre for Diffraction Data 2002, Advances in X-ray Analysis, Volume Based on the results of the solid-state analysis, it was expected that the Pechini method, involving atomic scale mixing, would provide faster reaction times for preparing Bi 2 Sr 2 Nb 2 TiO 12. However, a processing problem ensued with the gelation of NbCl 5 on heating of the precursor solution. It should be noted that any precipitation means that the atomic scale mixing for that element is lost. Regardless, the sample was heated to 130 o C for the requisite time until the polymer formed. Figure 5 shows the high temperature diffraction pattern of a sample that was previously ashed to 350 o C. It was unfortunate that in this system, the peak resolution was extremely poor and the initial reaction sequence was not understood. It appears that the desired phase formed between o C and appeared to grow strongly between o C. Phase purity of Bi 2 Sr 2 Nb 2 TiO 12 appeared to be complete by 900 o C, however the peak widths were still extremely broad. Further heating to 1200 o C did not appear to affect the peak widths. On the contrary, it appeared that the peak width of the 100% peak broadened and some peak splitting occurred. This phenomenon appeared in each case as the sample was heated above 950 o C. The peak width of the 100% peak (~45 o 2θ) broadened and the peak at 49 o 2θ split into two peaks. One possible explanation was that the material was decomposing into two separate phases. On a separate note, the Aurivillius phases are well known to exhibit stacking faults. 8 Any such stacking fault would cause the peak widths to broaden. A bulk sample was prepared and held in the furnace at 1000 o C for several hours in attempt to further crystallize the material and therefore increase the peak sharpness but to no avail. It is possible that controlling the gelation of the NbCl 5 would help in the matter of the peak broadening. It is also possible that much longer reaction times are necessary to achieve a more crystalline specimen. TEM analysis will be performed to determine if stacking faults are present or if two phases crystallize. Figure 5. HTXRD patterns of Bi 2 Sr 2 Nb 2 TiO 12 Pechini sample. Powder sample placed on alumina sample strip. Each measurement represents a different temperature as noted in the figure.

8 Copyright (c)jcpds-international Centre for Diffraction Data 2002, Advances in X-ray Analysis, Volume SUMMARY The use of HTXRD was extremely helpful in determining the processing conditions necessary for synthesis of the Aurivillius phases studied. The simplest of the phases, Bi 4 Ti 3 O 12, formed rapidly regardless of processing method. The Bi 2 Sr 2 Nb 2 TiO 12 phase proved to be difficult to form under any processing conditions. The reaction of four cations in the Bi 2 Sr 2 Nb 2 TiO 12 system proved to be a complicating factor, which may have hindered the reaction. While it was clear that phase purity was achieved after long times in the solid state method, it remained uncertain as to whether the sample prepared via the Pechini method achieved phase purity due to the extremely broad peaks. The broadness of the peaks in this case indicates a structural defect; possibly stacking faults, which require further characterization and possibly a different synthesis technique. REFERENCES 1 B. Aurivillius, Mixed Bismuth Oxides with Layer Lattices, Arkiv For Kemi, 1, # , United States Patent # 3,330,697. M. Pechini, July 11, P. Duran, F. Capel, C. Moure, M. Villegas, et al. Processing and Dielectric Properties of the Mixed-layer Bismuth Titanate Niobate Bi 7 Ti 4 NbO 21 by the Metal-organic Precursor Synthesis Method, Journal of the European Ceramic Society, 21, 1-8, Jade+ version 5. Materials Data Inc. software Powder diffraction file. 6 C. Hervoches, P. Lightfoot, Cation Disorder in Three-layer Aurivillius Phases: Structural Studies of Bi 2-x Sr 2+x Ti 1- xnb 2+x O 12 (0 < x < 0.8) and Bi 4-x La x Ti 3 O 12 (x = 1 and 2), Journal of Solid State Chemistry, 153, 66-73, S. Blake, M. Falconer, M. McCreedy, P. Lightfoot, Cation Disorder in Ferroelectric Aurivillius Phases of the Type Bi 2 ANb 2 O 9 (A = Ba, Sr, Ca), Journal of Materials Chemistry, 7, , S. Horiuchi, T. Kikuchi, M. Goto, Structure Determination of a Mixed-Layer Bismuth Titanate, Bi 7 Ti 4 NbO 21, by Super-High-Resolution Electron Microscopy, Acta Crys. A33, , 1977.