In situ high temperature neutron diffraction study of a liquid-phase-sintered alumina ceramic

Abstract

Neutron powder diffraction has been extensively used to study mineral and ceramic phases over the past decade [1]. Central to neuron diffraction (ND) analysis is the attribute that measurements provide the bulk specimen character of a material. This is important in quantitatively extracting phase composition information. Whilst many diffraction studies of the nature of phases, the crystallization and phase transformations of ceramic systems have been limited to room temperature measurements, little is known of the material characteristics and behaviour at speci®c temperatures. Most diffraction measurements are conducted at ambient or static temperature conditions. When studying sintering processes, measurements are typically collected after specimens have been heat-treated and then cooled. Such data may yield misleading information particularly in relation to non-equilibrium phases. Hence, dynamic measurements are clearly preferable as a direct means of con®rming sintering processes. We present here a real time study of the phase evolution, dissolution and crystallization behaviour in a liquidphase-sintered (LPS) alumina. Powder neutron diffraction data of initially green ceramic specimens was collected in real-time at temperatures up to 1400 8C. The dynamics of the phase changes which occur on heating and the crystallization of minor phase components on cooling the LPS alumina are explored. Emphasis is placed on differences observed in the ND patterns with the crystallization of minor phases subjected to three distinct cooling treatments from the sintering temperature. Alumina powders (A13, Alcoa of Australia Ltd., Kwinana, Western Australia) doped with various sintering aids were used as starting materials for the synthesis of liquid-phase-sintered (LPS) alumina ceramics. Batches were prepared comprising a mixture of 84 wt % alumina with 9 wt % kaolinite, 5 wt % talc and 2 wt % calcite (supplied by Commercial Minerals, Kewdale, Western Australia) powders. Approximately 10 g of powder was placed in a cylindrical steel die assembly and uniaxially dry-pressed at 150 MPa to form compacts of dimensions 19 mm diameter 320 mm thickness. Hot stage ND measurements were performed using the High Flux Australian Reactor (HIFAR) neutron source operated by the Australian Nuclear Science and Technology Organisation (ANSTO) at Lucas Heights, NSW, Australia [2]. Three greenbody compacts were stacked and then glued together with a high-temperature alumina-based cement (Cotronics) to achieve a total specimen height of 60 mm in order to maximize specimen irradiation. The specimens were lowered into a custom-made MoSi2 resistance-heated vertical alumina-tube furnace (Model P1750, Ceramic Engineering) and carefully aligned with the incident neutron beam using a cadmium ` cross''. ND patterns were collected using Debye±Scherrer optics with 24 He3 detectors, on the medium-resolution powder diffractometer (MRPD) at intervals of 0.18 over the 2e range 08y1388 with e ˆ 0:1664 nm and a ®xedcount monitor. The temperatures at which data were obtained ranged from 20 8C to 1400 8C and the heating and cooling rates used were controlled and monitored automatically (Model ITC 502, Oxford Instruments) to simulate the heat-treatment regimes employed in the fabrication of these materials [3]. The nature of the crystalline phases in the specimens were identi®ed by search-match analysis using the International Centre for Diffraction Data (ICDD) database. Conventional laboratory X-ray powder diffraction was used in conjunction with neutron data analysis for phase identi®cation. Three dynamic ND experiments were performed. The compacts were heated to 1400 8C and held for 4 h, at which time two patterns were collected (counting time 2 h each), and then cooled at three different rates whilst continuously collecting diffraction patterns such that counting statistics and pattern information were not degraded. The different cooling schedules were: (i) rapid cooling (the fastest cooling rate) depending on the natural cooling of the furnace 30 8C miny1 (non-linear rate) to room temperature; (ii) slow cooling at a rate of 3 8C miny1 to room temperature; and (iii) rapid cooling to 1050 8C, then annealing for 18 h before natural cooling to room temperature. The ND patterns displayed, are in several instances slightly more noisy due to the