JOURNAL OF MATERIALS SCIENCE 29 (1994) 6458-6462 High-temperature creep of yttrium-aluminium garnet single crystals S. KARATO, Z. WANG, K. FUJINO Department of Geology and Geophysics, University of Minnesota, Minneapolis, MN 55455, USA High-temperature creep in single crystals of Y3AI~O12 (YAG) was studied by constant strain- rate compression tests. The creep resistance of YAG is very high: a stress of ~300 MPa is needed to deform at a strain rate of 10 -6 (s -1) at a temperature as high as 1900 K (~0.84 Trn, (melting temperature)). YAG deforms using the (1 1 1~ {1 10} slip systems following a power law with stress exponent n ,-~ 3 and activation energy E* ~ 720 kJ mo1-1. However, a small dependence of n on temperature and of E* on stress was observed. This stress-dependence of activation energy combined with the observed dislocation microstructures suggests that the high creep resistance of YAG is due to the difficulty of dislocation glide as opposed to the difficulty of climb. Present dislocation creep data are compared with diffusion creep data and a deformation mechanism map is constructed. Large transition stresses (2-3 GPa for 1 0 ~tm grain size) are predicted, implying that deformation of most fine-grained YAG will occur by diffusion creep. 1. Introduction High creep resistance of garnets has been well docu- mented [1]. Among the various materials that crystallize into garnet structure, Y3AtsO12 (YAG) has the highest melting point (Tin = 2273 K) and hence is likely to be the most resistant to plastic deformation at high temperatures [2-4]. However, previous studies have several limitations and high-temperature creep in YAG has not been well characterized. The limitations include: (i) very small strains (< 0.3%) in creep tests [3] leading to large uncertainties in creep law para- meters, and (ii) the absence of detailed microstructural studies [2-4] causing uncertainties in determining the slip systems and in the interpretation of the mechan- ical data. This paper presents a first comprehensive mech- anical data set on high-temperature creep of single crystals of YAG together with microstructural obser- vations of deformed specimens using etch pit tech- nique and transmission electron microscopy (TEM). The results are compared with those on other oxide garnets [1, 5] to better understand the deformation mechanisms in single crystals, and also With data on diffusion creep [6] to estimate the relative importance of dislocation and diffusion creep. 2. Materials and methods A single crystal of Czhokralsky-grown YAG was ob- tained from MIMports (California); it was ~ 2 cm in diameter and ~ 5 cm long and of more than 99.999% purity. It was oriented using the Laue back reflection technique within 1~ to the ( 1 0 0 ~ orientation. Paral- lelpiped samples (,-~ 2 • 2 x 5 mm 3) were cut from the same single crystals and both ends were polished with a diamond slurry down to 0.3 lam. Compression tests were carried out using an MTS servo-controlled test- ing machine in air in a MoSi a furnace. The pistons are made of SiC and the displacement of the specimen was measured using a custom-built in situ extensometer by which the relative movement of two sets of SiC rods which touch the bottom of the top piston and the top of the bottom piston respectively, is measured (Fig. 1 [7]). The pistons showed no appreciable deformation up to the maximum temperature (1963 K) and stress (~ 300 MPa) used in this study. However, it was found that YAG reacts with SiC in air at high temperatures, and therefore we inserted corundum (with orientation parallel to the c-axis) in between the YAG and the SiC. To minimize the deformation of corundum, thin discs of YAG were inserted in between the specimens and the corundum platens, which distributes the load on the corundum platens (Fig. 1). The deformation of corundum was found to be significantly less than that of YAG. In most of the experiments a constant strain-rate (displacement-rate) mode was adopted, in which the change in load (or stress) was measured as a function of strain (time). The strain rate ranged from 0.3 x 10 -6 to 4.5 x 10 -6 S -1. Temperatures ranged from 1853 to 1963 K (TIT m = 0.82-0.86). The compression tests were made along the (10 0) orientation. In this ori- entation, of the two possible types of dislocations [1] those with Burgers vectors b = (1/2)( 1 1 1 ) are activ- ated but those with b = (1 0 0) are not. After the constant strain-rate tests the specimens were quenched under load and recovered for micro- structural observations using an optical microscope 6458 0022-2461 9 1994 Chapman & Hall