Determination of non-symmetric secondary creep behaviour of ceramics by residual stress measurements using neutron diffractometry

Abstract

e.g. c~ ~ 7 was reported for hot-pressed silicon nitrides [31. In Equation 2 the same exponent n for tension and compression was assumed. Chuang [4] used a relationship with different n-values. The coefficient c~ can be determined directly from creep tests using tensile loads. Unfortunately, such tests are not easy to perform. An alternative is given by bending creep tests on a specimen of trapezoidal cross-section. The stationary creep rates are expected to be different if the narrow side is subject to tensile or to compressive stresses. Formulae for the evaluation of ~ from the observed creep rates can be taken from [2]. Another method of determining ~ is based on measurements of the dimensional changes of a bending bar produced by the creep test as described in [5]. Both methods yield rather rough values of a. In this letter we present a method which is based on the determination of the residual stress distribution in a crept bending bar by neutron stress analysis. This method was found to be rather simple and yields a-values of superior precision. Creep tests were performed on specimens with dimensions 3 .5mm x 4 .5mm x 45ram made from the following materials: an AlzO3 ceramic containing 2.7% SiO2, 1.3% MgO, 0.02% CaO and 0.02% Fe203 (V38, Feldmiihle AG, Plochingen, FRG) and hotpressed silicon nitride (HPSN), containing 2.5% MgO. Due to the additives, a grain boundary glass phase is obtained in these materials after sintering. This is why creep can occur at relatively low temperatures. For these materials the Nor ton exponent was determined to ben = 2.25 (A1203) [5] and n = 2 (HPSN) [6]. The bending creep tests were performed with a load corresponding to an initial outer fibre bending stress of a0 = 60MPa (A1203) and o 0 = 135MPa (HPSN). The specimens crept for 100 h at 1100°C (A1203) and 1200° C (HPSN) were cooled under load and unloaded at room temperature. High-resolution neutron diffractometers were used to determine the residual stress distribution in the samples after unloading. Neutron stress analysis is analogous to the well-known method of X-ray stress analysis, i.e. it is based on precise measurements of lattice strain distributions. This technique has been described in previous publications [7-9], and therefore we confine ourselves here to giving some important details. Preliminary measurements were performed on a triple-axis spectrometer located at the Melusine Research Reactor, Grenoble, France. The final data sets were taken on the high-resolution diffractometers D I A at the high-flux reactor of the ILL, Grenoble. Slits in the incident and the diffracted beam restricted the irradiated volume to 0.4ram x 3 mm x 3mm, with the smallest dimension in the y-direction (see Fig. 1). For the A1203 ceramic the d-spacings were measured on the { 1 1 3} planes, whereas for HPSN the {1 1 1} and {2 1 1} planes were chosen. With a neutron wavelength of 2 = 0.298 nm the corresponding Bragg angles 20 were between 80 and 105 ° . The lattice strains were determined in the central part of the samples for the strain components G and ez as a function of y. Under the assumption that the residual stresses G are uniaxial and directed along the x-direction, the strain components investigated are related to the residual stresses G by