Modeling of the degradation of TiB2 mechanical properties by residual stresses and liquid Al penetration along grain boundaries
Abstract
Titanium diboride (TiB2) is a potential candidate as cathode material in the electrolysis of aluminum. The employment of such material, instead of carbon, could induce considerable savings of energy during the production of aluminum; therefore the interest towards this material has been increasing in the last decades. Titanium diboride is a relatively strong ceramic material, characterized by high wetting ability by liquid Al, low solubility in liquid Al and cryolite, and high electrical conductivity. All these properties are advantageous for the above mentioned application. On the contrary, some limiting factors prevented so far the use of TiB2 in the Al production industry. When in contact with liquid Al at high temperatures, TiB2 is subjected to penetration by liquid Al along TiB2 grain boundaries and cracking at micro-level; these phenomena reduce the material strength and induce failure. The failure mechanism in TiB2 has been identified as probably being similar to the one of Liquid Metal Embrittlement (LME) observed in several ‘solid metal-liquid metal’ systems. It is thus very important to investigate the material behavior, before and after exposure to liquid Al, and understand the mechanisms of degradation.
The present study is an investigation of the behavior of TiB2 at micro-level by Finite Elements models, in connection with the microstructure properties. It consists of three steps where different FE models were employed in order to describe different aspects of the degradation problem.
The first paper of this thesis work concerned the investigation of the residual stresses effect on the grain boundary microcraking. Residual stresses along grain boundary usually build up during the last production stage of ceramic materials, when they are cooled down from high temperatures (1000-1500°C) to room temperature. This is due to thermal anisotropy at grain level which is a typical characteristic of ceramics. Therefore the model was chosen as a combination of an existing model, Clarke’s model (1980), to take into account thermal anisotropy, and cohesive zone models, to represent the grain boundary and its microcracking. The stress field and the grain boundary microcracking as a result of residual stresses were analyzed by varying grain size, grain boundary energy density, and sintering temperature. The residual stress distribution depends only on parameters of the microstructure and the sintering process, while residual-stress-induced microcracking depends strongly on the grain size. In addition a quantitative relation among grain boundary energy density, temperature variation, and microcrack length was established, so that critical combinations of grain size and grain boundary energy density can be identified. No measurements of TiB2 grain boundary energy were available; nonetheless, the calculated critical energy density for microcracking induced by residual stresses was compared to the macroscopic fracture toughness measurements from literature (Munro, 2000).
In the second paper, the effects of the exposure of TiB2 specimens to liquid Al were investigated. Three types of TiB2, differing by grain size, density and secondary phases, were studied. The characterization of the materials’ microstructure and the exposure of four-point bending bars to liquid Al were carried out in a parallel PhD project. The TiB2 material with grain size 7.6 μm was also characterized from a mechanical point of view. Micro-indentation and fourpoint bending tests on unexposed and exposed bars at room temperature, were performed in cooperation with the PhD student colleague Morten Jensen. Reduced hardness, stiffness, and flexural strength, were observed with increasing penetration depth. A FE model was created in order to simulate the reduced stiffness observed in experimental measurements. The model does not include any thermal anisotropy and no residual stresses were taken into account. Instead, the presence of Al along grain boundary was modeled using elements with Al properties. The parameters mainly affecting the reduction of stiffness in the penetrated material are the equivalent volume percentage of Al present in the specimen, and the penetration depth. The grain boundary thickness was found to be less relevant.
The work presented in the last paper consisted in the creation of a FE model which can describe the failure strength of 4-pt. bending specimens which are subjected to residual stresses and external load. Moreover both pristine and Alinfiltrated materials were analyzed and the reduction of grain boundary energy due to Al penetration was well described by the model. The effect of grain size, grain orientation and number of grains, was investigated. The employment of cohesive models to model the grain boundary and its cracking behavior caused some convergence problem; a regularization technique was used and the effect of the main parameter (the viscosity parameter) on the results was evaluated for the present case.