Thermal and mechanical properties of Mg2Si1-xSnx (x = 0 – 1) for thermoelectric generators
Thermoelectric technology is a good option for electricity generation due to its capacity to turn waste heat directly into employable electrical energy. Thermoelectric modules are the basis of this technology and are fabricated from doped n- and p-type semiconductors. Mg2(Si,Sn) thermoelectric material is one of the top candidates for module ... assembly due to its good thermoelectric properties coupled with low density and cost. The low toxicity and high availability of the precursor elements give this material system crucial advantages in comparison with other competitors. Thermoelectric module operation requires a temperature difference, which inevitably causes differential thermal expansion within a module. Such an expansion in a device composed of different materials with different expansion coefficients could lead to failure due to stress-induced fracture, posing a serious threat to reliability and applicability of thermoelectric modules. It is therefore important for module design to take into account the different thermal and mechanical properties of the materials involved in the assembly. Most of the research on thermoelectric applications, however, is focused on the optimization of the thermoelectric performance of the materials. Other properties like elastic modulus, hardness and coefficient of thermal expansion are studied with substantially lower intensity. This thesis aims at filling the gap of missing information regarding the mechanical and thermal properties for the solid solutions Mg2Si1-xSnx with x = 0 – 1. This work starts with hardness measurement, Vickers indentations were performed on the sintered pellets to identify the effect of Sn content in Mg2(Si,Sn) on the hardness exhibited by the material. Increasing the amount of Sn in the solid solution decreases the hardness values in a linear relationship. Mg2Si has the highest value at 5 GPa and Mg2Sn the lowest at 2 GPa. The fracture toughness of the studied samples did not, however, follow the same trend, as the material Mg2Si0.6Sn0.4 exhibited the highest value. It was found that Si-rich regions in the microstructure left over from the synthesis and pressing processes were strengthening the material by adding interfaces, which deflected or otherwise impeded the growth of the cracks produced by indentation. The next step towards filling the gap in missing information was to characterize the elastic moduli of the solid solution series Mg2Si1-xSnx with x = 0 – 1. Two non-destructive characterization methods were employed and compared, the Resonant Ultrasound Spectroscopy and the Impulse Excitation Method. This work innovates in the parallel measurement and comparison between the results provided by both of these techniques. The difference between the measurement results is below 9%, which suggests that using both techniques interchangeably is possible. The main differences between the techniques are the sample size required for testing, as well as the ease at which high temperature measurements can be implemented. This work presents the first ever report of Young’s modulus of Mg2(Si,Sn) as a function of composition and temperature, finding a linear dependence of both. Using these results, a bilinear dependence was proposed to predict the Young’s modulus of any material within the solid solution and at any temperature between 300 K and 623 K. Joining a fast quantification method to estimate the local composition using back-scattered electron images to the bilinear equation, the effective Young’s modulus of several samples was estimated. For this estimation both the Voigt and Reuss approximations for a composite material were used. The results show that the composite material approach and the bilinear equation can be used to accurately predict the effective elastic modulus of typical, not completely homogenized, Mg2(Si,Sn) material. To test the effect of doping species on the thermal and mechanical properties of Mg2(Si,Sn), the materials Mg2Si0.3Sn0.665Bi0.035 and Mg1.97Li0.03Si0.3Sn0.7 were compared to undoped Mg2Si0.3Sn0.7 and low doped Mg2Si0.3Sn0.6925Bi0.0075. This information is crucial for accurate module design as any possible effect has not been identified before. Room and high temperature Young’s modulus was measured for all the mentioned compositions. All of them exhibited a linear behavior, albeit with Bi containing samples having different slopes. Both materials of interest show, however, very similar values at application temperatures. The coefficient of thermal expansion for all the aforementioned samples was measured from room temperature to 440 °C. It was proposed to use a linear fit and extrapolation to describe the temperature dependent thermal expansion of the material instead of the mean value usually given in literature. When the equation obtained from the extrapolation is used to estimate the room temperature value, the comparison to the mean value results in a difference <3%. This work concludes with the simulation of a thermoelectric uni-couple using Finite Element Modelling. For this simulation, temperature dependent data presented in this work is used and compared to modeling results based on constant values. The stress distribution is described using three main stress components, the von Mises stress, the principal stress 1 and the shear stress along the contact surface. A comparison between constant values with temperature dependent data for Mg2(Si,Sn) shows that using constant room temperature or temperature averaged values gives similar results as full temperature dependent calculations. However, when only one of the main variables, Young’s modulus or coefficient of thermal expansion, is employed with the correct temperature dependence, the stress values can be off by more than 10%.