Data driven parametric modelling and multi response optimization of ECAE processed Bi85Sb15 thermoelectric materials

The advancement of high-performance thermoelectric (TE) materials is sometimes hindered by the inadequate capacity of conventional experimental methods to clarify complicated processing–property interactions. This study proposed a data-driven computational–experimental framework that integrates Response Surface Methodology (RSM), based on a Central Composite Design (CCD), with a Multi-Objective Genetic Algorithm (MOGA). This hybrid technique proposed to optimize the thermoelectric performance of Bi85Sb15 alloys processed via Equal Channel Angular Extrusion (ECAE). Three critical parameters are measurement temperature, processing temperature, and number of ECAE passes were systematically analysed to quantify their influence on electrical resistivity (ρ), thermal conductivity (k), Seebeck coefficient (S), power factor (PF), and figure-of-merit (ZT). RSM was used to generate reliable surrogate models, and MOGA was utilized to quickly find the best trade-offs between the different TE responses. The optimal solution shows that eight ECAE passes at a processing temperature of 73 °C give a maximum ZT of 0.32, which is 12% better than what was measured in the experiment. These results show the effectiveness of the proposed data-driven process optimization approach (RSM–MOGA) for accelerating the design and optimization of Bi–Sb thermoelectric material. It also provides a basic structure for making new TE materials with superior performance. The suggested data-driven optimization methodology supports global goals in clean energy, energy efficiency, and engineering for sustainable materials. This work helps improve waste-heat recovery, solid-state cooling, and decentralized energy solutions by making it easier to rapidly design TE materials with effective ZT.