Abstract—A modified Genetic Algorithm (GA) have been used to obtain the optimum dimensions for a large area pn-junction thermoelectric power generator device. Optimization routine allows for obtaining maximum output thermoelectric power. The results which obtained by using computer simulations has confirmed better performance in compare to those obtained from design in a vast range of temperatures. Index Terms—Thermoelectric, device, pn-junction, large area, Genetic Algorithm. I. INTRODUCTION Thermoelectric power devices provide attractive solutions for direct conversion of heat to electricity. Advantages of these devices include simplicity, low cost and lack of moving parts which results in a long lifetime and practically no need of device maintenance. Unfortunately, state of the art devices suffer from their low efficiency and thus, they normally employed in situations where low maintenance is critical such as satellites and spacecraft systems. Fig. 1 shows schematic of a simple semiconductor thermoelectric power generator. p-doped and an n-doped semiconductor are electrically connected at the heated side using a metallic joint. The temperature gradient along the semiconductors causes a flux of the available carriers from the higher temperature end to the lower temperature end. For this reason, a voltage drop appears across width and Carrier transmission of the two lower temperature ends. Novel, efficient and highly reliable thermoelectric structures and materials suitable for operation in a wide range of temperatures are obviously needed. Recently, several approaches such as heterostructures, nanowires and superlattices using novel complex materials for improving the efficiency and power output of thermoelectric generators have been investigated with limited success [1-4]. Wagner et al. [5] proposed use of a large area Si/SiGe pn-junctions to provide higher efficiencies at lower production costs in thermoelectric power generation. In this paper, we employed a Si large area pn-junction thermoelectric module as shown in Fig. 2. By using this structure, the strong correlation between thermal and electrical properties can be minimized .i.e. both effects can be Manuscript received July 19; revised August 30, 2012. The authors are with the Department of Electrical and Computer Engineering, Islamic Azad University, Shahrood branch, Shahrood, Iran (e-mail: Ali_eftekhary2003@yahoo.com; mfathi@ut.ac.ir; dramalavi_gharah@yahoo.com). optimized independently from each other [6]. In this study, ATLAS Device Simulation Framework in Silvaco software has used to run simulations of this thermoelectric device [7] and a modified genetic algorithm [8] in MATLAB is applied in combination with ATLAS for dimensional optimization of this structure. Fig. 1. Principle sketch of a conventional thermoelectric generator with applied temperature gradient. A p-doped and an n-doped semiconductor are connected on the higher temperature end of the device. Electric contacts are applied on the lower temperature end of the device. Fig. 2. Large area pn-junction with temperature gradient II. MODEL The structure of the thermoelectric device employed in this study is shown in fig. 2. First, a temperature gradient is applied along the pn-junction which causes a flux of both carrier types from the high temperature to the lower temperature regions .This is due to the fact that thermally generated electron-hole pairs are separated by the built-in potential gradient of the pn-junction. Band gap changes with the temperature as: ( ) () 2 0 - (1) T E T E g g T α β = + Design of a Large Area Pn-Junction Si Thermoelectric Device via Genetic Algorithm Ali Eftekhari, Morteza Fathi-Pour, and Abdorreza Alavi Gharahbagh International Journal of Machine Learning and Computing, Vol. 2, No. 5, October 2012 658 10.7763/IJMLC.2012.V2.209