Parallel Molecular Dynamic Simulation Study on Nanoscale Thermal Conductivity

• Main Authors: Kenny Lin, 林培凱
• Other Authors: Rong-Tai Hong
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/06319552292371108988

碩士 === 中華技術學院 === 機電光工程研究所碩士班 === 97 === The thermal conductivity is studied by using equilibrium molecular dynamics simulation (EMD) with nanoscale with bulk Silicon. The parameters are investigated including the different crystalline plane, (100), (110) and (111) and different temperatures. The Tersoff potential is used is this work which is local environment potential and consider the bond order. Due to considering of the three-body f potential function in MD, the computational time will be increased in calculating the interacting force. Therefore the parallel computation is needed for this potential function. A parallel program is developed in shared-memory multi-core PC with OpenMP. Based upon the results of heat current autocorrelation, it decays over a time range which increases as the system temperature is lowered. The thermal conductivity with different crystal plane is changed with temperature as T-1 as the temperature is above Debye temperature. Furthermore, the overall parallel efficiency is increased by increasing the number of simulating atoms. The code is executed on Intel Xeon and IBM p9119-595 respectively. The efficiency will be decreased by increasing the number of processors and atoms. In this work, the parallel efficiency remains 80% by using 16 processors whereas the efficiency is near ideal with fewer processors.