A Novel Approach for the Development of Accurate ab inito Potential Energy Surfaces for Atomistic Simulations of MEMS Applications

开发用于 MEMS 应用原子仿真的精确从头势能面的新方法

• 批准号: 0457663
• 负责人: Ranga Komanduri
• 金额: --
• 依托单位: Oklahoma State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2005
• 资助国家: 美国
• 起止时间: 2005-09-01 至 2009-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0457663&HistoricalAwards=false
• 关键词: Novel; Approach; Development; Accurate; ab

The objective of the proposed research is to present an integrated approach for conducting simulations at the atomistic level using ab initio quantum mechanical potential-energy surfaces and force fields computed at high-level with extended basis sets. The approach involves (1) electronic structure calculations (using GAUSSIAN 2003 software) of non-equilibrium clusters within the cut-off radius of a material, (2) sampling of the subset of the system configuration space that is important in the dynamics using MD methods augmented with novelty sampling procedures, and (3) accurate interpolation between the computed points using neural network (NN) with early stopping and regularization methods employed to improve network performance. The accuracy of the method will be tested and validated in wide range of manufacturing processing including nanometric cutting, tribology, and material testing. This approach will be extended this to other applications involving carbon nanotubes and synthesis of microcrystalline diamond by chemical vapor deposition (CVD).Molecular dynamics (MD) and Monte Carlo (MC) simulations can be applied for modeling a wide range of manufacturing processes including ultraprecision machining, grinding, and polishing as well as materials testing, and tribology. The method proposed is a generalized method in that it can be applied for a wide range of fields in engineering as well as in basic chemistry, physics, biology, and other scientific fields of endeavor. This problem will be addressed by an interdisciplinary group of researchers from chemistry, electrical engineering, and mechanical engineering, math and materials science background working with graduate and undergraduate (through NSF REU program) students from these disciplines. Research conducted will be fully integrated with the graduate level education. We intend to interact with industry and national laboratories involved in this area. Attempts will be made to recruit U.S. born women, minority, and physically impaired students to work on this project.

提出的研究目标是提出一种综合方法，在原子水平上使用从头算量子力学势能面和扩展基集在高级计算的力场进行模拟。该方法包括(1)在材料的截止半径内对非平衡簇进行电子结构计算（使用高斯2003软件），(2)使用带有新颖性采样程序的MD方法对系统配置空间子集进行采样，这在动力学中很重要，以及(3)使用神经网络（NN）在计算点之间进行精确插值，并采用早期停止和正则化方法来提高网络性能。该方法的准确性将在广泛的制造加工中进行测试和验证，包括纳米切削，摩擦学和材料测试。这种方法将扩展到其他应用，包括碳纳米管和化学气相沉积（CVD）合成微晶金刚石。分子动力学（MD）和蒙特卡罗（MC）模拟可以应用于各种制造过程的建模，包括超精密加工，磨削和抛光以及材料测试和摩擦学。所提出的方法是一种广义的方法，可以广泛应用于工程领域以及基础化学、物理、生物和其他科学领域。这个问题将由来自化学、电气工程、机械工程、数学和材料科学背景的跨学科研究人员与来自这些学科的研究生和本科生（通过NSF REU项目）一起解决。所进行的研究将与研究生水平的教育充分结合。我们打算与参与这一领域的工业和国家实验室进行互动。将尝试招募在美国出生的妇女、少数民族和身体有缺陷的学生参与这个项目。