ITR/AP: Large-Scale Quantum Mechanical Molecular Dynamics Simulations: Challenges, New Directions, and Applications to Carbon-Based Nanostructures

该奖项是提交给信息技术研究倡议的提案的结果。 计算材料科学的发展依赖于有效和可靠的大规模量子力学分子动力学（MD）模拟计算方法的发展。 线性标度或阶数N（O（N））方法已经被开发，目的是克服在基于量子力学的模拟中计算总能量和原子力时与N3标度相关的瓶颈。 尽管取得了这些进展，O（N）- ab initio-MD仍然局限于包含相对较少原子数的系统，因为ab initio MD中与自洽计算相关的开销。 另一方面，基于双中心紧束缚（TB）哈密顿量的MD方案比O（N）-从头算MD方法快几个数量级。 然而，已经发现它们在电荷转移或键断裂/重新键合起重要作用的系统中是不令人满意的。 因此，迫切需要开发可转移的半经验哈密顿量，其将上级于传统的双中心TB哈密顿量，但将包括从头算哈密顿量的所有基本成分而不需要计算过多。 采用这种哈密顿量的O（N）-MD方法可以精确地预测大尺寸复杂系统的性质，本研究将发展一种通用的方法来构造这种可靠的、可传递的材料半经验哈密顿量（金属或半导体）在原子轨道线性组合（LCAO）的框架中，该框架明确地包括自洽（SC）确定的电荷转移和环境依赖（艾德）多中心互动。 SCED-LCAO哈密顿量将在之前开发的O（N）-MD方案中实现。 利用O（N）/SCED-LCAO-MD作为模拟工具，我们将研究碳多壁纳米管和碳纳米棒的性质，并评估它们作为分子尺度器件组件的潜力。 具体而言，我们将研究以下三个当前感兴趣的项目：（1）碳多壁纳米管（MWNT）的能量学、结构、电子、力学和振动性质，（2）金属电极与MWNT之间的接触性质的研究，和（3）碳纳米棒（CNR）的能量学、结构、电子、力学和振动性质。 项目（2）将与路易斯维尔的Alphenar教授的实验组合作完成，而项目（3）将与中央研究院（台湾）的陈教授的实验组合作完成。%**

This award is the result of a proposal submitted to the Information Technology Research initiative. Advances in computational materials science depend on the development of efficient and reliable computational methods for large-scale quantum mechanical molecular dynamics (MD) simulations. Linear-scaling or order-N (O(N)) methods have been developed with the aim of overcoming the bottleneck associated with the N3-scaling in the computation of the total energy and atomic forces in quantum mechanics-based simulations. Despite this progress, O(N)- ab initio-MD is still limited to systems containing relatively small numbrs of atoms because of the overhead associated with the self-consistent calculations in ab initio MD. On the other hand, MD schemes based on two-center tight-binding (TB) Hamiltonians are orders of magnitude faster than O(N)-ab initio-MD methods. However, they have been found to be unsatisfactory in systems where charge transfer or bond breaking/rebonding plays a significant role. Therefore, there is a pressing need for developing transferable semi-empirical Hamiltonians that would be superior to traditional two-center TB Hamiltonians, but would include all essential ingredients of ab initio Hamiltonians without being computationally excessive. The O(N)-MD scheme using such a Hamiltonian can predict accurately the properties of complex systems of large sizes.In this research a general scheme will be developed to construct such reliable and transferable semi-empirical Hamiltonians for materials (metal or semiconductor) in the framework of linear combination of atomic orbitals (LCAO) that explicitly includes the self-consistently (SC) determined charge transfer and environment-dependent (ED) multi-center interactions. The SCED-LCAO Hamiltonian will be implemented within the O(N)-MD scheme developed previously. Using O(N)/SCED-LCAO-MD as our simulation tool, we will investigate the properties of carbon multi-wall nanotubes and carbon nanorods, and evaluate their potential as components of molecular-scale devices. Specifically, we will study the following three projects of current interest: (1) Energetics, structure, electronic, mechanical and vibrational properties of carbon multiwall nanotubes (MWNT), (2) A study of the properties of contacts between metal elctrodes and MWNT, and (3) Energetics, structure, electronic, mechanical and vibrational peoperties of carbon nanorods (CNR). Project (2) will be done in collaboration with the experimetal group of Professor Alphenar of Lousiville, whereas project (3) will be done in collaboration with the experimental group of Professor Chen of Academia Sinica (Taiwan).%%%***