Bridging magnetic and electronic structure techniques via atomistic approaches

构建磁性材料结构-性能相图的新能力。在上个世纪这类材料的研究中，一个难题是总能量计算（现在基于密度泛函理论（DFT））和磁结构计算（使用模型哈密顿方法）之间存在明显的障碍。在前一个领域，磁性组分在结构性质随温度、压力或化学势的演化中所起的作用往往是不清楚的。从DFT的角度来看，磁性作为自旋或磁矩的表现是立即明显的，但特定的相变可能是由系统的其他性质（如热膨胀）“驱动”的。从磁的角度来看，从电子结构计算中获得的信息反过来可能是令人沮丧的不透明的，因为它没有阐明哪种类型的物理相互作用在特定条件下占主导地位。为了将这两种方法结合起来，我们建议从电子结构的许多细节中抽象出来，只关注那些直接影响磁性的细节。系统的剩余相互作用可以用传统的原子间势来解释。将原子视为（潜在极化的）点电荷的原子论方法的威力将通过用点状磁矩增加这种描述而得到增强。因此，组合的总能量描述将包括半经典原子贡献和磁（自旋）模型项，该模型项也可以半经典地或通过适当的量子力学方法获得。后者通常需要大量的晶格蒙特卡罗模拟，这将需要利用高性能计算机。学生的申请项目将涉及三元锰基氧化物，最初将集中于CaMnO3、LaMnO3及其固溶体的建模。这项工作将涉及三个方面：(i) DFT计算，研究小单元中的磁有序，旨在获得将用于（ii）磁蒙特卡罗模拟，旨在从模型哈密顿量和实验数据中获得有序参数信息；（iii）两个纯系统的原子GULP模拟，包括两种类型的相互作用。当主要模型参数建立后，工作将转移到固溶体和相应相图的构建。

New capability of constructing structure-property phase diagrams of materials with magnetic properties. One of conundrums in the last century of research in this class of materials was a distinct barrier between total energy calculations, nowadays based on the Density Functional Theory (DFT), and magnetic structure calculations using model Hamiltonian approaches. In the former field it is often unclear what roles the magnetic components play in the structure property evolution with temperature, pressure or chemical potentials. The magnetism from the DFT point of view is immediately apparent as a manifestation of spin or the magnetic moment, but a particular phase transition may be "driven" by other properties of the system such as thermal expansion. From the magnetic perspective the information available from the electronic structure calculations in turn can be frustratingly opaque in that it does not clarify which types of the physical interactions are dominant for a particular set of conditions. To bring these two methods together, we propose to abstract from many details of the electronic structure and concentrate only on those that will directly affect the magnetic properties. The remaining interactions of the system can be accounted for using traditional interatomic potentials. The power of the atomistic approaches that treat atoms as (potentially polarisable) point charges will be enhanced by augmenting this description with point-like magnetic moments. A combined total energy description therefore will include semi-classical atomistic contribution and a magnetic (spin) model term that can be obtained also semi-classically or with an appropriate quantum-mechanical approach. The latter will typically require extensive lattice Monte Carlo simulations that will need to exploit High Performance Computers. The student's application project will concern ternary manganese based oxides initially concentrating on modelling CaMnO3, LaMnO3 and their solid solution. The work will involve three aspects: (i) DFT calculations studying magnetic ordering in a small unit cell aimed at obtaining parameters of magnetic interactions that will be used in (ii) magnetic Monte Carlo simulations that will aim to derive order parameter information from model Hamiltonians and the experimental data; and (iii) atomistic GULP simulations of the two pure systems including both types of interactions. When major model parameters will be established - the work will be moved to the solid solution and construction of the corresponding phase diagram.