Models and Dynamics for Energy and Charge Transfer in Light Harvesting

来自波士顿大学的David Coker获得了化学理论、模型和计算方法项目的奖励，他开发了新的理论和计算工具，用于理解在植物和细菌等自然系统以及合成系统中如何通过光合作用收获光。他开发了建模、分析和预测能量传输和电荷分离过程的光谱特征的方法，这些过程是光合作用的基本组成部分。大自然已经发展出非凡的方法，在分子水平上调整纳米级结构，以优化不同条件下的功能。在新的生物启发纳米材料和技术中模仿这种多功能性需要对这些微观过程的工作原理有基本的了解。这种理解目前是缺失的或不发达的。本研究中探索的新的理论和计算方法将提供解释和指导实验的手段，以发展对这些系统行为的基本理解。这些知识是设计用于光能收集的多功能功能分子纳米结构的关键。在这个项目中，Coker和他的团队将首先专注于扩展，第一性原理，激发态量子化学方法和构象采样技术，以计算最近在超快非线性光谱学研究中受到广泛关注的生物光收集系统模型中的参数分布。这种模型通常用来解释这些平均实验的结果。这些最佳拟合的平均模型有许多难以估计的参数，而且它们通常不是唯一的，经常导致模棱两可的解释。然而，Coker小组正在开发的理论方法能够详细分析平均值之下的波动，并对一组独特模型进行抽样，这些模型包括，例如，高性能结构异常值，其特征将对最佳设计（而不是平均行为）提供重要理解。在第二个项目中，使用耗散量子动力学方法计算光谱性质，并使用计算模型的集合研究包括能量输运和电荷分离在内的弛豫过程。一系列技术将被实现，从标准的微扰近似，包括雷德菲尔德理论和福斯特理论的变体，到更一般的非微扰方法，包括路径积分、半经典和混合量子经典方法，如在以前的工作中开发的部分线性化密度矩阵动力学方法。在不同的动态处理之间切换的有效、自动化方案将使用标准来决定哪种方法将提供最佳的动态描述，平衡给定采样模型的准确性和可靠性与计算成本。该方案的发展对于使这些广泛波动的模型的采样有效是重要的，因为集合可以在非常不同的动力状态下探索模型。这种计算集成动力学的自适应方法将在研究大规模现实系统之前，对可以在较小的简化模型上执行的精确计算进行基准测试。

David Coker from Boston University is supported by an award from the Chemical Theory, Models and Computational Methods program to develop new theoretical and computational tools for understanding how light is harvested through photosynthesis both in natural systems like plants and bacteria, and synthetic ones. He develops methods for modeling, analyzing and predicting spectroscopic signatures of energy transport and charge separation processes that are essential components of photosynthesis. Nature has developed remarkable ways of adapting nano-scale structures at the molecular level to optimize function under varying conditions. Mimicking this versatility in new bio-inspired nano-materials and technologies requires a fundamental understanding of the workings of these microscopic processes. Such understanding is currently missing or poorly developed. The new theoretical and computational methods explored in this research will provide the means to interpret and guide experiments with the goal of developing fundamental understanding of the behavior of these systems. Such knowledge is key to the design of versatile functional molecular nano-structures for light energy harvesting. In this project, Coker and his team will first focus on extending, first principles, excited state quantum chemical methods and conformational sampling techniques to compute the distributions of parameters in models of the biological light harvesting systems that have received much attention in recent ultrafast nonlinear spectroscopy studies. Such models are usually employed to interpret the results of these averaged experiments. These best-fit, average models have many parameters that can be difficult to estimate and they are not generally unique, often leading to ambiguous interpretation. The theoretical methods being developed by the Coker group, however, enable detailed analysis of fluctuations underlying the average and the sampling of an ensemble of unique models that include, for example, highly performing structural outliers whose characteristics will give important understanding for optimal design, rather than mean behavior. In the second project, dissipative quantum dynamical methods are employed to compute spectroscopic properties and study relaxation processes including energy transport and charge separation using the ensembles of computed models. A suite of techniques will be implemented ranging from standard perturbative approximations including Redfield theory, and variants of Forster theory, to more general non-perturbative approaches including path-integral, semi-classical, and mixed quantum-classical methods like the partial linearized density matrix dynamics approaches developed in previous work. An efficient, automated scheme for switching between different dynamical treatments will be implemented using criteria to decide which approach will provide an optimal description of the dynamics, balancing accuracy and reliability with computational cost for a given sampled model. The development of this scheme is important to make the sampling of these widely fluctuating models efficient, as the ensemble may explore models in very different dynamical regimes. This adaptive approach to computing the ensemble dynamics will be benchmarked against exact calculations that can be performed on smaller simplified models before it is implemented to study large-scale realistic systems.