Unraveling the nature of inter and intra-molecular interactions

加州大学伯克利分校的Martin Head-Gordon获得了化学学部化学理论、模型和计算方法项目的奖励，他开发了计算工具，为量子化学计算的数值结果提供了物理洞察力。今天，大约有一半的化学新研究出版物使用量子化学计算作为实验的补充，以预测化学反应机制、结构和各种分子性质。大多数关于量子化学方法论的研究都集中在数值方法和算法的改进上，以帮助这些应用。虽然这类研究很有价值，因为它能提供更好的预测，但最终的预测可能只产生数字，而不是物理洞察力。相比之下，本项目开发的工具专注于解释量子化学计算的结果。例如，计算出的相互作用能，例如分子与酶活性位点的结合，可以分解为具有不同物理来源的附加贡献，例如永久电荷-电荷相互作用、诱导相互作用和电荷转移效应。观察在给定的分子相互作用中哪些贡献占主导地位可以解释结合的起源，从而为通过进行化学修饰来调整相互作用提供合理的基础。这些工具有许多用途，从增加对简单键合过程的理解，到理解新型键合（如卤素键），再到能够合理设计改进催化剂或药物的预期目标。这种分解的理论被称为能量分解分析。现有的EDA方法中，只有少数几种是成功的，并被广泛应用于量子化学。每种方法都有明显的局限性，这促使本项目开发和测试新的和改进的EDA。在更高的技术层面上，目标是开发新一代EDA，以消除基于绝对定域分子轨道（ALMO-EDA）的现有EDA的三个关键限制。这将对更广泛的分子问题产生更可靠的解释，在这些问题中，化学家除了寻求数字之外还寻求洞察力。第一个目标是发展一种静电极化的收敛方法，该方法不依赖于底层原子轨道（AO）基础来定义重叠区中的极化。因此，这种方法将具有定义良好的基集极限。第二个目标是使用新的极化处理和新的电荷转移的非微扰处理，将EDA扩展到键合的分子内相互作用。由于这种相互作用非常强，但可能涉及更大的正负项之间的抵消（例如，正排斥和几何畸变与负极化和电荷转移），因此EDA的绝热推广的发展也是一个优先事项。这个绝热EDA将被定义为所有的贡献都是负的半确定的，这样对强相互作用的解释就更直接了。第三个主要目标是发展EDA的扩展，以超越单行动式波函数的框架（即Hartree-Fock或密度泛函理论）。要正确地处理包含相关的波函数方法，需要发展相关对冻结和极化相互作用的贡献理论，以及负半确定的色散贡献理论。最初的目标是二阶Moller-Plesset理论，但随后探索耦合簇理论的扩展也是非常可取的。

Martin Head-Gordon of the University of California, Berkeley is supported by an award from the Chemical Theory, Models and Computational Methods program in the Chemistry Division to develop computational tools that provide physical insight into the numerical results of quantum chemistry calculations. Roughly half of new research publications in chemistry today use quantum chemistry calculations as a complement to experiment, to predict chemical reaction mechanisms, structure, and diverse molecular properties. Most research on quantum chemistry methodology is focused on improvements in the numerical methods and algorithms to aid such applications. While very valuable because such research enables better predictions, the resulting predictions may yield only numbers and not physical insight. In contrast, the tools developed in this project are focused interpreting the results of the quantum chemistry calculations. For example, a computed interaction energy, such as the binding of a molecule to an active site of an enzyme, can be decomposed into additive contributions which have different physical origins, such as permanent charge-charge interactions, induced interactions, and charge-transfer effects. Seeing which contributions are dominant in a given molecular interaction can explain the origin of the binding, and thus provide a rational basis for tuning the interaction by making chemical modifications. There are many uses for these tools ranging from increased understanding of simple bonding processes, to understanding new types of bonding such as halogen bonds, to the desired goal of being able to rationally design improved catalysts or drugs. The theory of such a breakdown is called an energy decomposition analysis (EDA). There are only a few other successful existing EDA methods, which are widely used in quantum chemistry. Each of these has demonstrable limitations, which motivates this project on the development and testing of a new and improved EDA. At a more technical level, the objective is to develop a new generation EDA that lifts three key limitations of an existing EDA based on absolutely localized molecular orbitals (ALMO-EDA). This should yield more reliable interpretations on broader classes of molecular problems where chemists seek insight in addition to numbers. The first objective is the development of a convergent method for electrostatic polarization which does not rely on an underlying atomic orbital (AO) basis to define polarization in the overlapping regime. This method will therefore have a well-defined basis set limit. The second goal is to use the new polarization treatment together with a new non-perturbative treatment of charge-transfer to extend the EDA to bonded intra-molecular interactions. Since such interactions are very strong and yet can involve cancellation between even larger positive and negative terms (e.g. positive repulsions and geometric distortion vs negative polarization and charge-transfer), the development of an adiabatic generalization of the EDA is also a priority. This adiabatic EDA will be defined so that all contributions are negative semi-definite, such that interpretation of strong interactions is more straightforward. The third principal objective is to develop an extension of the EDA to go beyond the framework of single determinant wave functions (i.e. Hartree-Fock or density functional theory). To treat wave function methods that include correlation correctly requires developing theory for the correlation contribution to the frozen and polarization interactions, as well as a dispersion contribution that is negative semi-definite. The initial target is second order Moller-Plesset theory, but it will also be highly desirable to subsequently explore extensions to coupled cluster theory.