Understanding the Nature of Proton Transport and Hydrogen Bond Networks in Complex Environments by Accelerated Quantum Simulations

氢键(H键)的形成和质子传输在许多与生物和技术相关的过程中起着关键作用。酶中的氢键网络是加快化学反应速度的关键，而高质子迁移率对水化学和燃料电池的发展具有重要意义。然而，通过实验很难获得关于复杂体系中质子位置和质子转移机制的准确信息。此外，较小的氢质量会引起核量子效应，如隧道效应、零点运动和量子离域等，这些效应往往是不能忽略的。虽然从头算计算机模拟明确考虑了电子和原子核的量子性质，对氢键的性质提供了重要的见解，但其高昂的计算成本严重限制了它们的应用。该项目的目的是建立一个计算框架，允许通过准确和高效的量子模拟来详细研究复杂凝聚系统中的质子输运和氢键。通过将有效的机器学习势与加速量子模拟收敛的方法相结合，在不损失精度的情况下，传统从头计算方法的计算负担将减少两个数量级以上。这一新方法将被应用于研究浓酸溶液的量子动力学和研究HIV天冬氨酸蛋白酶活性部位的氢键网络，该酶是抗HIV药物的重要靶点。除了这些应用之外，所提出的方法代表了一个加速量子模拟的一般框架，该框架将适用于大量系统，在这些系统中，光原子的存在需要明确包括原子核的量子性质。

Hydrogen bond (H-bond) formation and proton transport play a key role in numerous biologically and technologically relevant processes. H-bond networks in enzymes are critical in accelerating the rate of chemical reactions while the high proton mobility has important implications for aqueous chemistry and the development of fuel cells. However, accurate information on proton location and proton transfer mechanism in complex systems is difficult to obtain by experiment. Moreover, the small hydrogen mass gives rise to nuclear quantum effects such as tunneling, zero-point motion and quantum delocalization which can often not be neglected. While ab initio computer simulations which explicitly consider the quantum nature of the electrons and the nuclei give important insights into the properties of H-bonds, their high computational cost severely limits their application.The aim of the proposed project is to develop a computational framework which allows for a detailed investigation of proton transport and H-bonds in complex condensed systems by accurate and efficient quantum simulations. By combining efficient machine learning-potentials with methods to accelerate the convergence of quantum simulations, the computational burden of conventional ab initio methods will be reduced by more than two orders of magnitude without loss in accuracy. This novel approach will be applied to investigate the quantum dynamics of concentrated acid solutions and to study the H-bond network in the active site of the enzyme HIV aspartic protease, an important target for anti HIV-drugs. Beyond these applications, the proposed method represents a general framework for accelerating quantum simulations which will be applicable to a large range of systems in which the presence of light atoms requires to explicitly include the quantum nature of the nuclei.