光化学反应是自然界中最重要的过程之一，与人类的生活息息相关：光合作用为人类提供赖以生存的能源和氧气；DNA的光稳定性保证了人类的生存与繁衍；视紫红质的光异构化是引起视觉的重要原因，可以说，没有光化学反应就没有如此丰富多彩的世界。理解光化学反应背后的微观机制对能源，生物，环境和材料等领域都有着非常重要的意义。然而，由于光化学反应往往具有过剩的内部能量，相关电子态寿命短，反应通道繁多且复杂等特点，除了需要借助于超快、高分辨的光谱技术外，也需要高精度的理论计算和模拟，对理论和实验研究都是挑战性的前沿课题。 光化学反应总是始于电子激发态，但最终产物是在基态。该过程必然涉及不同电子态势能面之间的跃迁，是一个典型的非绝热过程。借助于静态电子结构计算，探索势能面的结构和不同势能面之间的交叉情况，是光化学反应中的一个核心问题；此外，由于有大量过剩的内部能量，光化学反应并不一定沿着极小能量路径进行，并不满足统计平衡，需要借助于非绝热动力学模拟，这是光化学反应的另外一个核心问题。本文旨在发展和应用适用于有机分子、过渡金属配合物以及周期性材料体系的非绝热动力学模拟的算法，希望可以针对不同类型的体系采取适当的非绝热动力学模拟算法，从而高效解决相关问题。因此，本文分为如下几个部分: （1）拓展了基于Zhu-Nakamura非绝热动力学算法，在高精度的QM(CASPT2)/MM的水平上，研究了有机小分子体系在溶液中的光化学反应过程。随后采用该方法研究了1，2，3-噻重氮的光解离及重排机理，模拟表明该体系在受到光激发后，体系到达S2态后，由于能量与S1态能量近简并，因此在Franck-Condon区域超快的失活到S1态，在S1态上，沿S-N键断裂的反应坐标进一步失活到反应基态，并在此过程中生成了S-N键断裂的反应中间体，最终由该反应中间体在基态通过协同非同步的重排反应生成不同的反应产物，纠正了此前实验中提出的在激发态发生重排的反应机理。 此外，对于某些不含过渡金属的有机分子体系，除了采用高精度的电子结构计算方法外，还可以通过高精度的静态电子结构计算，结合半经验的OM2/MRCI非绝热动力学模拟，研究一些光物理过程。基于该方案，研究了绿色荧光蛋白（GFP）发色团类似物o-LHBDI的荧光增强机制，研究表明，体系受光照跃迁至激发态后，会发生超快的分子激发态质子转移过程到达酮式构型，在酮式构型，由于化学修饰，酮式圆锥交叉结构能量较高，体系难以通过该通道超快失活至基态，荧光增强。通过与此前的研究相比较，光诱导的激发态质子转移及化学修饰两者的协同作用对GFP发色团类似物的荧光强度具有非常重要的影响。 （2）对于电子态性质复杂的过渡金属配合物体系，发展并程序化了一种基于含时密度泛函理论（TD-DFT）的广义最少面跳跃方法，该方法采用了可以高效计算体系非绝热耦合项以及自旋-轨道耦合的算法，结合经典路径近似，该方法可以模拟过渡金属配合物超快的内转换和系间窜跃过程。利用该方法系统研究了不同取代的Au（I）配合物超快的系间窜跃过程和Ir（III）配合物的短时动力学，解释了a）Au（I）配合物取代基不同取代位置引起的系间窜跃速率的变化问题，较小的能量差以及较大的旋轨耦合是造成该类配合物超快系间窜跃过程的主要原因，而AU-2相对于AU-3明显减小的旋轨耦合是造成两者系间窜跃速率差异的主要因素；b）Ir（III）配合物不同配体造成的各个体系迥异的电子和空穴转移过程，其中Ir1主要发生的是Ir到三个ppy配体的空穴转移过程，Ir2主要发生的则显示配体bpy到ppy的电子转移，随后电子再从ppy转移回bpy配体，而Ir3则同时发生了ppz配体到dipy配体的电子和空穴转移过程。 （3）对于周期性材料体系，在Prezhdo等人提出的基于Kohn-Sham轨道的非绝热动力学方法基础上进行了拓展，将该方法由平面波基组拓展到了高斯基组，研究了锌酞菁配合物与二硫化钼的界面电子转移过程。研究发现，在受光激发后，体系的一个电子由锌酞菁配合物的最高占据轨道（HOMO）跃迁到该配合物的最低空轨道（LUMO），之后LUMO轨道上的电子超快的转移至二硫化钼表面，电子转移过程的时间尺度为10 fs。进一步的分析表明，分子振动导致的绝热电子态的变化是引起该超快过程的主要因素。此外，还研究了聚合物PTB7和二硫化钼界面间的电子和空穴转移过程，研究发现电子转移与PTB7层的厚度关系较小，对于单层PTB7和五层PTB7吸附的异质结结构，可以在10 fs以内完成，而空穴转移过程则与PTB7的层数密切相关，当PTB7层数由一层增加到五层时，空穴转移时间尺度由70 ps加速缩短至1 ps。这些工作证明了新发展方法的可靠性，对进一步理解有机物/聚合物-过渡性金属硫化物异质结界面的电子和空穴转移过程有一定的贡献。

Photochemical reaction is one of the most important processes in nature and is closely related to our human life: photosynthesis provides human energy and oxygen to survive; the light stability of DNA guarantees the reproduction and continuation of human beings; the isomerisition of retinal is an important cause of vision. It can be said that there is no such rich world without photochemical reactions. Understanding the microscopic mechanisms behind these processes is very important in areas such as energy, biology, environment, and materials. However, photochemical reactions are often challenging due to their high energy, short lifetime, and complex reaction channels. To investigate these processes, we not only need the help of ultra-fast time-resolved spectroscopy, but also need to conduct high-level static electronic calculations and non-adiabatic simulations. Photochemical reaction often starts from the excited state and finally returns to the ground state, which is a typical non-adiabatic process. The process necessarily involves multiple potential energy surfaces. The static electronic structure can be used to calculate the minima, transition states and minimum energy paths of these potential energy surfaces as well as conical intersections between the potential energy surfaces. This is a core issue in photochemical reactions. However, due to its high energy, the photochemical reaction process does not always follow a minimum energy path, and some of the reaction processes are ultrafast and do not satisfy the statistical balance. The relevant dynamics cannot be described by relying on static electronic structures alone. The process requires simulation with non-adiabatic effects, which is another core issue in photochemical reactions. This paper aims to develop and apply non-adiabatic dynamics simulation algorithms suitable for different scale systems. It is hoped that appropriate non-adiabatic dynamics simulation algorithms can be adopted for different scale systems to solve related problems efficiently. Therefore, this article is divided into the following sections: (1) We extend the non-adiabatic dynamics algorithm based on Zhu-Nakamura to the high-level QM(CASPT2)/MM level, which can be used to study the photochemical reaction process of small molecule systems in solution. We have studied the photodissociation and rearrangement mechanism of 1,2,3- thiadiazole by this method, and corrected the mechanism of the excited state rearrangement proposed in the previous experiment. The simulation shows that after being excited to S2 state, the molecule quickly relaxed to the ground state, forming an intermediate in the ground state, through which the dissociation and rearrangement process is completed to produce different products. In addition, for some transition-metal-free organic molecular systems, we can also combine semi-empirical OM2/MRCI non-adiabatic dynamics simulations with high-precision static electronic structure calculations to study its photophysical processes. Based on this scheme, we studied the fluorescence enhancement mechanism of the green fluorescent protein (GFP) chromophore analogue o-LHBDI. Our study reveal that photoinduced excited-state intramolecular proton transfer and chemical modification have an important influence on the fluorescence intensity of GFP chromophore. (2) For transition metal complex systems with complex electronic states, we have developed and programmed a generalized trajectory surface hopping method based on time-dependent density functional theory (TD-DFT) method. We have programed new algorithms that can be used to efficiently calculate non-adiabatic coupling terms and spin-orbit coupling algorithm. In combination with the classical path approximation, this method can simulate the ultrafast internal transition and intersystem crossing processes of transition metal complexes. We have systematically studied the ultrafast intersystem crossing processes of different substituted Au(I) complexes and the early-time dynamics of Ir(III) complexes. The results not only agree well with previous experimental studies, but also provide some valuable insights into their microscopic excited-state relaxation dynamics. (3) For the periodic systems, we extended the non-adiabatic dynamics based on the Kohn-Sham orbital proposed by Prezhdo et al. from the plane wave basis sets to the Gaussian basis sets. This method was used to study the interfacial electron transfer between zinc phthalocyanine and molybdenum disulfide. It is found that after photoexcitation one electron of the system transitions from the highest occupied orbital (HOMO) of the zinc phthalocyanine to the lowest empty orbit (LUMO) of the complex, and then the electron in the LUMO is ultra-fastly transferred to the the surface of molybdenum sulfide, the time scale of the electron transfer process is 10 fs. Further analysis indicates that changes in the adiabatic electronic state caused by molecular vibration are the main factors that cause this ultrafast process. In addition, we also studied the electron and hole transfer processes in two PTB7-nL@MoS2 models (n=1 and 5). The interfacial electron transfer is found to be ultrafast and completes within ca. 10 fs in both PTB7-1L@MoS2 and PTB7-5L@MoS2 models, which demonstrates that the electron transfer is not sensitive to the thickness of the PTB7 polymer. Differently, the interfacial hole transfer is sensitive to the thickness of the PTB7 polymer. The transfer time is estimated to be ca. 70 ps in PTB7-1L@MoS2 while it is significantly accelerated to ca. 1 ps in PTB7-5L@MoS2.These works demonstrate the reliability of our newly developed methods and contributes to the the design of excellent charge-separated interfaces of mixed-dimensional TMD-based heterojunctions for a variety of optoelectronic applications.