Development And Application Of Quantum Vibrational Perturbation Method And Comparative Study Of Semi-empirical Methods In Combustion Reaction Simulation

Studying systems at larger scales and giving more accurate predictions have long been a constant goal of researchers in theoretical and computational chemistry.The simulation of complex systems poses a great challenge to both of them,and the development of various approximation theories,localization theories and parametric methods has become an important research content.Among the research works on many complex systems,the simulation of infrared spectra of probe molecules and the simulation of combustion reaction processes are two areas that we focus on.In this paper,the authors’work on the comparative study of computational methods for the simulation of infrared probe molecules and the simulation of combustion reactions are presented with the aim of developing accurate and efficient methods for the simulation of infrared spectra and finding low-cost computational methods suitable for the simulation and calculation of complex systems.To solve the problem that spectral peaks are difficult to be identified and interpreted theoretically due to the ultrafast time scale of molecular vibrations,the quantum effect of vibrations,and the complex solvent environment,we further developed the Quantum Vibration Perturbation（QVP）method,which combines quantum mechanical calculation of vibrational frequencies with classical molecular dynamics and describing the wave functions of vibrational ground and excited states with a small number of grids through discrete variable representations,the vibrational potential energy functions of molecules can be calculated in real time using quantum chemical methods,thus calculating the time-dependent quantum vibrational frequencies of infrared characteristic groups in molecules more accurately and quickly,and using perturbation methods to obtain time-containing vibrational frequency shifts in environments such as clusters or solvents,thus avoiding the need to solve the Schr（?）dinger equation of a one-dimensional potential energy curves at each time step,thus greatly reducing the computational effort,a method that has yielded good results in previous infrared spectra simulations of diatomic molecules in their aqueous cluster system by our research group.However,for polyatomic molecules,there is no reliable algorithm to embed vibrational reference state wave functions into molecular dynamics simulations.Polyatomic molecules have multiple intramolecular degrees of freedom,and in addition to intermolecular interactions between the probe molecular chromophore and the environment,other intramolecular motions may also affect the vibrational frequency shift of this chromophore.In this paper,we extend the QVP method to the case of polyatomic chromophores and construct a reliable algorithm for embedding quantum vibrational effects into molecular dynamics trajectories.At the same time,we designed an energy scaling method to reduce the computational cost of the cluster system to ensure both efficiency and accuracy.The improved QVP method was applied to a variety of systems,We first simulated the C=O stretching vibration infrared spectrum of formic acid in formic acid-water cluster,and the position of the simulated vibration peak deviated from the experiment by less than 1 cm^-1,which reproduced the local continuous spectral peak pattern of HCOOH-n H₂O（n=1,2）very well.The C=O stretching vibrational infrared spectrum of formic acid in formic acid-water solution was also simulated,and the position of the simulated vibrational peak deviated from the experiment by only 13 cm^-1,reproducing the local experimental spectral peak pattern.The performance of the QVP method in systems dominated by dispersive interactions was also examined by simulations of NMA-d IR probe molecules in a variety of solutions.In the work on the simulation of combustion reaction systems,we investigate the performance of various semi-empirical methods in the simulation of combustion reaction-related systems with the formation process of soot in combustion reactions.Soot molecules originate from the incomplete combustion of hydrocarbon fuels in a variety of environments and are substances that are hazardous to human health and affect climate change on Earth.Computational chemistry provides an effective tool to study their properties and formation processes,however,accurate quantum chemical calculations are too expensive for simulations of combustion reactions and high throughput calculations.Given our computational experience in infrared spectroscopy simulations,we believe that semi-empirical quantum chemistry methods are a viable option for balancing accuracy and computational cost.In this work,the performance of a variety of widely used semi-empirical methods in describing the soot formation process is studied in comparison.Our benchmark study focuses on,but is not limited to,validating the performance of semi-empirical methods in the calculation of reactive and non-reactive molecular dynamics trajectories.We also investigate the accuracy of semi-empirical methods in predicting the structure of soot precursors and the energy distribution along the intrinsic reaction coordinates.Finally,we discuss the performance of these semi-empirical methods for predicting electron spin density.The semi-empirical methods in the comparison include AM1,PM6,PM7,GFN2-x TB,DFTB2（with spin polarization）,and DFTB3.Taken together,the relative energies and molecular structures of the molecular dynamics trajectories predicted by the semi-empirical methods are relatively accurate,indicating that the semi-empirical methods are adequate for sampling carbon soot formation in large-scale reactions and for calculating the main reaction mechanisms,while the results also suggest that they are not suitable to be used to calculate quantitatively accurate data related to thermodynamics and reaction kinetics,but the findings of this work support our work on spectroscopic simulations of the NMA-d system during the same period,which largely improved the results of single-point energy calculations by using a more appropriate semi-empirical method.The chapter 1 of this paper will focus on the background and purpose of the study.The chapter 2 enumerates the main theoretical and methodological tools used in the study.In the chapter 3,we develop a suitable method for the simulation of infrared spectra of polyatomic probe molecules,which can simulate the infrared spectra of probe molecules in clusters and solutions more accurately at a lower computational cost,and apply it successfully with the example of formic acid-water clusters.Chapter 4 focuses on the simulation of the IR spectra of formic acid C=O stretching vibrations in formic acid-water solutions by the QVP method,and the verification of the QVP method in dispersion force dominated systems by simulating the IR spectra of NMA-d probe molecules in different solvents.Chapter 5 covers a comparative study of the performance of various semi-empirical methods in the calculation of soot molecular systems.Chapter 6 provides a summary and outlook.