针对航空煤油燃烧和碳烟形成机理进行分子水平的深入研究，可为提高发动机燃烧效率和控制颗粒污染物排放提供理论支持。但航空煤油和碳烟颗粒的分子结构极为复杂，微观反应机制的认识是长期的挑战性任务。ReaxFF MD模拟是结合新一代反应力场ReaxFF和分子动力学模拟的计算方法，可连续描述化学反应的演化，无需预先设定反应路径，计算准确度接近于密度泛函方法但大幅减少了计算代价，可用于~1,000–10,000原子量级或更大的复杂分子模型。这些特性使其成为有潜力研究复杂燃料燃烧及碳烟形成的微观反应机制的新方法。以深入认识复杂多组分航空煤油燃烧和碳烟形成的化学反应机理为目标，本论文基于作者所在团队自主研发的、国际领先的基于GPU的ReaxFF MD模拟程序GMD-Reax和反应机理分析可视化程序VARxMD，通过采用ReaxFF MD对航空煤油RP-1燃料混合物24组分模型的模拟，系统考察了RP-1热解、氧化和碳烟形成过程中主要中间物和产物的演化规律以及相关的化学反应。主要工作结果如下：基于文献报道的航空煤油燃料RP-1的代表性组成信息，构建了RP-1的24组分燃料分子模型和3组分化学替代燃料分子模型。采用ReaxFF MD模拟方法对RP-1燃料热解和氧化过程进行了研究，重点考察了两种RP-1燃料模型在升温条件下的热解反应性差异。模拟结果发现：两种RP-1燃料模型热解得到的反应物和重要小分子产物乙烯演化的整体趋势相似，但3组分替代模型中RP-1总体消耗比24组分模型略慢，质量分数最多可相差20%。将RP-1按直链烷烃、支链烷烃及环烃进行分类，考察了两种模型的反应性后发现，环烃对两种模型中RP-1热解消耗的质量分数的差异贡献最大，主要原因是3组分模型中环烃分子的热稳定性较好，质量分数占比也较大。进一步采用VARxMD对两种RP-1模型中燃料热分解的完整化学反应路径进行了考察，揭示了两种RP-1模型热解反应路径的差异：对于支链烷烃，3组分替代燃料模型中的七甲基壬烷是以季碳原子甲基取代为特征的多分支结构，容易得到支链烯烃2-甲基丙烯，但24组分模型中的支链烷烃组分多为叔碳原子甲基取代，热解产物主要为直链烯烃；对于环烃，3组分模型的甲基环己烷是支链很短的单环结构，开环路径相对简单，而24组分模型的环烃燃料含有双环结构和复杂支链的单环结构，开环路径更加复杂。该模拟结果不仅为进一步优化3组分替代燃料模型组成提供了线索，更为重要的是提出了一种有潜力通过计算评价替代燃料模型化学性质的新途径。采用单组分晕苯分子模型，通过ReaxFF MD模拟探索了热解过程中由多环芳烃前驱体向碳烟纳米颗粒成核及随后演化的化学反应路径。1 ns的模拟结果表明：温度在2600–3000 K内的升高有利于最大碳烟纳米颗粒的碳原子数增长，但温度进一步升高到3400 K时，其原子数会有下降；以形貌和H/C比为标志，模拟温度升高有利于最大纳米颗粒的成熟程度增加。在3400 K的高温条件下，可观察到由多环芳烃分子向碳烟成核及纳米颗粒进一步结构转变的化学反应路径。多环芳烃之间首先形成碳碳键得到二聚体或更大的低聚体，意味着碳烟纳米颗粒的成核；低聚体和多环芳烃、链烃分子之间的反应使颗粒生长；颗粒继续生长会形成有长侧链和大环结构的无定形颗粒；进而环上碳碳原子之间桥接成键得到小环，使颗粒的多环芳烃状结构尺寸进一步增长；成环和热裂解反应使侧链长度与数量减少；最终形成具有较好形貌的纳米颗粒。这些结果从全原子层面展示了由多环芳烃到碳烟纳米颗粒形成过程的化学结构变化，有利于丰富碳烟形成相关反应路径的已有认识。进一步采用24组分RP-1燃料模型，通过5–10 ns长时间ReaxFF MD高温氧化模拟，研究了复杂组分RP-1高温贫氧氧化时碳烟纳米颗粒形成的全景式演化路径。模拟得到的碳烟形成过程可划分为三个阶段。(1)初始环的形成和生长：初始环的形成来自于燃料分子脱氢得到的活化长链类炔烃分子的环闭合；环上碳原子之间桥连成键和环的缩合使分子的环数进一步增加，形成带侧链的类多环芳烃。这是在普遍接受的多环芳烃形成的HACA机理之外可能存在的一种新的成环方式。(2)颗粒成核：距离相近的两个多环芳烃状分子会发生侧链相连成环、环内部碳原子之间桥连成键的环缩合反应，两个多环芳烃的核合并将形成一个新的多环芳烃大核结构。这一观察表明多环芳烃的侧链在碳烟成核过程中发挥着重要作用。(3)颗粒的石墨化过程：碳烟纳米颗粒的碳骨架结构会缓慢重整，由形貌较不规则的颗粒逐步形成形貌良好的富勒烯状颗粒结构，主要是5元环或7元环通过3元环中间体重整为6元环结构。本论文还通过5 ns的长时间ReaxFF MD模拟，系统考察了模拟温度（1500–4000 K）、化学计量比（0.5–10）、和体系密度（0.01–0.5 g/cm3）变化对碳烟纳米颗粒形成的影响。本工作从全原子层面揭示了由多组分燃料分子向碳烟纳米颗粒形成过程的全景式化学反应路径和分子结构信息，为丰富已有碳烟形成机理的认识提供了线索。本论文通过ReaxFF MD模拟RP-1航空煤油多组分模型、替代燃料模型和多环芳烃模型化合物，较为系统地揭示了RP-1燃料的热分解反应机理以及从燃料分子到碳烟纳米颗粒形成全过程的反应路径，其中对RP-1航空煤油多组分模型与替代燃料模型反应性比较的方法有望发展为评价替代燃料模型化学性质的新途径。论文工作表明ReaxFF MD模拟是从分子层次上系统认识复杂航空煤油燃烧过程化学反应细节的有前景的方法。;Deep understanding on the reactions of aviation fuel combustion and soot formation at molecular level is vital for improvement of fuel combustion efficiency, particulate emission control and for design of aviation engines. Since the components of aviation fuel and the molecular structures of soot particles are very complicated, it is challenging to fully understand the chemical reaction mechanism therein. ReaxFF MD is a computational approach combining the novel reactive force field ReaxFF with molecular dynamics (MD) simulations. The ReaxFF force field was developed on the basis of bond order, which allows for the descriptions of breaking and formation of chemical bonds for chemical reactions in a reactive molecular model. The accuracy of ReaxFF force field is close to DFT method with much less computational cost than DFT. Molecular models of ~1,000–10,000 atoms or even larger models have been used in ReaxFF MD simulations for varied applications. Driven by the reactive potential, pre-defined reaction pathways are not required in ReaxFF MD simulations. We believe that ReaxFF MD simulation is a potential new method to reveal the underlying complex reaction pathways occurred in the combustion of aviation fuels and the generation of soot particles.This thesis attempts a new approach to reveal chemical reaction pathways occurred in the combustion of aviation fuel and the formation of soot particles by ReaxFF MD simulations of complex molecular models of RP-1. By using the leading codes of GMD-Reax for high performance computing and VARxMD for reaction analysis, this thesis presents the comprehensive chemical reaction details revealed for intermediates and products generated in the combustion of aviation fuel and the soot formation process.Based on experimental RP-1 component data reported in literatures, a complex RP-1 molecular model containing 24 fuel components and a 3-component surrogate model were constructed. By a series of ReaxFF MD simulations for RP-1 fuel in pyrolysis and oxidation conditions, reactivity differences were discussed between the two RP-1 fuel models, especially for the heat-up simulations in the pyrolysis condition. It can be obtained that the weight fraction evolution of fuel molecules and the important product ethylene are similar between the two RP-1 models. However, RP-1 component consumption is slightly slower in the 3-component surrogate model than that of the 24-component model, and the weight fraction difference can be as much as 20% between the two RP-1 models in pyrolysis conditions. Then RP-1 fuel components in the two RP-1 models are classified into three categories as normal paraffins, branched paraffins and cyclic hydrocarbons. Cyclic hydrocarbons contribute mostly to the RP-1 evolution differences, because the cyclic hydrocarbons in the 3-component surrogate model are stable and their weight fraction is up to 43.6% that is higher than that of the other two components in the surrogate. By investigating the overall chemical reaction pathways of RP-1 fuel pyrolysis of the two RP-1 models with the aid of VARxMD, the differences of RP-1 pyrolysis reaction pathways between the two RP-1 models were revealed. For the branched paraffins, the heptamethylnonane in the 3-component surrogate contains multi-branched structure of quaternary carbon, the branched alkene of 2-methylpropene will be formed in the pyrolysis, which is not a major pyrolysis product for the 24-component model. Most of the branched paraffins in the 24-component model are characterized by tertiary carbons, their alkene pyrolysis products are linear alkenes. For the cyclic hydrocarbons, the methylcyclohexane in the 3-component model is mono-ring structure containing very short side chain, its ring-opening reaction pathways are relatively simple. However, the cyclic hydrocarbon fuel components in the 24-component model consist of double-ring structures and mono-ring structures with complex branched side chains. More versatile ring-opening pathways can be observed. This work can provide clues for further optimizing the fuel components in the 3-component surrogate model. More importantly, this work proposes a potential computational approach for evaluating chemical properties of exising surrogate fuel models.Reaction pathways for the nucleation of soot nanoparticles from polycyclic aromatic hydrocarbon (PAH) precursors were investigated by ReaxFF MD simualtions in pyrolysis conditions. A model of coronene was constructed and used as a representative PAH precursor in soot formation process. The 1 ns simulations indicate that the growth of the maximal nanoparticles observed can be promoted at 2600–3000 K with temperature increasing, but its C number decreases slightly at higher temperature conditions. In addition, the increasing of temperature contributes to the growth of maturity of the nanoparticles in terms of H/C ratio and morphology. At 3400 K, chemical reaction pathways of soot nanoparticle formation from PAHs can be observed as follows. The C–C bonds formed between PAH molecules will result in the formation of PAH dimers or even larger oligomers, which suggests the nucleation of incipient soot nanoparticles. The size of incipient nanoparticles will increase by reactions with oligomers, PAHs and acyclic molecules. The continuous size increase of the nanoparticle will result in the formation of amorphous nanoparticles with long side chains and large rings. The size of PAH sheets in the nanoparticle will increase by inter-bridging of C–C atoms inside the large rings. The decrease of length and number of side chains will occur in the nanoparticle. The continuous size decrease of these side chains will lead to well-formed soot nanoparticles. These simulation results uncover the chemical structure evolution of soot nanoparticle formation from PAHs at all-atom level, which help for enriching the knowledge of the reaction pathways for soot formation.Based on the constructed 24-component RP-1 fuel models, the overall scenario of reaction pathways of soot nanoparticle formation was investigated by performing the long time 5–10 ns ReaxFF MD simualtions of RP-1 oxidation at fuel-rich conditions. The soot nanoparticle formation obtained from the simulations can be divided into three stages. The first stage is characterized with incipient ring formation from fuel molecules and the ring growth. The incipient ring formation was initiated by ring closure reactions of long acetylene-like hydrocarbon chains generated from dehydrogenation of fuel species. Then PAH-like molecules with side chains can be formed by bridged C–C bonds of the incipient ring structures, leading to the ring condensation and resulting in the increase of ring number. The reaction pathways for the formation of PAH-like molecules from fuel molecules obtained in this work may suggest new ring formation pathways beyond the well-known HACA mechanism of PAH generation. In the second stage, nucleation can be observed through the C–C bond formation reactions of side chains between two adjacent PAH-like molecules. Then large rings form by further C–C bonding between side chains of the two PAH-like molecules. The continuous ring condensation through the internal bridging between carbon atoms of the formed large rings leads to the born of new and large PAH-like core. This observation indicates that the side chains of PAH-like structures play significant role in the nucleation of soot nanoparticles. The third stage corresponds to the graphitization of the soot nanoparticles. Slow reorganization occurring in the whole carbon skeleton of the soot nanoparticles can be observed in this stage from irregular shaped nanoparticles to well-formed fullerenic nanoparticles, through the long time transformation reactions from C5 and C7 rings into C6 rings via the intermediates of C3 rings. Systematic investigations on the simulation condition effects of temperature (1500–4000 K), equivalence ratio (0.5–10), and density of 0.01–0.5 g/cm3 were also carried out by performing long time ReaxFF MD simulations up to 5 ns. The work reveals the overall scenario of molecular structures and chemical reactions in soot nanoparticle formation from fuel molecules in fuel rich RP-1 oxidation condition at full atom simulations, which is of help for enriching the knowledge of reaction details of soot formation mechanism.In this thesis, the ReaxFF MD simulations of constructed multi-component RP-1 molecular models, RP-1 surrogate models, and PAH precursor model compound were performed. Comprehensive reaction pathway details were revealed for thermal decomposition of RP-1 and for the full view scenario of soot nanoparticle formation in fuel-rich oxidation conditions from the reactions of RP-1 fuel molecules or PAH precursors. Particularly, the reactivity difference comparison strategy proposed in this work is promising for a new method to evaluate the chemical properties of existing surrogate fuel models. This work demonstrates that ReaxFF MD simulations can provide a promising alternative approach for enriching the available knowledge of complex reaction pathway details in RP-1 combustion and soot nanoparticle formation at molecular levels.