Creation of complex chemical mechanisms for hydrocarbon pyrolysis and combustion is challenging due to the large number of species and reactions involved. Reactive molecular dynamics (RMD) enables the simulation of thousands of reactions and the discovery of previously unknown components of the reaction network. However, due to the inherent imprecision of reactive force fields, it is necessary to verify RMD‐obtained reaction paths using more accurate methods such as Density Functional Theory (DFT). We demonstrate a method for identification and confirmation of reaction pathways from RMD that supplement an established mechanism, using the example of benzene formation from n‐heptane and iso‐octane pyrolysis. We establish a validation workflow to extract reaction geometries from RMD and optimize transition states using the Nudged‐Elastic‐Band method on semi‐empirical and quantum mechanical levels of theory. Our findings demonstrate that the widely recognized ReaxFF parameterization, CHO2016, can identify known pathways from a established soot formation mechanism while also indicating new ones. We also show that CHO2016 underestimates hydrogen migration barriers by up to as compared to DFT and can lower activation barriers significantly for spin‐forbidden reactions. This highlights the necessity for validation or potentially even reparametrization of CHO2016.

中文翻译：

---

碳氢化合物热解分子动力学模拟的反应路径识别和验证

由于涉及大量物质和反应，创建碳氢化合物热解和燃烧的复杂化学机制具有挑战性。反应分子动力学 (RMD) 能够模拟数千种反应并发现反应网络中以前未知的成分。然而，由于反作用力场固有的不精确性，有必要使用密度泛函理论（DFT）等更准确的方法来验证RMD获得的反应路径。我们以苯形成的例子展示了一种识别和确认 RMD 反应途径的方法，该方法补充了既定机制n‐庚烷和异‐辛烷热解。我们建立了一个验证工作流程，从 RMD 中提取反应几何形状，并在半经验和量子力学理论层面上使用微推弹性带方法优化过渡态。我们的研究结果表明，广泛认可的 ReaxFF 参数化 CHO2016 可以从已建立的烟灰形成机制中识别已知的途径，同时也可以指示新的途径。我们还表明，与 DFT 相比，CHO2016 低估了氢迁移势垒，并且可以显着降低自旋禁止反应的活化势垒。这凸显了对 CHO2016 进行验证甚至重新参数化的必要性。