CDS&E: Catalytic Kinetics of Hydrocarbon Transformations from Dynamic Experimental Approaches Combined with on-line Machine Learning

催化剂和催化化学过程对于我们日常使用的产品的制造是必不可少的。催化材料和反应器的合理设计需要确定化学转化的顺序，以及将原料转化为所需产品的各个步骤的速率。该项目开发了将人工智能方法与系统工程技术相结合的计算方法，以加速发现改进的催化剂和催化过程。开发的方法将在经过充分研究的系统上得到验证，培训活动将有助于将该方法转移到其他学术和工业催化研究人员。催化机理和动力学参数识别传统上涉及通过多个实验获得稳态反应速率，产生有限数量的离散数据。瞬变反应堆实验可以提供更丰富的力学信息的时间分辨数据。气相浓度或温度的时间动态扰动表面覆盖率(S)，这将通过时间分辨红外光谱进行探测，并与在固定床反应器-红外光谱仪装置中通过质谱仪测量的气相浓度相关联。系统理论概念将改变用途，以确保在给定反应机理和实验装置的情况下，动力学参数的结构可辨识性。人工智能将探索输入搜索空间，以确保入口反应堆条件的持续激励，从而诱导出丰富的机械信息的连续数据流。因此，这项拟议的工作将结合当前的非线性系统识别方法和最先进的实验仪器来推导出一种自动化的实验设计(DOE)程序，该程序在最少的实验者监督下告知催化过程的机械模型。将在以前研究的乙烯加氢和一氧化碳氧化催化系统上进行验证。该项目的成果将是一套软件和硬件工具，以及一套定制进气扰动的综合程序，该程序将持续激发固定床反应器的动力学，以确定催化动力学参数。这一新的瞬变方法将在学术和工业应用中对催化材料和工艺的发展有用。这一瞬时方法允许快速开发机理，比较催化材料之间的本征动力学，并估计对催化反应器系统设计有用的参数。除了向催化社区提供这项技术外，还计划开展教育和推广活动，为STEM代表性不足的群体和催化社区的学生提供研究培训。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Catalysts and catalytic chemical processes are essential for the manufacture of products we use every day. The rational design of catalytic materials and reactors requires determining the sequence of chemical transformations, and the rates of individual steps, that convert raw materials to desired products. The project develops computational approaches that integrate artificial intelligence methods with systems engineering techniques to accelerate the discovery of improved catalysts and catalytic processes. The developed approach will be validated on well-studied systems, and training activities will facilitate transfer of the approach to other academic and industrial catalysis researchers.Catalytic mechanism and kinetic parameter identification traditionally involve acquiring steady-state reaction rates over multiple experiments producing a limited amount of discrete data. Transient reactor experiments can provide time-resolved data that are richer in mechanistic information. Temporal dynamics of gas-phase concentrations or temperatures perturb surface coverage(s), which will be probed by time-resolved infrared spectroscopy and correlated with gas-phase concentrations measured by mass spectrometry in an operando packed bed reactor-infrared spectrometer set-up. System theoretic concepts will be repurposed to ensure structural identifiability of the kinetic parameters given the reaction mechanism and experimental apparatus. Artificial intelligence will explore the inputs search space to ensure persistent excitation of the inlet reactor conditions that induce a continuous data stream rich in mechanistic information. The proposed effort will thus combine current nonlinear system identification methods and state-of-the-art experimental apparatuses to derive an automated Design of Experiments (DoE) procedure that informs mechanistic models of catalytic processes with minimal experimentalist supervision. Validation will be performed on previously studied ethylene hydrogenation and CO oxidation catalytic systems. The outcome of this project will be a set of software and hardware tools, and an integrated procedure to tailor inlet perturbations, that continuously excite the dynamics of operando packed bed reactors to determine catalytic kinetic parameters. This new transient approach will be useful, both in academic and industrial applications, for the development of catalytic materials and processes. This transient approach allows for rapid mechanism development, comparison of intrinsic kinetics among catalytic materials, and estimation of parameters useful for catalytic reactor system design. In addition to providing this technique to the catalysis community, educational and outreach activities are planned to provide research training to students from STEM underrepresented groups and the catalysis community.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.