Ten Recommendations for Literature Data Generation for Chemical Process Optimization in Renewable Feedstock Conversion

Process development of recyclable and renewable feedstocks entails intensive and tedious catalyst and solvent screening and multiparameter process exploration. The solid nature of these feedstocks necessitates highly time-consuming, batch experiments, leading to limited published datasets. Our ability to rapidly advance process development requires a systematic approach to the simultaneous optimization of catalysts, solvents, and processing conditions. One approach toward this goal is to harness and analyze accumulated literature data to develop performance-process correlations. With this goal, we study a prototype biomass reaction, the fructose dehydration to 5-hydroxymethyl furfural (HMF) over zeolite catalysts. We explore literature data to optimize the vast descriptor space of zeolite characteristics, solvents, and process conditions using correlation analysis. We assess the quality of reported literature data and identify where provenance is lacking. We develop apparent kinetic analysis to correlate performance with solvent and catalyst properties. We showcase that solvents and to an extent processing conditions dominate HMF yield; in contrast, zeolites play a secondary role for this chemistry. We provide ten recommendations for improving this methodology to extract further knowledge and improve process optimization.可回收与可再生原料的工艺开发，需要开展高强度且繁琐的催化剂与溶剂筛选，以及多参数工艺探索。这类原料的固态属性，使得实验必须采用耗时极久的间歇式操作，进而导致已公开的数据集十分有限。我们若要快速推进工艺开发，就需要一套能够同时优化催化剂、溶剂与工艺条件的系统性方法。实现这一目标的一种途径，是利用并分析已积累的文献数据，构建性能与工艺之间的关联模型。基于此目标，我们研究了一个典型的生物质反应：在沸石（zeolite）催化剂作用下，果糖脱水生成5-羟甲基糠醛（5-hydroxymethyl furfural，HMF）。我们通过文献数据挖掘，利用关联分析方法，对沸石特性、溶剂与工艺条件的庞大描述符空间进行优化。我们对已报道的文献数据质量进行评估，并识别出数据溯源缺失的环节。我们构建了表观动力学分析模型，以关联反应性能与溶剂、催化剂的性质。我们的研究结果表明，溶剂以及一定程度上的工艺条件，是影响HMF收率的主导因素；相比之下，沸石在该反应体系中仅起到次要作用。我们针对该研究方法提出了十条改进建议，以期进一步挖掘其中蕴含的知识，并优化工艺开发流程。