A corpus of CO2 electrocatalytic reduction process extracted from the scientific literature

Abstract

The electrocatalytic CO2 reduction process has gained enormous attention for both environmental protection and chemicals production. Thereinto, the design of new electrocatalysts with high activity and selectivity can draw inspiration from the abundant scientific literature. An annotated and verified corpus made from massive literature can assist the development of natural language processing (NLP) models, which can offer insight to help guide the understanding of these underlying mechanisms. To facilitate data mining in this direction, we present a benchmark corpus of 6,086 records manually extracted from 835 electrocatalytic publications, along with an extended corpus with 145,179 records in this article. In this corpus, nine types of knowledge such as material, regulation method, product, faradaic efficiency, cell setup, electrolyte, synthesis method, current density, and voltage are provided by either annotating or extracting. Machine learning algorithms can be applied to the corpus to help scientists find new and effective electrocatalysts. Furthermore, researchers familiar with NLP can use this corpus to design domain-specific named entity recognition (NER) models.

CO2 Reduction beyond Copper-Based Catalysts: A Natural Language Processing Review from the Scientific Literature

“…In fact, they are the most commonly used in each of these conditions, with KHCO 3 being the most frequent electrolyte overall. Additionally, we conducted a comparison of Figure b with the manually extracted results published in ref , which also focused on the same objective: electrolytes employed in CO 2 ER. Notably, the top five electrolytes identified in their analysis are also present in our top five, albeit in a different order.…”

“…This rigorous evaluation helped to ensure the accuracy and effectiveness of our strategy. It is worth noting that similar approaches have been attempted recently in the field of the CO 2 RR, but with a narrower focus on Cu-based catalysts for CO 2 ER and relying on only a few articles , or using these NLP tools in a general fashion . Unlike these previous analyses, our approach considers a larger corpus and has a broader scope: CO 2 catalysts beyond Cu-based materials.…”

“…This decision was based on heuristics since details about reaction conditions are typically described in this section and oftentimes dismissed in the abstract unless the work emphasizes the impact of electrolytes on CO 2 ER. To validate our approach, we utilized the manually extracted data from ref . By comparing our results to their analysis, we were able to assess the accuracy and reliability of our methodology.…”

Large language model enhanced corpus of CO2 reduction electrocatalysts and synthesis procedures

“…Furthermore, NER algorithms are closely tied to their assigned tasks, lacking generalizable ability and thus making direct transfer to other tasks challenging. Additionally, extracted information tends to be intricate, heterogeneous, and diverse in the field of catalysis, leading to unsatisfied NER performance and reduced accuracy 12 . Therefore, the development and utilization of more general and robust methods for extracting domain knowledge are becoming increasingly imperative.…”

“…In recent work, our team has developed a text-mining pipeline to construct a dataset describing the CO 2 reduction process catalyzed by copper-based electrocatalysts, which specifically includes material, regulation method, product, Faradaic efficiency and relevant conditions 12 . In the current work, we built a more advanced extraction pipeline based on the knowledge system of CO 2 electrocatalytic reduction (Fig.…”