Electrothermal methods, such as flash Joule heating (FJH), have recently become a popular strategy for the scalable synthesis of nanoparticles, 2D crystals, and other high-entropy materials. All of these endeavors would benefit from a deeper understanding of the FJH method. However, FJH invokes a large and complex parameter space. The yield and structure of the target material may be affected by the interplay between many variables, including the batch size, reaction time, electrical profile of the flash, and the properties of the starting material. Here, we employ interpretable machine learning models to explore the parameters that drive crystal formation in FJH. Several “boosted trees” machine learning models were constructed and trained to predict the nanostructure and yield of flash-synthesized graphene from various feedstocks of amorphous C. Preliminary results have revealed the importance of the flash's charge density as a predictor of yield, as well as the importance of the pulse width and current profile in controlling the turbostraticity of the resultant graphene. These results illustrate the power of machine learning as a scientific tool to analyze complex nanomanufacturing processes. Furthermore, our findings will inform the design of new strategies for the synthesis of 2D crystals with desirable properties by FJH.

Machine learning guided synthesis of turbostratic graphene by flash joule heating

ACS Spring Meeting 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.