Computational Models Used to Advance Environmentally Friendly Processes — ScienceDaily
Rice University scientists are using machine learning techniques to streamline the process of synthesizing graphene from waste by Joule flash heating.
The process discovered two years ago by chemist James Tour’s Rice Lab has extended beyond making graphene from various carbon sources to extracting other materials like metals from municipal waste, with the promise of more environmentally friendly recycling to come.
The technique is the same for all of the above: send a high-energy jolt through the source material to remove all but the desired product. But the details for flashing each raw material are different.
The researchers describe in Advanced materials how machine learning models that adapt to variables and show them how to optimize procedures help them move forward.
“Machine learning algorithms will be key to making the flash process fast and scalable without negatively affecting the properties of the graphene product,” Tour said.
“In future years, flash parameters may vary depending on the feedstock, whether it’s petroleum-based, coal-based, plastic-based, household waste, or whatever,” he said. “Depending on what type of graphene we want – small flake, large flake, high turbostrate, level of purity – the machine can discern on its own which parameters to change.”
Because flash produces graphene in hundreds of milliseconds, it’s difficult to disentangle the details of the chemical process. Tour and his company have therefore taken inspiration from materials scientists who have incorporated machine learning into their daily discovery process.
“It turned out that machine learning and Joule flash heating had a really good synergy,” said Jacob Beckham, Rice’s graduate student and lead author. “Flash Joule heating is a very powerful technique, but it is difficult to control some of the variables involved, such as the rate of current discharge during a reaction. And this is where machine learning can really shine. is an excellent tool for finding relationships between multiple variables, even when it is not possible to do a full search of the parameter space.
“This synergy allowed graphene to be synthesized from waste entirely based on understanding models of the Joule heating process,” he said. “All we had to do was do the reaction – which can potentially be automated.”
The lab used its custom optimization model to improve the crystallization of graphene from four starting materials – carbon black, plastic pyrolysis ash, pyrolyzed rubber tires and coke – over 173 trials, using Raman spectroscopy to characterize graphene starting materials and products.
The researchers then fed more than 20,000 spectroscopy results to the model and asked it to predict which starting materials would provide the best yield of graphene. The model also considered the effects of charge density, sample mass and material type in its calculations.
Co-authors are Rice graduate students Kevin Wyss, Emily McHugh, Paul Advincula, and Weiyin Chen; Rice alumnus John Li; and postdoctoral researcher Yunchao Xie and Jian Lin, associate professor of mechanical and aerospace engineering, from the University of Missouri. Tour holds the TT and WF Chao Chair in Chemistry as well as a professor of computer science, materials science and nano-engineering.
The Air Force Office of Scientific Research (FA9550-19-1-0296), US Army Corps of Engineers (W912HZ-21-2-0050), and Department of Energy (DE-FE0031794) supported the research.
Source of the story:
Material provided by rice university. Original written by Mike Williams. Note: Content may be edited for style and length.