Researchers use AI to trace the greenway to nylon

The chemical and related industries face challenges such as easy access to reliable energy sources, waste reduction, water conservation and energy efficiency. Organic electrosynthesis, an energy-efficient and energy-efficient process that can easily integrate with renewable energy sources, could help to solve them.

A team from the NYU Tandon School of Engineering has announced that it has developed an innovative and environmentally friendly process to manufacture adiponitrile (DNA), the main precursor of nylon 6 and 6, and to find a way to Significantly improve the efficiency of organic electrosynthesis. . The researchers attributed some of their success to what they consider to be the first use of artificial intelligence to optimize an electrochemical process.

Miguel Modestino, professor of chemical and biomolecular engineering, and PhD student Daniela Blanco have refined the way electrical current is routed to catalytic electrodes, and then applied to artificial intelligence to optimize the reaction. In doing so, they achieved a 30% improvement in DNA production

The detailed results in the Proceedings of the National Academy of Sciences (PNAS), could have major implications since the team is targeting one of the most important organic electrosynthesis processes in the chemical industry: the electrohydrodimerization of acrylonitrile (AN) into DNA.

The demand for DNA is high and growing: the nylon market is expected to increase by 4% per year until 2023. Only one company currently uses an electrochemical process invented by Monsanto to manufacture DNA; the lion's share of the nylon precursor comes from a toxic and energy-intensive thermal hydrocyanation of butadiene. In contrast, DNA electrosynthesis is a green and efficient chemical process that uses water-based electrolytes and can be directly coupled to renewable electricity sources such as wind or sun.

The standard electrosynthetic process for DNA uses a continuous "continuous" electric current sent to the electrocatalytic site. NYU Tandon researchers, however, found that a direct current did not maximize DNA production and generated a large proportion of propionitrile (PN), an undesirable by-product. They decided to design a system providing an intermittent current in order to continually replenish the reagent concentration at the electrocatalytic site (a phenomenon called mass transport) and to improve the flow of the DNA.

The pair provided an artificial neural network with data from 16 different experimental cases of pulse time.

"By analyzing electrochemical pulse techniques with AI, we were able to visualize the efficiency of DNA conversion over a range of pulse time without having to do more than just a few physical experiments. "said Modestino. "This innovative and integrated approach has led to an unprecedented 30% increase in DNA production and a 325% increase in DNA-to-PN ratio, mainly due to a significant drop in production. of the last."

Blanco explained that this technique could advance the sector in the adoption of more sustainable processes. That's exactly what she and a former Modestino laboratory student had imagined when they founded a young green chemistry company, Sunthetics, to commercialize a sustainable nylon production process based on their research.

"With this new study, we wanted to show that we could make the electrochemical DNA process more competitive," she said. "At present, only 30% of global DNA production uses electrosynthesis, the rest of the production involves treatment on a catalytic reactor with high energy and oil intensity", a- she declared.

The next step for the team will be to use this artificial intelligence approach to accelerate their research efforts. "Instead of using a classic research model involving long experimental campaigns, artificial intelligence tools can help us predict the experimental results." To the best of our knowledge, this is the first time that AI is used to optimize an electrochemical process, "said Modestino.