10 billion.it is How many molecules are currently commercially available? When he begins to examine them in groups of five (typical combinations used to make battery electrolyte materials), the number increases to the 10 to the 47th power.
For those keeping count, that’s a lot.
In the world of batteries, the combination of all these things is important. Finding the right mixture of electrolyte materials can lead to faster-charging, more energy-dense batteries for EVs, power grids, and even electric airplanes. What are the drawbacks? As with the drug discovery process, finding the best fit can take more than a decade and thousands of failures.
That’s where startup founders are. Ionics They say their AI tools can speed things up.
“The problem is there are too many candidates and not enough time,” Ionics co-founder and CEO Austin Sendek told TechCrunch during a recent AppSummit event in Dallas. told.
Dr. Renson Perchaud, Co-Founder and Chief Technology Officer.Dr. Austin Sendek, Co-Founder and CEO and Dr. Venkat Viswanathan, Co-Founder and Chief Scientist image credits: Ionics
The encounter between electrolytes and AI
Lithium-ion batteries contain three important components. There are two electrodes, an anode (negative) on one side and a cathode (positive) on the other side. Typically, the electrolyte is centrally located and acts as a carrier that moves ions between electrodes during charging and discharging.
Ionics focuses on electrolytes and uses an AI toolkit to accelerate discoveries and ultimately deliver better batteries. Ionics’ approach to catalyst discovery is also appealing to investors. Founded in 2020, the Palo Alto-based startup has raised $3.5 million to date, including a $3.2 million seed round from investors including UP.Partners.
The startup has already partnered with several companies, including Porsche’s battery manufacturing subsidiary Cellforce. The company has also worked with energy storage company Form Energy, Japanese materials and chemicals maker Showa Denko (now Resonac), and battery technology company Qberg.
This entire process starts with a company’s wish list (or performance profile) for the battery. Ionics scientists can use AI-accelerated quantum mechanics to perform experiments on existing databases of billions of known molecules. This allows the company to consider 10,000 candidates every second, Sendek said. His AI model learns how to predict the outcome of the next simulation and helps select the next molecule candidate. Each run generates more data and improves problem solving.
Introducing generative AI
Ionics takes this a step further by, in some cases, incorporating generative AI into the mix. This year, Ionics announced the launch of generative AI models trained on existing battery materials data to create or design new molecules targeted for specific applications, instead of relying on billions of known molecules. I have started using it.
The company is using software developed by Carnegie Mellon University’s Accelerated Computational Electrochemical Systems Discovery program to enhance its efforts. Venkat Viswanathan, an associate professor at CMU who led the program, is Aionics’ co-founder and chief scientist.
Aionics has also started using large language models built on OpenAI’s GPT 4 to help scientists narrow down millions of possible formulations before they start running them in the database. I am. This chatbot tool is trained on Ionics’ selection of chemistry textbooks and scientific papers, and while it is not used for actual discovery, scientists can eliminate certain molecules that are not useful for a particular application. , Sendek said. He explained.
After training with these textbooks, scientists can use LLM to query models. “If you could talk to a textbook, what question would you ask?” Sendek said. But he is quick to point out that this is no different than someone curating a scientific paper. “This is just providing the next level of interaction,” he said, adding that everything is verifiable by looking at the sources used to train the chatbot.
“I think what’s good for our field is that we’re not looking for specific facts, we’re looking for design principles,” he said while explaining chatbot capabilities.
pick a winner
Once billions of candidates have been screened and narrowed down to just two or designed using a generative AI model, Aionics sends customer samples for validation.
“If you don’t make it to the first round, you can iterate until you reach a winner and do some clinical trials to prove it,” Sendek said. “And once we find a winner, we work with our manufacturing partners to scale its production and bring it to market.”
Interestingly, this process is also used in some new fields such as cement. KementThe startup, co-founded by Viswanathan and also in partnership with Ionics, is working on ways to use renewable electricity and raw materials to accelerate chemical reactions and make zero-emission products such as cement. .