Ten billion. It is how many commercially available molecules are available today. Start looking at them in groups of five – the typical combination used to make electrolyte materials in batteries – and it goes up to 10 to the 47th power.
For those counting, that’s a lot.
All of these combinations matter in the world of batteries. Find the right mix of electrolyte materials and you can end up with a faster-charging, higher-energy-density battery for the EV, the grid, or even an electric airplane. The downside? Like the drug discovery process, it can take more than a decade and thousands of failures to find the right solution.
That’s where the founders of startup Aionics say their AI tools can speed things up.
“The problem is that there are too many candidates and not enough time,” Aionics co-founder and CEO Austin Sendek told TechCrunch at the recent Up Summit event in Dallas.
Electrolytes, meet AI
Lithium-ion batteries contain three critical building blocks. It has two electrodes, an anode (negative) on one side and a cathode (positive) on the other. The electrolyte is usually found in the middle and acts as a messenger to move ions between the electrodes during charging and discharging.
Aionics focuses on the electrolyte and uses a suite of artificial intelligence tools to accelerate discovery and ultimately deliver better batteries. Aionics approach to catalyst discovery also attracted investors. The Palo Alto-based startup, which was founded in 2020, has raised $3.5 million to date, including a $3.2 million seed round from investors including UP.Partners.
The startup already works with several companies, including Porsche’s battery-making subsidiary Cellforce. The company has also worked with energy storage firm Form Energy, Japanese materials and chemicals maker Showa Denko (now Resonac) and battery company Cuberg.
This whole process starts with the company’s wish list — or performance profile — for a battery. Aionics scientists, using AI-accelerated quantum mechanics, can run experiments on an existing database of billions of known molecules. This allows them to screen 10,000 applicants every second, Sendek said. This AI model learns how to predict the outcome of the next simulation and helps select the next candidate molecule. Each time it runs, more data is generated and it gets better at solving the problem.
Enter generative AI
Aionics has taken this a step further, in some cases by bringing generative AI into the mix. Instead of relying on the billions of known molecules, Aionics this year began using generative AI models trained on existing data about battery materials to create or design new molecules aimed at a specific application.
The company topped the effort by using software developed in the Accelerated Computational Electrochemical Systems Discovery Program at Carnegie Mellon University. Venkat Viswanathan, who was an associate professor at CMU and led this program, is co-founder and chief scientist at Aionics.
Aionics has also started using large language models built on GPT 4 from OpenAI to help its scientists sift through the millions of possible wordings before they even start putting them into the database. This chatbot tool, which is trained on chemistry textbooks and scientific papers selected by Aionics, is not used for the actual discovery, but can be used by scientists to eliminate certain molecules that would not be useful in a particular application, Sendek explain.
Once trained with these textbooks, LLMs allow the scholar to query the model. “If you could talk to your textbook, what would you ask it?” Sendek said. But he’s quick to point out that it’s no different than someone who curates scholarly articles. “It’s just providing next-level interaction,” he said, adding that everything can be verified by pointing to 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 as he explained the chatbot’s function.
Picking a winner
After the billions of candidates have been screened and narrowed down to just a pair—or designed using the generative AI model—Aionics sends samples to its customers for validation.
“If we don’t get to the first round, we iterate and we can do some clinical trials to prove it until we get to the winner,” Sendek said. “And once we find the winner, we work with our manufacturing partners to scale up that production and bring it to market.”
Curiously, this process is even being used in some new fields such as cement. Chement, a startup co-founded by Viswanathan and also partnered with Aionics, is working on ways to use renewable electricity and raw materials to drive chemical reactions to produce zero-emissions products like cement.