Researchers at the University of Toronto are using artificial intelligence (AI) to accelerate the development of sustainable energy solutions. With the help of the University of Saskatchewan’s Canadian Light Source, the team validated an AI-designed catalyst that they say efficiently produces hydrogen fuel.
Green hydrogen is produced by passing electricity from renewable sources between two metal electrodes in water, releasing hydrogen and oxygen gases.
However, this process currently requires a lot of electricity and consumes rare, expensive metals.
“In short, the problem is that we want renewable hydrogen, and to do that we often split water into hydrogen and oxygen. The current Nafion-based membranes are acidic, and many metals dissolve in the acidic conditions,” explains John Kitchin, a member of the research team involved in the project.
This encouraged scientists to look for an alloy that could serve as a more efficient and cost-effective catalyst.
Finding the right metal combination is usually a lengthy process of trial and error. However, artificial intelligence can simplify this search considerably.
Speaking to Canadian Light Source, researcher Dr. Jehad Abed said: “We are talking about hundreds of millions or billions of alloy candidates, and one of them could be the right answer.”
The team used AI to predict the most effective metal combination for forming a catalyst.
According to CLS, the team examined over 36,000 different metal oxide combinations and conducted virtual simulations to determine which combination of ingredients might work best.
The winning candidate was a combination of ruthenium, chromium and titanium in certain proportions.
“Now that we have identified the candidate, we can go into the lab, make the candidate material and then test it in a real device,” Abed said.
At the CLS, the researchers have access to beamlines, i.e. stations that can direct a very bright beam of light into the catalyst material during the reaction.
“This not only enabled us to develop the most efficient catalyst, but also to understand important things about the underlying mechanism.”
Abed announced that the computer-recommended alloy performed 20 times better in terms of strength and durability than the team’s reference metal.
“The computer was right that this alloy is more effective and stable. This was a breakthrough because it shows that this method of developing better catalysts works,” Abed told CLS. “What a human would test for years, the computer can simulate in a few days.”
Reducing emissions with AI
AI increases sustainability in the areas of industrial gases and energy through various innovations. In chemical production, AI optimizes processes to reduce energy consumption and waste, such as in improving ammonia production at BASF.
In carbon capture, AI at NET Power maximizes carbon capture efficiency. Safety and efficiency are improved through AI-driven predictive maintenance deployed by ExxonMobil.
A study by the University of Surrey found that using AI would enable carbon capture plants to capture 16.7% more CO2 while using 36.3% less energy from the national grid in coal-fired power plants.
Companies like Siemens Gamesa are using AI to effectively integrate renewable energy into hydrogen production, while industrial gas giant Linde is using AI to reduce energy consumption in air separation plants.
AI also supports circular economy practices, such as Covestro’s CO2 recycling, optimizing the supply chains of companies like Air Products and thus reducing emissions.
Honeywell uses AI to develop energy efficient devices and Dow Chemical uses AI for real-time energy management to reduce waste.
