Rebranding carbon

What if there was a way to close the carbon loop by making fossil fuels renewable? Dr Ziyun Wang and his academic group at the University of Auckland, the Theoretical Catalysis (TCAT) Group, are working to harness the power of catalysts to make it possible.

Wang explains a catalyst, in chemistry, as a hand fitting Lego pieces together. Its role is to cause something new to happen with chemicals, something novel, something potentially valuable to science. Like Lego, it’s a game of trial and error. This is where Artificial Intelligence (AI) and machine learning can help.

Wang is fortunate to have a team to experiment with solutions for what he calls the ‘CO₂ problem’. The TCAT Group comprises around 15 University staff and postgraduate students and their diverse backgrounds allow for fresh takes on complex problems.

By experimenting with catalysts, Wang and his team have made progress converting carbon dioxide (CO₂) to formic acid, a liquid which has broad usage in the chemical industry and potential as a fuel. They did this by taking a used lead-acid battery, from an Internal Combustion Engine (ICE) vehicle, and extracting the lead element for use as the catalyst that enabled the conversion.

If formic acid in fuel-use comes to fruition, the fuel-cell vehicles that could potentially use it would employ much smaller, lightweight Li-ion batteries. Rather than being the sole source of power for the vehicle, like the much larger batteries in the now widespread Battery Electric Vehicles (BEVs), these act only as a support to the fuel-cell system. In a win for the environment, it would mean less extraction of minerals, like lithium, cobalt and nickel, in the battery’s production and less environmental impact in its end-of-life disposal.

The other side of this potentially two-pronged solution is the reuse pathway it could open for electrical waste, like the lead batteries Wang experimented with. Though not present in most modern EVs, they are still commonly used in e-bikes in many parts of the world, particularly developing countries, and in some lower-end EVs.

Scientists have long grappled with the problem of closing the carbon loop: achieving net zero emissions by reusing CO₂, rather than letting it accumulate in the atmosphere. Finding a productive use for CO₂ has proved challenging, because its level of volatility makes it notoriously difficult to work with. Formic acid, as a compound, has a good level of stability.

Testing out endless catalysts in the lab is time-consuming work. This is where Wang’s work is different, by harnessing the power of AI and machine learning. Sometimes, identifying a suitable catalyst involves detective work – drilling down through a vast range of possible states or energy levels for a particular element or material. Figuring this out takes time. “We may know that lead works well, but we don’t know which type specifically – should we use lead carbonate, or something else,” says Wang.

Traditional lab work requires manual processes, trial and error. With machine learning, a calculation can be run with hundreds of thousands, even millions of combinations at the same time. The use of AI to learn from previous data also drastically improves precision: by selecting the perfect catalyst, the resulting reaction will perform at exactly the level it needs to, thus wasting less energy in the process.

Wang and his team are currently working on how to make widespread carbon capture a reality at the industrial level. Capturing pure CO₂ is an energy intensive process, but understanding the use of industrial flue systems (which contain a mixture of other chemicals too, including sulphur) to achieve the same purpose is now virtually assured. Thus, the CO₂ to formic acid initiative now primarily relies on industry effort. Scientifically, the exploring of new products from CO₂ would be the next logical research goal.

A fascination with all things technical has been a lifelong thread for Wang. As a teenager, he loved computer games and diving into virtual worlds. “I wanted to play computer games all day every day,” he says, and even thought about becoming a professional gamer. He later realised that computational chemistry was essentially a computer game to solve a real-world problem.

“Our system and a computer game are actually very similar. The way we experiment with different catalysts is like tweaking design layouts and managing budgets in SimCity.

“In both cases, you’re constantly looking at cause-and-effect dynamics – a computer algorithm gives you feedback about whether you’re doing well or not.”

The combined strengths of the TCAT team are key to exploring the opportunities that present themselves. “Some have a background in pure Physics, some Chemistry and others Computer Science,” says Wang. “Everyone brings something to the table. One member works for a large IT company and brings experience from the industry side; others are doing a lot of theoretical work.” The team clearly works in the here and now but has a foot firmly in the future – daring to look ahead to what might develop.

This future focus is part and parcel of cutting-edge science. “One thing about machine learning is that we always underestimate how fast it will move and how impactful it will be,” says Wang, citing how critical people were of ChatGPT just three years ago and how underestimated it was. This is why he finds it such an exciting field.

“Everything is moving very fast and we’re very lucky to be in the game earlier than most people.”