Commentary

Beyond Lithium: an AI Test Case

Chris Taufatofua, Partner, Vinson & Elkins, explains how technology innovation to identify alternatives to critical minerals can accelerate the wider energy transition.

The accelerating rate of climate change and the consequent need to transition away from carbon-intensive sources of energy have led to a surge in investment into electric vehicles (EVs) and battery energy storage solutions (BESS). Lithium, an essential component in batteries used for both EVs and BESS, is already in great demand; and that demand is expected to skyrocket over the next five to ten years. However, this highly flammable material has become increasingly controversial, with the long-term negative environmental and social impacts of the lithium extraction process coming increasingly into focus. These factors, coupled with (i) the rate at which demand for lithium is expected to increase and (ii) the perceived over-reliance on China and a select few other countries for the supply of economically and strategically important minerals, including lithium and other rare earth minerals required for batteries, have led to a push for research and development into alternative materials for use in battery technologies.

To date, the search for alternative materials has not borne fruit. However, earlier this year Microsoft and the Pacific Northwest National Laboratory were able to find alternative battery materials in a matter of hours using a combination of advanced artificial intelligence (AI) and high-performance computing (HPC), showing their potential to rapidly expedite the energy transition.

The need for alternatives

Lithium has a high electrochemical potential, making it a perfectly suited component for use in high energy density rechargeable batteries. As of 2023, demand for lithium-ion batteries across EVs and BESS rose by 53% year-on-year to reach 950 gigawatts per hour (GWh); and is expected to exceed 4,500 GWh by 2030. While BESS are expected to have a compound annual growth rate of 30%, the bulk of the rise in demand for lithium-ion batteries will be driven by the demands of the burgeoning EV industry.

As things currently stand, lithium supply makes up less than 50% of the projected demand for 2030. Theoretically, there are enough global lithium reserves to supply this projected demand. However, current mining and extraction processes cannot easily be scaled, and projects have uncertain lead times. For example, one method of lithium extraction involves brine being pumped to the surface and into evaporation ponds, where the liquid brine is left to evaporate to yield lithium. Such projects can take up to eight years to reach commercial production.

These forecasts indicate a significant supply gap, with this scarcity potentially making lithium-ion batteries an attractive investment opportunity. But production lags are far from the only challenge. The extraction of lithium has the potential to be carbon-intensive and environmentally harmful, with a significant proportion of lithium currently mined in regions where ethical mining practices have not yet been established.

Carbon-intensive and unethical

Lithium extraction can take place through several processes. Traditional hard rock mining releases 15 tonnes of CO2 for every tonne of lithium mined. Alternatively, extraction from underground reservoirs releases a slightly more conservative — but by no means environmentally friendly — five tonnes of CO2 per tonne of lithium mined. However, this alternative process is significantly more water-intensive, using multiple millions of litres of water to extract the equivalent tonnage of lithium needed for just a single EV, not to mention the resulting groundwater contamination.

Indeed, in Asia’s lithium capital — China’s Yichun City — lithium production was forced to halt back in 2022 amid an investigation into the abnormal water quality of the Jin River. This followed a series of incidents in Tagong, China, where a toxic chemical leak from the Ganzizhou Rongda Lithium mine found its way into the Lichu River, resulting in masses of dead fish and farm animals, devastating the local ecosystem. Groundwater contamination has also meant the area is no longer suitable for agriculture, impacting the livelihood of the indigenous people in the surrounding areas.

The perceived over-reliance on China for the supply of economically and strategically important minerals such as lithium has seen western economies introduce legislation aimed at reclaiming control of critical raw material supply and manufacturing capacity. However, given the time needed to scale lithium extraction projects, it is highly unlikely that domestic production in western economies will be able to meet expected lithium demand in the coming years.

Consequently, there is a pressing need to find a commercially viable alternative to lithium that is sustainable and ethical.

New potential battery material

The issues discussed above have led companies like Microsoft to try and find novel methods to source viable alternatives to lithium. Using an AI model combined with HPC, Microsoft was able to take 32 million potential materials and narrow them down to 18 viable alternative combinations in just 80 hours. Factors that were filtered for included, amongst others, stability of chemical composition, energy density, availability, and cost.

From the candidates, scientists were able to develop a working prototype, with the entire end-to-end process taking less than nine months — a process that would take decades in a typical research laboratory. The alternative currently being evaluated uses a combination of sodium and lithium, reducing lithium content by 70% compared to typical lithium-ion batteries. It was previously thought lithium and sodium could not be used together.

Scientists have flagged that it is not the material discovered (which is yet to be proven at scale), but the speed of the discovery that represents the most significant breakthrough. Krysta Svore, who leads the Microsoft quantum computing team at Microsoft Research, stated that in order to reach net zero by 2050, “we need to really compress the next 250 years of chemistry material science into the next two decades”.  Microsoft’s groundbreaking use of AI and HPC represents a positive step toward achieving that goal.

Wider application

While this article is focused on lithium, it will be interesting to see how AI and HPC are deployed to find alternatives to other materials and scarce metals/minerals. For example, 95% of solar PV modules use a material called polysilicon; as of 2021, 45% of polysilicon was being produced in the Uyghur region in China, where it has been linked to forced labour regimes. A number of alternatives to polysilicon are currently being assessed, and methods such as those used by Microsoft could help to quickly identify materials that are less carbon intensive, more commercially suitable, and reduce reliance on countries where there have been proven instances of forced labour practices in the supply chain.

A shifting landscape

As we have discussed, the rate at which demand for lithium is expected to increase over the next decade, the environmental and social impacts of the lithium extraction process, and the over-dependence on China for supply mean that Microsoft’s employment of AI and HPC to identify a viable alternative to lithium represents a welcome breakthrough in innovation.

While the use of AI and HPC may expedite the search for alternatives to other economically and strategically important minerals critical to the energy transition, there may be other, unintended consequences. For example, the speed at which AI and HPC can analyse vast datasets and simulate an infinite number of scenarios at unimaginable speeds could see the maturity lifecycle of new energy transition technologies compress. Currently, inventions in the energy sector can take between two to three decades to reach mass market from inception. Use of AI and HPC to accelerate the rate of innovation may create a risk for developers, lenders and other project participants that their technology may be surpassed by the time they are able to effectively get it to market, as the rate of innovation outstrips the timeline for developing their project.

In harnessing the power of AI and HPC, Microsoft’s breakthrough underscores the potential of these technologies to expedite innovations in the energy transition, though it also raises questions about the implications for project developers looking to get their technologies out to market.

This article was co-authored by Afzaal Abidi, Associate, and Ellen Swarbrick, Trainee Solicitor, Vinson & Elkins.

The practical information hub for asset owners looking to invest successfully and sustainably for the long term. As best practice evolves, we will share the news, insights and data to guide asset owners on their individual journey to ESG integration.

Copyright © 2024 ESG Investor Ltd. Company No. 12893343. ESG Investor Ltd, Fox Court, 14 Grays Inn Road, London, WC1X 8HN

To Top
Share via
Copy link
Powered by Social Snap