The right elements aren’t enough: steel, magnets and lightweight alloys

Decarbonisation, supply chain security and energy efficiency are usually discussed as policy or process problems. Underneath each of them sits a materials problem.

For example, making steel cleaner changes what ends up in it. Replacing rare earths in magnets means finding new atomic arrangements that work. Making vehicles lighter means balancing several elements in one crystal lattice. In this blog post, we look at three material families: low-carbon steel, rare-earth-free magnets and lightweight alloys. All three lead to the same conclusion:

Changing what a material is made of, or how it is made, changes what it can do. The challenge is not only finding the right elements. It is knowing how they will arrange themselves, how that arrangement behaves, and whether there is a realistic route to make it.

Low-carbon steel: cleaner process, different material

Steel is the backbone of modern infrastructure. The industry estimates it is responsible for 7 to 9% of global CO₂ emissions. Cutting those emissions means changing how steel is made. One route is to use more recycled scrap, melted in electric arc furnaces. But every change to the process also changes the material.

Scrap carries leftover elements, mainly copper, that conventional refining can’t remove. During hot rolling, that copper can collect along the boundaries between the steel’s grains, melt and cause cracking. This defect is known as hot shortness. Elements like tin make it worse.

So the question for steelmakers isn’t only “can we make it cleaner?” It is also “will it still perform the same?” Answering that means knowing how small shifts in composition move a steel’s structure and properties, ideally before a full production trial.

Rare-earth-free magnets: the right atoms in the right order

The strongest permanent magnets in electric motors and wind turbines depend on rare-earth elements. The supply chain behind them is highly concentrated. According to the IEA, China accounts for 59% of rare-earth mining and 84% of refining for the rare earths used in permanent magnets.

Replacing those elements is not a simple swap, because a magnet’s performance depends on how its atoms are arranged in the crystal. Take tetrataenite, a compound of iron and nickel. Both elements are abundant, and it is considered a promising rare-earth-free magnet. But it only works when the iron and nickel atoms sit in a precise, alternating order. In nature, it is found in trace amounts in iron meteorites. In the lab, that ordered structure is extremely hard to reach. At the temperatures where it is stable, atoms move so slowly that conventional heat treatment would take “an astronomical amount of time”.

So, the right elements aren’t enough, they need the right arrangement, and a realistic route to get there.

Lightweight alloys: trade-offs in every composition

Every kilogram removed from a vehicle saves energy over its whole life. The US Department of Energy estimates that a 10% reduction in vehicle weight can improve fuel economy by 6 to 8%.

That’s why aluminium, magnesium and advanced high-strength steels keep gaining ground. But lightweighting is never as simple as swapping one metal for a lighter one. Strength trades against formability, performance against cost, new compositions against proven ones.

Underneath every one of those trade-offs is the same question: how several elements share one crystal lattice, and what that arrangement does to the final part. Change the ratio slightly, or the way it is processed, and the properties can shift a lot.

The common thread: composition, structure and route

Cleaner steel, rare-earth-free magnets and lighter alloys all come back to how atoms arrange themselves, and to whether that arrangement can be made reliably. Exploring these options by hand, one composition and one production trial at a time, takes years. Physics-based simulation can predict how a composition will behave before anything is cast or rolled. AI can then search those predictions much faster than trial and error. The point is to spend lab time on the candidates most worth testing.

Physics first, AI on top: that’s how we approach the problem at PhaseTree.