Method combines quantum mechanics with machine learning to accurately predict oxide reactions at high temperatures when no experimental data is available; could be used to design clean carbon-neutral processes for steel production and metal recycling.
