Tritium is scarce. It basically isn’t there.
We need it for nuclear fusion, obviously. The donut-shaped tokamak reactors run on a mix of deuterium and tritium. Deuterium? Stable, everywhere. Tritium? Radioactive, fleeting, nearly invisible on Earth except for those trace bits born from cosmic rays in our upper atmosphere.
When those two hydrogen isotopes fuse, the payoff is nice. One helium nucleus, a free neutron, 17.6 million electronvolts of energy released. The reaction rate is high. The yield is massive.
It’s the best we’ve got for fusion energy.
But you can’t fuel a future power grid with trace amounts. Scientists need to ‘breed’ more of it.
Enter the quantum computers. Not just any supercomputers—specifically, a quantum-centric approach borrowed from a totally different field.
“Quantum computers are key tools that accelerate the discovery… needed to produce sufficient trit.”
Tom Beck, Oak Ridge National Laboratory
The Cleveland Clinic has been using these techniques to simulate proteins. Researchers at Oak Ridge, IBM, and Michigan State figured they could apply the same muscle to fusion. They turned to a substance called FLiBe.
What is FLiBe? A molten salt. Lithium fluoride and beryllium fluoride mixed together.
In a reactor, FLiBe sits in the ‘breeding blanket’. It absorbs the stray neutrons from the fusion reaction and converts lithium into fresh tritium. Simple in theory. Complicated in practice because the atomic interactions inside that molten salt are a nightmare to calculate using classical physics.
So the team let the quantum simulators loose.
They found nine distinct molecular configurations. Just nine.
Why does this matter?
Before this, you had to guess. You mixed salts in a lab, heated them to infernal temperatures, waited to see if anything exploded or produced fuel, and prayed for data. It was slow. It was expensive. It often led nowhere.
Now? They can see the electronic structure. The atomic behavior. The bond strength. All in silico.
They know which configurations are worth the trouble before they even heat up the salt.
Is it done? No. This is simulation work. A preprint on arXiv, nothing more yet. They still have to go back to the physical world to see if the atoms actually behave the way the quantum model predicts.
Fusion has been stuck in the lab for decades. Sure, Lawrence Livermore finally hit breakeven at the end of 2032—wait, 2022. More energy out than in. And we’re holding plasma hotter for longer; a recent record was 1,337 seconds, which sounds like an eternity for plasma physics.
But none of that means much if you run out of fuel.
Jerry Chow from IBM calls this “practical.” Maybe. The quantum computers aren’t perfect yet, scaling is still a headache, and simulating chemistry isn’t the same as building a power plant.
It’s a promising step though. One where the computer stops guessing and starts pointing the way.
What happens when the lab tests fail? We don’t know yet. But the tools are sharper than they used to be.
