Fusion’s Time is (Finally) Coming
Every aspect of nuclear fusion is like something out of science fiction: particles flying at hypersonic speeds, synchronised laser arrays, liquid metal and super-strong magnets, all in the service of abundant clean energy.
It sounds too good to be true – and for a long time, it has been. Fusion has always been beyond reach, always twenty or thirty years away. But that might finally be changing. On this week’s podcast, I spoke to one of the people working to make it a reality.
Nick Hawker is the co-founder and CEO of First Light Fusion, an Oxford-based company using a novel method to tackle one of science’s hardest problems. He and I discussed how and when nuclear fusion might become part of our energy mix, and what hurdles we’ll have to jump to get there.
🎧 You can listen to our conversation here, or 📝 read a full transcript.
The Big Idea
The general principle of nuclear fusion is relatively simple: two light atoms are bonded together to form one, heavier atom. That process releases energy. One example: an atom of deuterium and an atom of tritium – both isotopes of hydrogen – can be fused to make helium (and release energy):
That process can be carried out in two main ways. Magnetic confinement fusion squeezes the particles together using magnetic fields, inducing fusion. The most common type of magnetic confinement system is the tokamak, but there are other approaches including stellarators and field-reversed configuration systems. In inertial confinement fusion, the most common approach is to use lasers to heat and compress the deuterium and tritium fuel.
While fusion has largely been funded by public sector investments, small amounts of venture capital have been flowing into teams for about a decade. Things had started to get more serious by 2017... but 2021 has proven to be a stonking year with more than $2.5bn of capital flowing into fusion startups.
Venture capital investment into nuclear fusion startups
A New Approach
First Light’s method is a variation on inertial confinement fusion. The firm doesn’t use lasers to implode hydrogen isotopes:
[W]hat we’re doing is a new method for inertial fusion, which we call projectile fusion. What we do is we fire a high-velocity projectile that flies for a short distance and it hits the target... we launch our projectile with electromagnetic forces, so it’s like a railgun.
Nick’s approach has a number of advantages (over and above the fact the company gets to put out press releases with titles like ‘First Light Fusion Installs UK’s Biggest Two-Stage Hyper-Velocity Gas Gun’). Not only is the projectile fusion approach more energy-efficient than using lasers, but it does away with the need to build geometrically complicated laser arrays in an extremely hostile environment.
First Light Fusion’s approach has another, fascinating advantage: its experiments lend themselves to simulation. Instead of conducting onerous, energy-intensive & time-consuming experiments, First Light can mimic them on a computer:
[T]he advent of the computing power that we now have, and the simulation tools and methods that we now have, make a huge, huge difference to the rapidity with which we can explore the parameter space. We’re doing a campaign right now, an experimental campaign. It’s a 20-shot campaign and we’ve done over 100,000 simulations going into the design of that campaign. So all of that learning that we’re able to do in silico allows us to make much more rapid progress than we’d be able to make otherwise.
Comparing the results of physical experiments with simulations allows Nick and his team to refine their method:
[W]e’re constantly going back and forwards between the simulations and the experiment, and whenever we’re looking at the experimental results, we take the simulation results as gospel... When we're looking at the simulations, we’re assuming that the experiment has been executed perfectly; where is our physics modelling incorrect, where could there be some additional process or some different thing happening?
The real-world results and the simulations hone each other: iron sharpens iron.
What Progress Looks Like
Getting a reactor to produce more energy than it consumes has long been fusion’s holy grail. That phenomenon, known as net gain, will mark a major milestone when scientists achieve it – and Nick is confident that one of the many approaches currently in train could get there this decade. Discussing the National Ignition Facility – a leading, California-based inertial fusion lab – Nick suggested major progress could be on the horizon:
They’ve been operating and doing these experiments for 10 years [and] there’s been a succession of improvements coming through... I think of it like overlapping S-curves: [you] get on a new idea, progress is slow to start with because you don’t know what it is, and you don’t understand, and [then] you really get the hang of it... they’re right in that bottom of the curve where they’re still exploring and understanding. I’m very, very confident they’re going to make that progress.
Achieving net gain is a big step, but it isn’t the same as building a commercially viable fusion reactor. So when will we have workable fusion power? Nick suggests there’s a “very high” probability fusion will be contributing to our energy mix in 2070, or even 2050, as a range of startups work to tackle the problem. But his sights are set even higher:
I think there’s a very, very high chance that there will be a fusion reactor connected to the grid somewhere [by 2030]. If it’s one of ours, I think it will be economically viable…
Watch this space.
Nick and I also discuss:
☀️ The sun’s inefficiency, and how science can do better [4.53]
⛳ How crazy golf can explain nuclear fusion [9.39]
🔫 How many lasers it takes to turn on a light bulb [16.26]
Listen to this, too 🎧
Earlier this series, I spoke to Michele della Vigna, who runs the Carbonomics research programme at Goldman Sachs. His work is all about putting a price on the green transition, and costing the various technologies that will wean us off fossil fuels. We talked about all manner of methods for doing that, both established and emergent. It was a fascinating conversation, and a really clear-eyed run-through of the problem. You can listen to it here.