Hydrogen, the universe’s most abundant element, is the fuel for any potential fusion reactor. The endgame is an energy source so cheap and clean and plentiful that it would create an inflection point in human history, an energy singularity that would leave no industry untouched. Fusion would mean the end of fossil fuels. It would be the greatest antidote to climate change that the human race could reasonably ask for. Saving the world: that is the endgame.
People have been talking about it way too much for way too long. The theoretical underpinnings go back to the 1920s and serious attempts to produce it date to the 1940s. Fusion was already supposed to save the world 50 years ago. The running joke about fusion energy is that it’s 30 years away and always will be. It’s not a very funny joke, but historically it’s always been true.
There are many prototype fusion reactors being built around the world. If one of them ever works, it will transform the world as completely as any technology in the past century. Most were built by universities, large corporations and national governments – with all the speed, parsimony and nimble risk taking that that implies. Fusion research has a reputation for consuming time, money and careers in huge quantities while producing a lot of hype and not much in the way of actual fusion.
But over the last 10 years, a new front has opened up. The same engine of raging innovation that’s been powering the high-tech economy, the startup, has taken on the problem of fusion. There is now a stealth scene of virtually unknown companies working on it, doing the kind of highly practical rapid-iteration development you can do only in the private sector. They’re not funded by cumbersome grants – the money comes from heavy-hitting investors with an appetite for risk – most companies most people have never heard of. But you’ve heard of the people who invest in them: Bezos Expeditions, Mithril Capital Management (a.k.a PayPal co-founder Peter Thiel), Vulcan (a.k.a. Microsoft co-founder Paul Allen), Goldman Sachs.
Fusion also gets mixed up with nuclear fission, the kind or nuclear power we now have, but they are very different animals. Fission involves splitting atoms, big ones like uranium 235 into smaller atoms. This releases a lot of energy, but it has a lot of drawbacks too. Uranium is a scarce and finite resource, and nuclear plants are expensive and hazardous – Three Mile Island, Chernobyl, Fukushima – and produce huge quantities of toxic waste that stays hazardously radioactive for centuries.
Nuclear fusion is the reverse of nuclear fission: instead of splitting atoms, you’re squashing small ones together to form bigger ones. This releases a huge burst of power too, as a fraction of the mass of the particles involved gets converted into energy (in obedience to Einstein’s famous E=mcsquared). It’s what makes the sun shine. The sun is a titanic fusion reactor, constantly smooshing hydrogen nuclei together into heavier elements and sending us the by-product in the form of sunlight.
As an energy source, fusion is so perfect, it produces 3-4 times as much power as nuclear fission. Its fuel isn’t toxic, or fossil, or even particularly rare: fusion runs on common elements like hydrogen, which is in fact the most plentiful element in the universe. If something goes wrong, fusion reactors don’t melt down, they just stop. They produce little to no radioactive waste. They also produce no pollution: the byproduct of fusion is helium, which we can use to inflate the balloons for the massive party we’re going to have if it ever works.
A 1-GW coal-fired power station requires 10,000 tones of coal – 100 rail-wagon loads – every day. By contrast … the lithium from a single laptop battery and the deuterium from 45 liters of water could generate enough electricity using fusion to supply an average Western consumer’s energy needs for 30 years.
What makes fusion hard is that atomic nuclei don’t particularly want to fuse. Atomic nuclei are composed of protons (and neutrons), so they’re positively charged, and as we know from magnets, things with the same charge repel each other. You have to force the atoms together, and to do that you have to heat them up to the point where they’re moving so fast that they shake off their electrons and become a weird cloud of free-range electrons and naked nuclei called a plasma. If you get the plasma really hot, and/or smoosh it hard enough, some of the nuclei bang into each other hard enough to fuse.
The heat and pressure necessary are extreme, essentially you’re trying to replicate conditions in the heart of the sun, where the colossal mass – 330,000 times the mass of the Earth – creates crushing pressure, and where the temperature is 17 million degrees Celsius. In fact, because the amounts of fuel are so much smaller, the temperature at which fusion is feasible on Earth starts at around 100 million degrees Celsius.
That’s the first problem. The second problem is that your fuel is in the form of a plasma, and plasma is weird. It’s a fourth state of matter, neither liquid nor solid nor gas. When you torture plasma with temperatures and pressures like these, it becomes wildly unstable and writhes like a cat in a sack. So not only do you have to confine and control it, and heat it and squeeze it; you have to do all that without touching it, because at 100 million degrees, this cat will instantly vaporize solid matter.
You see the difficulty. Essentially you’re trying to birth a tiny star on Earth. It comes down to two challenges. Long enough and hot enough. Can you keep your plasma stable while you’re getting it up to these crazy temperatures? The severity of the challenge has given rise to some of the most complex, most extreme technology humans have ever created.
A common method for creating fusion is by controlling the plasma magnetically. One of the few breaks physicist found in the quest for fusion is the plasmas are extremely sensitive to electromagnetism, to the point where electromagnetic fields can actually be used to contain and compress them without physically touching them. It’s a feat most often performed using a device called a tokamak, a big hollow metal doughnut wrapped in massively powerful electromagnetic coils. The coils create a magnetic field that contains and compresses the plasma inside the doughnut.
Since they were developed in the Soviet Union in the 1950s, tokamaks have come to dominate fusion research: in the1980s enormous tokamaks were built at Princeton and in Japan and England, at a cost of hundreds of millions of dollars.
In a tokamak the particles in the plasma move in tight spiral orbits around lines of electric current. But it’s hard to keep those particles from being bumped out of their little orbits by electromagnetic turbulence, and when that happens the plasma becomes unstable and loses precious heat. One way scientists fight this instability is by building bigger and bigger tokamaks, but bigger means more complex, and more power-hungry, and more expensive.
When you are doing fusion, what atomic nuclei do you fuse? By far the most popular answer is deuterium and tritium, two isotopes of hydrogen. This is fusion’s low hanging fruit, because deuterium and tritium fuse at a lower temperature than any other option, a comparatively mild 100 million degrees Celsius. ITER uses D-T fusion, as do the NIF, the National Spherical Torus Experiment a Princeton, Lockheed Martin, General Fusion and almost everybody else.
But there are catches. One is that tritium is rare, so you have to make it. The other is that the reaction emits, along with an isotope of helium, a neutron, which is a problem because when you throw a lot of free neutrons at something it eventually becomes radioactive. That means you’re stuck regularly replacing parts of your reactor as they become too hot to handle. If it works, you’ve still got many decades of materials research to try to make something that lasts more than six to nine months, in the hellish bombardment of neutrons it is going to have to live in.
But there are engineering solutions to the problem.
The goal for all these machines is to pass the break-even point, where the reactor puts out more energy than it takes to run it. The big tokamaks came close in the 1990s, but nobody has quite done it yet, and some scientists find the pace frustrating.
Academics aren’t necessarily good at adhering to a schedule, promising something and delivering it on budget and on time. In a university lab the name of the game, the end product, is a paper. You want to get to making energy, but it’s not the primary goal. The primary goal is to publish a lot of papers, to go to conferences and understand very thoroughly all the little details of what is going on. Understanding is all well and good in an ideal world, but the real world is getting less ideal all the time. The real world needs clean power and lots of it.
Fusion research is too slow, too cautious and too focused on lavishing too much money on too few solutions and too many tokamaks. Because of their extreme size and complexity, and the political vagaries associate with their funding, fusion projects are bedeviled by cost overruns and missed deadlines. The federal process also doesn’t condition you to live in that mind per se. This is one of the failures of the governmental way of running it. It doesn’t create enough diversity of ides, and let those freely be pursued to failure.
A private sector is a better place to get things done than a university lab. The problem with fusion typically is that it’s driven by science – the most predictable next step, the one you’re comfortable with. So it doesn’t necessarily connect with what you want. You’ve got to look at the end in mind. You’ve got to unravel it, reverse-engineer it. What would a utility want? What would make sense? And design something from there, and be agnostic as to how hard the physics might be.
Raising money is a challenge: tokamaks were eating up all the grant money, and energy startups are expensive, risky long-term bets, especially to Silicon Valley investors spoiled from flipping web startups for quick paydays.
Recruiting was tough too. Building a fusion device requires a blended culture of physicists and engineers, two groups who don’t historically mix well. You have money for a year or two to develop something, deliver, and to the next chunk is not the academic risk profile.
To keep the pace up they freed themselves from the baggage of theory: as long as something worked, they didn’t analyze to death why. The idea was to stay pragmatic and literate rapidly, spend as little as possible and not fear failure. Say this is where we ultimately want to go, what are the critical steps to get there, what are the risk elements of the path to get there, and can I test for some of these risks without spending a hundred million bucks?
Private industry can be quite nimble, relatively speaking, in exploring ideas and testing them for the first time. It’s good to see private investment in fusion – there’s an impatience that drives the startups. If the science breaks go their way, they will be able to accelerate the pace of getting fusion energy on the grid.
International Thermonuclear Experimental Reactor (ITER)
The biggest prototype, the colossus of all tokamaks is under construction by a massive international consortium in a small town in the south of France outside Marseilles, with a price tag of $20 billion and projected due date of 2027.
The reactor will be 30 meters tall and weigh 23,000 tons. Its staff numbers in the thousands. It will hold 840 cubic meters of plasma, its magnets alone will require 100,000 kilometers of niobium-tin wire. Its stupendous cost is being paid for by a global consortium that includes the US, Russia, the European Union, China, Japan, South Korea and India.
ITER’s estimated date for full power operation has slipped from 2016 to 2027, and even that date is under re-evaluation. Its price tag has gone from $5 billion to $20 billion; for purposes of comparison, the Large Hadron Collider cost $4.75 billion.
And even when it does get up and running, ITER will never supply a watt of power
to the grid. It’s a science experiment, not a power plant. Proof of concept only.
Tri Alpha Energy
Housed in a building the size of a small house, it draws so much power that when turned on, they have to disconnect from the pubic grid and run off its own power to keep from shorting out the whole county. All the iron rebar in the building’s foundations has been pulled out and replaced with stainless steel rebar, because iron is too magnetic. For the first few years, the company ran on the brink of insolvency. Tri Alpha is so low profile, it didn’t have a website until a few months ago.
The driving force behind the founding of Tri Alpha was a physicist at UC Irvine named Norman Rostoker, who died in 2014. In the early 1990s, he was skeptical of the tokamak hegemony. Michl Binderbauer is also one of the founders of Tri Alpha, and now is the brains and drive behind Tri Alpha. It is probably the best funded of the private fusion companies – to date it has raised hundreds of millions, but still a tiny fraction of what’s being spent on the big government-funded projects.
Rostoker thought a better way might be particle accelerators, those colossal rings, like the Large Hadron Collider, that crash subatomic particles into each other. In accelerators, particles travel on wide, conspicuously stable orbits. By bringing accelerator physics into the realm of fusion, you can actually make a better-behaved plasma, one that can give you long timescales. Then you can invest energy and heat it. Tri Alpha has managed to build a prototype fusion reactor quickly on a tiny budget.
Its reactor is very different from the towering tokamaks that dominate the fusion skyline, or the supervillain lasers of the NIF. You could think of it as a massive cannon for firing smoke rings, except that the smoke rings are actually hot plasma rings, and the gunpowder is a sequence of 400 electric circuits, timed down to 10 billionths of a second, that accelerate that plasma ring to just under a million kilometers an hour.
And there are actually two cannons, arranged nose-to-nose, firing two plasmas straight at each other. The plasmas smash into each other and merge in a central chamber, and the violence of the collision further heats the combined plasma up to 10 million° Celsius and combines them into a single plasma 70-80 centimeters across, shaped more or less like a football with a hole through it the long way, quietly spinning in place.
But a fusion reactor’s work is never done. Positioned around that central chamber are six massive neutral bean injectors firing hydrogen atoms into the edges of the spinning cloud to stabilize it and keep it hot. Two more things about this cloud: one, the particles in it are moving in a much wider orbit than is typical in, say, a tokamak, and hence are much more stable in the face of turbulence. Two, the cloud is generating a magnetic field, instead of applying a field from outside, Tri Alpha uses a phenomenon called a field-reversed configuration, or FRC, whereby the plasma itself generates the magnetic field that confines it. It’s an elegant piece of plasma-physics bootstrappery. What you get within forty millionths of a second from the time you unleash your first little bit of gas is the FRC sitting fully stagnant, not moving axially, and rotating.
The machine that orchestrates this plasma-on plasma violence is something of a monster, 23 meters long and 11 meters wide, studded with dials and gauges and overgrown with steel piping and thick loose hanks of black spaghetti cable. Officially known a C-2U, it’s almost farcically complicated – it looks less like a fusion reactor than it does like a Hollywood fantasy of a fusion reactor. It sits inside a gigantic warehouse section of Tri Alpha’s Orange County office building surrounded by racks of computers that control it and more racks of computers that process the vast amounts of information that pour out of it – it has 10,000 engineer control points that monitor the health of the machine, plus over 1,000 diagnostic channels pumping out experiential data. For every five millionths of a second it operates it generates about a gigabyte of date.
So far the company’s primary focus has been on the long-enough problem, rather than the hot-enough part; stabilizing the plasma is generally considered the tougher piece in this two-piece puzzle. But they have done it and were able to hold its plasma stable for 5 milliseconds. That’s not a very long time, but it’s an eternity in fusion time, long enough that if things were going to go pear-shaped, they would have. The reactor shut down only because it ran out of power – at lower power, and hence with slightly less stability, they’ve gone as long as 12 milliseconds. They have totally mastered this topology and can hold this at will. It does not veer at all. Tri Alpha has tamed the plasma.
To solve the neutron issue with D-T fusion, Tri Alpha plans to fuse protons (hydrogen nuclei) with boron-11. This reaction produces no neutrons at all, and both elements are plentiful and naturally occurring. If you buy their plant, they will give you a lifetime supply of fuel for free. But nobody else is pursuing that route as proton-boron-11 fusion requires much higher temperatures, insanely higher: 3 billion ° C. No one knows how plasma will behave at that temperature and everyone is skeptical about Tri Alpha making it work. Fusion is hard already, even when it’s D-T, and this is much harder than D-T, so big a leap that you worry about its viability – it is outrageously, outrageously ambitious. But Binderbauer is not intimidated by this.
The next move is to tear down Tri Alpha’s current reactor and build a new one that will scale up to the necessary temperatures. Particle reactors can create temperatures in the trillions. Going to higher temperatures is not that hard – you use techniques much like in a microwave.
General Fusion
Another approach is called magnetized target fusion: a spinning vortex of liquid metal, inject some plasma into its empty center, then squeeze the vortex, thereby squeezing the plasma inside it and causing it to heat up and fuse.
They took the idea to investors and founded the company. Now with 65 employees, they have raised $94 million and built prototypes of the reactor’s major subsystems, including a spherical chamber for the liquid metal vortex with 14 large spikes projecting out at all angles – the spikes are massive hammers that do the squeezing. It looks even more like a fusion reactor than Tri Alpha’s. They have sped up the timeline, but still think they have a decade to producing energy.
To solve the neutron problem with D-T fuel, that vortex of liquid metal in General Fusion’s reactor is a mixture of lead and lithium, which catches the neutrons. As a bonus, when you hit lithium with neutrons, you get tritium – so two birds with one stone.
Helion Energy
Another venture in Redmond Washington, it is already on its fourth-generation prototype. Its approach has two plasmas colliding in a central chamber, but it will work in rapid pulses rather than sustaining a single static plasma. It is focused on developing a smaller-scale, truck-sized reactor, and doing it as fast as possible with a commercial reactor operational within six years.
For the neutron issue with D-T fusion, Helions’s reactor will fuse deuterium and helium-3, which produces fewer neutrons, though it requires more heat and raises the problem of finding enough helium-3, which is also rare.
The National Ignition Facility (NIF)
At Lawrence Livermore National Laboratory outside San Francisco, it is housed in a 10-story building with a footprint the size of three football fields. The NIF houses one of the most powerful laser systems in the world: 192 beams of ultraviolet light capable of delivering 500 trillion watts, which is about 1,000 times the power as the entire US is using at any given moment. All that energy is delivered in a single shot lasting 20 billionths of a second focused on a tiny gold cylinder full of hydrogen. The cylinder simultaneously explodes, and the hydrogen inside it fuses. This technique is called inertial confinement fusion.
The NIF was finished seven years late for $5 billion, almost the original budge.
Industrial Heat in Raleigh, NC, Lawrenceville Plasma Physics in New Jersey, Tokamak Energy outside Oxford, England and Lockheed Martin’s Skunk Works are developing fusion systems also.
Everybody in the fusion industry shares a worldview in which the transformation of the globe by fusion power is imminent. It will happen as the machines claw their way up to 3 billion ° C, and the theories suggest this is possible. The physics says it is, but to test it, nature is the ultimate arbiter. That is the risk.
A reactor in five years is impossible and how long is impossible to predict. Most believe it is 3-4 years from the point where the risk changes from a science risk to an engineering risk. Maybe in a decade, things can mature to the point to the first commercial steps.
There may be a lot of those steps. The utilities will be the ones making the actual transition, and for fusion to be of any earthly use, it will have to make business and engineering sense, as fusion plants will be expensive. Unlike solar and wind, fusion would provide energy constantly, not intermittently, but there would have to be enough of it. The gain (the ratio of energy-out to energy-in) of a commercial fusion plant would have to be in the 15-20 range, but no fusion reactor has reached a ratio of 1, the break-even point.
Then there’s the question of how exactly to extract the energy from the reactor in the form of heat, so that it can plug into the existing infrastructure. But those steps would be giant leaps for mankind.
Bill Gates is currently on a global campaign trying to raise awareness about how badly our addiction to energy is destroying the environment. He’s putting $2 billion of his foundation’s money into it. Innovation is needed that gives energy that’s cheaper than today’s hydrocarbon energy, that has zero CO2 emissions, and that’s as reliable as today’s overall energy system. An energy miracle is needed. Gates has personally invested in Terra Power, a maker of next-generation fission plants.
To assess the precise probability that fusion will or will not be that miracle needs a PhD in plasma physics, but it is looking a lot more plausible than most miracles. Many say the claims of the private sector to be overconfident but it is a question of not if, but when – it’s inevitable – not within 10 years, but commercial fusion on the grid by the 2040s. That’s a long way away but in terms of mitigating climate change, fusion will play a very critical role.
Fusion may turn out to belong to the category of human achievement – like powered flight and moon landings – that appeared categorically impossible right up until the moment somebody did it. At the very least, a lot of very smart people are betting their money and their careers on it. As for the rest of us, we may already have bet the planet.
Fusion power
The race to build a commercial fusion reactor hots up
A Canadian firm plans a demonstration machine in Britain
But that record does not dismay a growing group of “alternative fusion” enthusiasts. Through a combination of new technology and entrepreneurial derring-do they hope to beat ITER to the punch. On June 17th one of their number, a Canadian firm called General Fusion, put its investors’ money where its mouth is. It said it would build a demonstration reactor, 70% the size of a full-blown commercial one, at Culham, the site of the Culham Centre for Fusion Energy, Britain’s national fusion-research laboratory. Like ITER, it hopes its reactor will be up and running by 2025.
Power play
On paper at least, fusion is attractive. Existing nuclear plants rely on fission—the splitting of heavy atoms, usually of uranium, into lighter ones. The energy thus liberated is used to boil water into steam, which then turns turbines that make electricity. Fusion plants attempt to do the opposite, generating heat by combining light atoms to make heavier ones.
Unlike coal or natural gas, fusion would produce no planet-heating carbon dioxide. Unlike solar panels and wind turbines, fusion plants could operate in any weather. Unlike fission plants, they pose no risk of spreading nuclear-weapons technology, and should generate much less radioactive waste. They offer safety, too. “I like to say that fission is easy to start and hard to stop,” says Christofer Mowry, General Fusion’s boss. “Fusion is the opposite.”
Fusion is hard to start because it requires extreme conditions. Most Earthly fusion reactors aim to combine deuterium with tritium. (Both are isotopes of hydrogen, in which the single proton in that element’s nucleus is joined by either one or two neutrons.) Protons have a positive electrical charge, and like charges repel. Persuading two atoms to join forces therefore means overcoming this repulsion. And that requires a great deal of energy.
General Fusion’s idea is to forge a middle path between two existing approaches, magnetic-confinement fusion (MCF) and inertial-confinement fusion (ICF), with less need for heroic engineering than either. ITER is a doughnut-shaped type of MCF reactor called a tokamak. It is intended to use carefully controlled, high-intensity magnetic fields to heat a hydrogen plasma to hundreds of millions of degrees Celsius, and then hold that plasma stable while its atoms combine. The trick is to control the fields precisely enough to keep the super-hot plasma together for long enough to allow a significant amount of fusion to happen. The present record, held by an experimental reactor in France, is six-and-a-half minutes. ITER’s goal is a reaction that lasts up to ten minutes.
General Fusion calls its own approach “magnetised target fusion”. The basic concept dates back to the 1960s. The firm’s reactor, says Mr Mowry, does away with magnetic confinement by using powerful electric pulses to create self-stabilising blobs of plasma that are injected into the reactor’s core. He compares this to blowing a smoke ring, in which the air currents within the ring allow it to maintain its shape for a few seconds before it dissipates.
The puffs of plasma actually last around 20 milliseconds. That would not be long enough to extract much energy were they to be injected into an MCF reactor. But it is long enough for them to be compressed, as in an ICF machine—and by something far less exotic than banks of advanced lasers. The core of General Fusion’s British reactor will be lined with molten lithium and lead. Once a puff of plasma has been injected, ranks of gas-driven pistons will compress the core, changing it from a cylinder to a sphere and drastically boosting the fusion rate (see diagram).
But while laser compression happens in mere billionths of a second, General Fusion’s takes thousandths—comparable with the timescales on which internal-combustion engines operate, and well within the capabilities of digital electronics to fine-tune. The upshot, the firm hopes, is a reactor which should be cheaper and simpler to build and operate than either an MCF or ICF machine.
Critical mass
Besides compressing the plasma, the liquid-metal jacket serves to capture the energy from the reaction. Heated metal will be piped to a heat exchanger and used to raise steam. Neutrons from the fusion reaction, meanwhile, will transform some of the lithium into more tritium fuel, which would otherwise be rare and expensive. Or at least, it will one day. General Fusion’s demonstration reactor will only fuse deuterium with deuterium, to keep things simple.
Still, the firm hopes that a full-fledged commercial reactor—which might be built in the early 2030s—could compete with other forms of electricity. It is aiming at a cost of $50 per megawatt-hour, which Mr Mowry says ought to make it competitive with coal. Renewable energy may prove cheaper, he concedes, but it is hampered by intermittency. And the plant will have one more advantage over existing, fission plants, the electricity production of which cannot quickly be raised or lowered. General Fusion’s reactor can increase or decrease power output ten-fold by changing the speed at which the core cycles. That should allow it to “load-follow”, ramping up production when electricity prices are high and cutting back when they are low.
And General Fusion is not the only firm pursuing commercial fusion. On April 8th TAE Technologies, a rival based in California, which was founded in 1998, said it had raised $280m for a demonstration reactor of its own, bringing the total invested in the firm to $1.1bn (see chart). Like General Fusion, TAE relies on puffs of self-stabilising plasma. Unlike General Fusion, it aims to produce electricity by combining hydrogen with boron, a process that needs temperatures of billions of degrees, but which should require less radiation shielding.
General Fusion’s other rivals include two British firms, First Light Fusion and Tokamak Energy, both based near Culham, and a pair of American ones, Commonwealth Fusion Systems and Zap Energy. Nor are governments putting all their eggs in the ITER basket. The Max Planck Institute for Plasma Physics, a German government agency, is trying to build a power station based on a device called a “stellarator”—a twist on the MCF approach. Its Wendelstein-7X device began working in 2015. And the Culham Centre for Fusion Energy’s STEP reactor, scheduled to open in 2040, is intended to demonstrate the commercial practicality of fusion. One reason General Fusion chose Culham, says Stephen Dean, who runs Fusion Power Associates, a research and education foundation which covers the field, is that this lab has a focus on getting to market quickly.
For his part, Dr Dean sees no fundamental reason why one or more of the current crop of contenders should not succeed in building a reactor that generates useful quantities of energy. But it is economics, not physics, that will have the final word. High-tech fusion reactors, assuming they are ever built, will have to compete in a world in which the price of solar and wind power is falling steadily. Fossil-fuel companies, meanwhile, are trying to work out how to capture and bury the carbon dioxide emitted from their power stations. Advanced fission reactors are attracting private interest of their own—on June 2nd TerraPower, a firm backed by Bill Gates, a founder of Microsoft, announced plans for a high-tech nuclear plant in Wyoming. All that competition is good news in a warming world. But it promises a white-knuckle ride for investors.
