Compact Fusion Power

Warning: This section of my site describes my current understanding of topics I'm curious about. Parts of it will be wrong. I write authoritatively without hedging because it's more fun, but please don't mistake that for actual authority. Please do your research before basing non-trivial decisions on this, and submit any corrections here. Also, feel free to tweet at me about it

Teams around the world are trying to build compact nuclear fusion reactors. I'm in love with this mission for largely aesthetic reasons. I don't think we need fusion power to mitigate climate change, nor do I think we should bet much on it in that arena. However, if we want to escape "fusion is always 20 years away" small reactors that you can iterate fast on feel right to me. This article says Lockheed Martin's Skunkworks has already built five iterations of their reactor. We formally agreed to build ITER - the crown jewel of the "make them bigger" approach - in 2006, and it won't be ready to test until 2025.

For me, fusion power is romantic-verging-on-quixotic because it's not the solution 1 to our current problems: It's the key to new kinds of magic.

US Supercarrier Gerald R. Ford. Imagine ships like this, but not just for killing. Source: Wikimedia
US Supercarrier Gerald R. Ford. Imagine ships like this, but not just for killing. Source: Wikimedia

Merchant marine nuclear power has been a non-starter since we dropped the first bomb. Fission meltdowns, nuclear proliferation, and decommissioning are scary problems and we don't need giant ships badly enough to solve them. But, if Lockheed Martin actually delivers a fusion power plant in a shipping container, that completely flips the economics of shipping. I can't tell you what this looks like but I can tell you it's awesome. Maybe we'll see mind-bogglingly massive ships traveling at about the same speed. Maybe we'll sacrifice energy efficiency for higher speed hydrofoils or wingships to deliver faster and amortize the capital cost of the ship over more trips.

Maybe we'll ditch ships entirely and fusion powered aircraft that dwarf the AN-225 will take over all cargo service. Atomic Accidents by James Mahaffey has a great section on a 1950s project which tried to pull this off with early fission.

General Fusion

I don't know who's going to pull this off, or when anyone will, but for aesthetic reasons General Fusion is my favorite contender. Their approach is satisfyingly physical: pistons, pressure, and a whirlpool of molten metal. I can viscerally imagine groaning about some tedious problem with my fusion reactor like a bad carburetor on an old car. As I understand it, the three big issues for fusion reactors are

  1. Squeezing hydrogen plasma hard and hot enough that the atoms fuse together,

  2. Safely capturing the heat of that fusion to do work (thermonukes are not power plants)

  3. Keeping the flood of neutrons released from destroying every material in the reactor.

General fusion reactor diagram. Source: Wikimedia
General fusion reactor diagram. Source: Wikimedia

General Fusion is building a sphere of molten lead and lithium a few meters in diameter and using pumps to spin the metal so a siphon forms along its vertical axis. Hydrogen plasma is shot into the siphon, and pistons arrayed across the outer surface of the spherical chamber rapidly push in more molten lead. The additional metal crushes the central siphon to 1/1000th its original volume and adiabatic compression of the hydrogen plasma takes it to other-worldly temperatures 2. The plasma fuses during the crush and blasts the inner wall of molten metal with intense heat and neutron flux. The siphon grows again and the whole process repeats.

Here's the beautiful bit: Extracting heat and getting useful reactor lifetimes in spite of neutron damage are fairly open problems for most approaches. General Fusion has a thick layer of molten lead between the fusion and the nearest solid surface, so the neutrons just turn into extra heat3. They're also already pumping that lead through a loop to establish their siphon, so adding a heat exchanger to make steam and run boring conventional turbines is dead simple.

There are asterisks on this: Making plasma stable enough to endure the time between injection and squeezing is hard but they say they've done that. The magnetohydrodynamics of the spinning, shrinking internal surface of the molten metal are difficult to predict. But, the whole system makes sense. You can visualize it, and you can imagine a future where it's banal.

Footnotes

Footnotes

  1. If we actually got compact fusion working tomorrow it'd solve a lot of problems, when I say solution here I mean "safest satisficing solution." Fusion's pay-off curve is a huge swath of nothing, a narrow band of "oh yeah that fixed a few things but has limitations" and then utopia. I'm more comfortable betting most of the house on stuff like solar, where we know every marginal unit of effort has positive and increasing impact.

  2. It surprises me that fusion can occur at pressures compatible with physical pistons, and a physical pressure vessel for that matter. The lead is incompressible so the entire surface of the vessel and the head of each piston sees the same pressure as the fusion reaction. Fusion in stars like our sun happens at an absurd 250 billion times atmospheric pressure. It turns out we totally can't achieve that and all our fusion plans revolve around temperatures far hotter than the core of the sun and paltry pressures. The record plasma pressure was set by Alcator C-Mod and it's just over 2 atmospheres. I'm not sure plasma pressure is the same as "how hard it pushes against a tangible surface" but it seems to be.

  3. This is a bit of a hand-wave. The lead absorbs the neutrons, turning into other isotopes of lead and releasing EM radiation in the process. You can go through the list of isotopes yourself to follow that process - each time a lead atom absorbs a neutron its atomic weight increases by 1. The nice thing about lead is the overwhelming majority of lead atoms are 206, 207, and 208, which are all stable. 209 is terrifyingly violently unstable with a half-life of 3 hours, but it decays into Bismuth-209 which is effectively stable with a halflife of 10^19 years. 210 is actually bad with a 22 year halflife and worrying decay products, but you only get it in "double captures" where a 208 atom first captures a neutron to become 209, and then captures a second neutron in the handful of hours before it decays to Bismuth. Even lead-210 has a halflife of 22 years and all its decay products are less than half a year, so within a single lifetime almost all your lead-210 will turn back into stable lead-206. This is why General Fusion says they'll produce "no long lived waste." The other interesting bit related to neutron capture is tritrium breeding which General Fusion plans to achieve by mixing molten lithium into their molten lead. This will also improve the capture cross section of the mixture, further protecting the solid components from neutron damage.