Great ideas need to be a little crazy, but crazy in the right way. I have been down this rabbit hole of asking myself “could we economically fuse hydrogen, at Type I scale of civilization?”
So I have a particular design to present. I’m not at all saying that this is the best design. It’s closer to something like a “proof” that pure-proton fusion can be used as a power source of an advanced civilization, outside of stars. As I have discovered, I have no idea if any other humans on this Earth agree or disagree with the claim. If you have ever had any opinion on this matter, you need to leave a comment because you would be the first documented case aside from myself, as far as I can tell.
I have an answer, and my answer is “yes”. But I literally wasn’t sure until I ran deeper calculations and twisted myself into physics knots. This is almost entirely with the help with GPT-5.6-Sol. Believe me, the lower model doesn’t work. The detailed work is presented mostly in jypter notebooks here. So far, in 0-dimensional qualitative work, but I’m looking into spatial simulations next.
Motivation
My reason for looking at a Type I scale is the “speedrun” hypothesis. Perhaps we will never build a Dyson Swarm because we master controlled fusion before then.
Magnetic confinement fusion does get a lot better with increased scale, but not by enough margin to be superior to solar in the inner solar system because solar power is just that good.
But perhaps there’s potentially a bigger prize possible. Inspired by the book Lockstep (Schroeder), I picture a society spanning the space between stars using fuel from the cold bodies of the Oort cloud and further. This has the benefit of no real power limit (as opposed to our sun), and somewhat changes or solves the inter-stellar travel problem.
Implosion Fusion
Implosion fusion has, by far, the best raw energetic scaling of all the fusion methods. This has always led me to think about a “Project PACER in space” idea, under the thesis that implosion fusion will win in the long-term regardless of what happens in the near-term.
So I just hijacked that idea when I started to look at the CNO cycle. If its the best, then it likely has the best chance of being able to fuse the lower cross-section reactions that the cycle requires. I have not honestly looked into an alternative with magnetic confinement, and that might actually work.
Staging
For implosion fusion to really get good, however, you need multiple stages. This is analogous to, though not quite exactly, what thermonuclear weapons do. Those can have more than two stages, which means fusion triggering additional fusion reactions.
So yes, I’m saying that once you get past the initial ignition requirement, a specially designed fusion fuel target may be able to use the first burn to compress and ignite a second burn. This is a highly effective method, because it reduces the marginal need for external energy inputs to essentially zero.
We abuse that principle. Not just once, it turns out, but in multiple ways, to develop a modified CNO cycle.
Decay Steps
One interesting quirk of the CNO cycle is that it involves some steps with nuclear decay. This limits the reaction rate, and is the reason for a separate “hot” cycle with the other one. Most literature I saw on this subject had this happen within the reactor, and it’s just a rate-limiting thing.
For the approach with implosion fusion, that obviously won’t work. The entire game is a race between burning enough fuel compared to the disassembly time of the compressed fuel ball. The solution was incredibly obvious once you get here. Run each fusion reaction in a separate implosion fusion reaction. By the time you manage to process the products, it will have finished its decay anyway.
Deuterium Production
By far and away, DT is the easiest fusion fuel to use. You can also make it from simple protons if you have a neutron source. The problem, then, is how to get a neutron source. This is a surprisingly difficult problem that needed 2 key pivots.
Neutronic CNO Modification
The ordinary CNO cycle does not produce an extra neutron. However, since we are now in the business of engineering the whole fuel cycle reaction-by-reaction, we can make it. Ultimately, what was used is this:
¹³C(alpha,n)¹⁶O
There was another, weirder, candidate that was possible, and I looked into it a good deal. But it required an endothermic reaction for the “return” path. While that’s not a deal-breaker by itself, the ratios wouldn’t have worked well even if the best-case energy exchange. So it looks mostly like a dead branch if we limit ourselves to those same multi-stage implosion reactions.
This reaction would have to happen surrounded by H2 gas or liquid, with the idea that it acts as a blanket and eventually absorbs the neutron, even though its cross-section is bad, clean material would still eventually transmute into D.
This is still an extremely challenging task if you are producing D for use in the implosions themselves. So much so that if the main implosion pusher is DT, the scheme just doesn’t work. For this problem, I’ve invented an entirely different workaround.
15N Assisted Pusher
Remember how I was talking about “multiple” stages? Well we’ve hit the point of a need to add another stage to our implosion fusion reaction. Let me lay out the full process.
- Initially, an impacter hits a DT target, which is 25 cm large, about.
- This ignites additional DT fuel or DD fuel that is arranged in channels that wrap around a larger fuel ball.
- These D-based veins ignite the slightly-harder-to-use fuel 15N which is arranged in a shell around the “real” target fusion fuel.
- This main fuel ball compresses by a 1,000 or 10,000 ratio.
- Another DT ball around the center is engineered to start the main reaction at the time of greatest compression. This part may or may not be necessary, or be heavily modified.
The reason you can’t just simply use the 15N fuel is that the burn-front doesn’t move around the rest of the fuel fast enough, before the first-burned fuel flies away. Not true for the DT propagation speed. So we use the speed of DT in veins/channels that run through 15N fuel to ignite 15N which is the real pusher.
Where do we get the 15N from? Well it’s part of the main cycle! We would have otherwise had to implode it in a special reaction. It just happens to be the most “co-operative” reaction, so it can help us implode the other fuels. Kind of like if a sheep joined the sheepdog to help move the flock. Kind of.
Scale set by Closure
What I’ve described is almost guaranteed to work by physics in a implosion reaction sense. However, the problem shifts to cycle closure. Because low cross-section means low reaction rate, and we need reactions to happen in order to make more pusher material. So low burnup would mean that we spent our pusher, but didn’t generate enough pusher to replace it.
There are really 2 things we have to make sure we get enough of, one is D, which is really the same thing as excess neutrons, and the other is 15N which is the main pusher for compressing the other fuel. We can increase burnup by making the fuel ball bigger, but this doesn’t guarantee that 15N and D usage fractions are sufficiently low, even at larger scale. But it appears that they are.
Speculative solutions for the fuel cycle that get closure have been found. I am also fairly confident to say that these require fuel balls on the size between 1 meter and 500 meters in radius.
Blast Chamber
At the extreme limit of these speculative numbers, they would require a blast chamber on the scale of the Earth. Remember, we are aiming for a Type III kind of civilization eventually, so this isn’t necessarily out of the question.
However, what I have found so-far is a new limit. The materials needed to contain a gas (either Hydrogen or Helium) in such a blast chamber would be too massive and limit us to <1% of the output of the sun. I am actually convinced there are alternatives to this problem, but those get vastly weirder even than the things I’ve talked about so far.
I’m pretty excited about these results. I hope to make a number of new posts on it with images. This text is written by myself, a human. The github data is written by the AI agent, with a lot of direction from me.