Will better superconductors transform the world?
quantamagazine.org
quantamagazine.org
Some modest improvements in high temperature superconductors from MIT is allowing Commonwealth Fusion Systems to scale down ITER from being a 65 billion dollar five story behemoth, to something slightly taller than the average human.
With a few more iterations, it becomes very feasible to imagine fusion reactors the size of a rice maker to power homes and vehicles, or even smaller to just have personal, wearable fusion reactors that can provide near infinite power to any gadgets we want to carry with us.
Also I really enjoy the idea of reducing giant MRI machines down to the form factor of a hula hoop. If anyone could get highly detailed MRI scans whenever they wanted, a huge amount of disease could be prevented.
Room temperature superconductors that wouldn't require any cooling would also make lots of use cases for strong magnets obsolete, increasing power and lowering mass for e.g. EV motors, generators, etc.
It does not mean cheaper fusion. Fusion via magnets is a lie. For 70 years the fusion scientists have been telling us the same story: we are nearly there. We did the math, and if we can just increase the power of the tokamak/stellerator/z-pinch/whatever by a factor of 10, we'll get ignition.
The problem is that plasma shows all sorts of instabilities. All these projections need to come with an asterisk: "we did the math, and assuming there's no new instability that will show up this time, then ...". But each time there is a new instability.
ITER will get built one day (maybe) and they'll figure out that there's yet another issue that requires a few more tens of billions (or maybe hundreds).
They will not get the money this time. ITER did not start as a genuine quest for getting fusion. It was a political project designed to make the US and the USSR cooperate on something that both could perceive as helping humanity in general. Once the USSR disappeared, the impetus was gone, and a third of a century later it remains something that will be finished in the distant future.
Joint European Torus (JET) built in 1983! eventually hit a plasma heating Q of 0.67. Scaling that up isn’t some big leap of faith. ITER was started as an agreement between Regan in Gorbachev in 1985, but didn’t begin construction until 2013! based on a significantly scaled down and very conservative design from 2001, with an expected completion date of 2025-2030. It’s terrible, but under funded multinational projects don’t move quickly.
That’s why nothing got done with fusion, you need experiments to make progress running computer simulator and toy machines isn’t enough. Further, you can start site prep and building the facilities well before a design is finished. So much of these delays were completely political in nature not technical, we could start site prep for DEMO today but expect site selection and prep to add a few years of completely avoidable delay to that project as well.
Of course being able to build a working device is only part of the story, it also needs to make economic sense. But that’s a different matter from why things have taken so long.
Lawrence Lidsky pointed out back in the 1980s that there are serious arguments against DT fusion, even if you assume the physics isn't a problem at all. Just assuming you have a magic black box that can make DT go and you still need to capture the neutrons, and this simple sounding engineering problem makes DT fusion unattractive compared to alternatives, with volumetric power density at least an order of magnitude worse than fission reactors.
Subsequent experience bears this out. Even supposedly compact designs like ARC will still be a factor of 40 worse than PWRs by this metric. Note that better superconductors than the HTSs in ARC would not help, since the mass of ARC is dominated by the structural material needed to keep the magnets from flying apart. Any stronger magnetic field is ruled out by practical considerations of strength of this structure.
Maybe the limits can be relaxed somewhat if the reactor has thick liquid lithium as the first wall, so power/area can be increased (the liquid would be exposed directly to vacuum; at sufficiently modest temperature the vapor pressure of liquid lithium can be very low). Zap uses that approach, but notably Zap doesn't use external magnets (superconducting or otherwise); the magnetic field is generated by current flowing in the plasma. Zap doesn't cover all 4 pi steradians around the plasma with flowing metal though, so neutron load on exposed components may still limit their power density.
What matters is the cost of the walls which mostly scales with surface area not internal volume.
I'm talking about the volume of the reactor, not the plasma, and no, it's not "basically free". Very far from it! The size and cost of the reactor is the big issue. In a DT fusion reactor, you're basically (from an economic point of view) burning the reactor, not burning the fuel. The reactor is the consumable that drives the cost of produced energy. The cost of the reactor is a function of its size and complexity. A fusion is reactor is both much more complex than, and much larger than, a fission reactor of the same thermal output. So, if you're using both as heat sources, the fusion reactor will not be able to compete with a fission reactor.
Fusion reactors also have strong diseconomies of scale, since they are limited by power through the first wall, but (contrary to your assertion) cost scales more with volume. So, you want to make a DT fusion reactor as small as possible. However, size is limited from below by the need to absorb neutrons in the blanket, and this distance is set by nuclear cross sections.
These considerations are what drives the desire the use fusion directly to produce electrical energy, not produce heat, so the apples-to-apples comparison with fission can be evaded. For this reason, I consider Helion the least dubious of the fusion enterprises.
Again, the reactor’s physical structure and cost doesn’t scale 1:1 with the internal volume, so it’s a meaningless metric.
As to your assumptions, a smaller device may be cheaper or more expensive depending on what it takes to make it smaller. It’s a complex machine you can’t just price things by weight or something.
Really the argument is like complaining about fission reactors because all that water is heavy what matters is cost not vague proxies for cost.
One consideration that implies accelerating cost increases is reliability. A fusion reactor will have many Criticality 1 features (like welds that, if they fail, leak coolant into the vacuum vessel). The larger the reactor, the more reliable each such component must be. Reliability is expensive.
That assumes the larger device will be more complex when it’s often simpler to make something larger than it is to scale it down. At the extreme end the sun is vastly simpler than any reactor design we are considering, those are complex because we need to operate on a smaller scale.
Hell we could have built a combined fusion/fission reactor in the 1960’s using Hydrogen bombs and a large enough underground body of water to absorb the energy. Would need to be huge, but not that complicated.
Anyway, a fission reactor’s complexity is mostly outside the device. Things like multiply redundant cooling systems, backups for your backup generators, fuel processing, etc that are completely unnecessary or simpler on a fusion reactor. So overall a large fusion power plant could easily be a simpler overall system.
The magnetic pressure (and hence density, at given beta and temperature) scales as B^2. The fusion power density scales as density squared, or as B^4. Now, the mass of the support structure scales as volume x B^2. So, power per support structure mass (and hence, per cost of that structure) scales as B^2.
However, fusion power density is limited by what the first wall can withstand. Therefore, a large reactor must reduce B (or, reduce beta) to stay within this limit. Reducing B reduces the power/support mass, and hence power/$ of support structure cost.
Fusion reactors require equipment outside the reactor as well, probably much more complex than in a fission reactor. Online tritium recovery, robotics equipment for internal maintenance, and space for first wall replacement (since known materials do not survive the life of the reactor, unlike structures in a fission reactor core.) If you look at ITER, there's a large volume around the core where robotic maintenance devices can slide on a track. The ARC concept required volume and machinery for lifting the entire top of the reactor and replacing the vacuum vessel (then disassembling the old, activated vessel.)
No it does not, just as an example the fusion blanket is a constant thickness for a given level of neutron flux per m2 and sub linear scaling as flux increases. Same deal with your cooling loop etc.
You’re making several other mistakes, but because at a fundamental level the costs relationships are more complex than what you’re modeling.
Consider despite being a novel experimental design and having a huge research staff etc, Vogtle Electric Generating Plant still cost more than ITER to build. Sure the scale is an order of magnitude smaller, but fission has been around for a half century at this point is seemingly much simpler and you’d assume there was a great number of efficiency gains to be had. https://en.wikipedia.org/wiki/Vogtle_Electric_Generating_Pla...
Viewed another way: if we double B, we quadruple the forces on each segment of conductor, and therefore must quadruple the strength of the supports.
I explicitly said "support structure", the structure that is resisting the force of the magnets. There are other parts that scale more slowly, but I don't need to show that all components scale poorly to show that overall there are diseconomies of scale.
Note that this is NOT like in a fission reactor. In a fission reactor, everything scales just about linearly with power.
Yes you do. If A represents 0.001% of the cost then increasing it by 20x doesn’t matter and the support structure is cheap.
Same deal with a fission reactor there’s several diseconomies of scale for example around the length of control rods passive cooling after an accident etc, most of them just don’t matter much.
Further, if electricity from a fusion plant were effectively free (like it didnt need turbines or something) what would large-scale desalination cost even if it did use truly free energy? This is what I’d be excited for from fusion power, because potable water scarcity is a huge upcoming/ongoing problem in some developing nations with very large populations.
Deuterium requires processing a great deal of water, but can be moved around the world cheaply at the levels needed for fuel.
As to the costs of fusion or fission power if the nuclear bits were free, it’s roughly half the cost of coal.
It's a parallel to how with a fusion power plant, even if the reactor only cost $50,000 (some comically small amount near zero), the electricity from the overall power plant connected to the reactor might still be fairly expensive. Like - if you have access to free steam, is turbine generation actually cheaper than solar? Even assuming absolute best-case, fusion's place might not be "lowest cost source of energy" but rather "a notably cleaner and lower-cost base load" which can provide electricity when weather causes the solar and wind farms to underproduce. In 30 years in developed nations, its competition will be batteries, pumped hydro, and other energy storage farms.
The core issues is drinking water is isn’t generally a lack of water. People in rainforests still have issues getting clean water because maintaining a distribution system requires a functioning society/government which simply isn’t available in extremely poor countries.
Similarly keeping water supplies drinkable requires pollution controls which is basically non existent in such countries.
I'm pro-fusion, I'd like to see more resources allocated towards its development, but my outlook on its potential impact is very tempered.
The NSS, nuclear steam supply system, is estimated by the following link to be 12% of the cost of a NPP.
https://world-nuclear.org/information-library/economic-aspec...
So, a fusion reactor an order of magnitude more expensive would double the cost of the entire power plant.
Fuel rods end up 0.6c/kWh and on top of that requires weeks of downtime for refueling etc. People hear nuclear reactor and think that’s the most expensive bit, but it’s really not.
I personally doubt fusion will be cost effective vs renewables + storage, but there’s reasons to suspect it could beat fission at scale.
Keeping electrons cold helps limit both effects, which is one reason why Helion's approach is interesting: the ions there are much hotter than the electrons.
But that’s a lie too. If you check the budget of the Department of Energy, you’ll see that each year about half a billion dollars is allocated to fusion. For 2024 for the number is $790 MM. If that’s not “good funding”, then how do we know there is a number out there that will count as good?
Just for reference I looked up what the moon mission cost. Adjusted for inflation it was $182 billion(multi-year so not directly comparable to annual budget). And this was just for flexing on the soviets.
I didn't say I trust them. You are taking the words I have written and basically overridden what you think I said with what I actually said. No one else can be trusted to have a good take on this except the experts. Fusion is complex topic. You either know what you are talking about (Also known as, you are an expert) or you don't know. If you don't know what you are talking about your opinion on the matter is based on nothing. I rather listen to the expert even though he might have a conflict of interest.
Ok, let me refresh your memory: "I don't know, that's what the experts say". Does that not sound like rhyming with "I trust them" to you?
> No one else can be trusted to have a good take on this except the experts. Fusion is complex topic.
Yes, fusion is a complex topic. But so is building fission nuclear power plants. If the experts in fission power plants keep telling you they can build these plants for X, and they consistently build them for 5X, do you give them a pass? Building rockets is complex too. The SLS system (also known as Senate Launch System) is permanently delayed and over budget. When the experts say they need more money, do you give them a pass, just because their domain of expertise is complex?
Why are you treating fusion any differently?
Take the famous "fusion never" graph [1]. According to that graph, if we had invested an average of about $5 BN over about 15 years, then it would have been the "maximum effective effort". What exactly would we have done in 1976-90 with $50 BN (in 2012 dollars) that would have made a difference? We are now half a century later. An iPhone has infinitely more computational power than the top supercomputer of the time (which was Cray-1, with 160 MFlops). Ok, not infinitely more, but at 2 TFlops and iPhone 15 Pro has more than 10000 time more compute, which means it most likely has more than the entire available compute in the world in 1976. The $50 BN in 1976 could not have bought you a time machine. You would have been stuck with Cray-1 and the likes. Similarly, we have made a lot of progress in the last 50 years in many areas, like software engineering, numerical methods, plasma modeling.
Yet, nobody would think now, in 2024 that you can achieve fusion with $50 BN. It is completely preposterous to think the same amount of money (adjusted for inflation) would have yielded better results in 1976. With what? With slide rules?
You see why I'm calling them liars? In the graph (more like a scribble) they put in front of you telling you that they would have done so much better if only they had better funding, if you actually read the numbers, you can see that they could not have achieved fusion.
Given that, why should the politicians give them money? Imagine you go back in time and some mad scientist asks for for some huge amount of money to build a supersonic plane in 1905. With your knowledge of today, you know that supersonic planes are possible, but you also know that no amount of money in the world can make that happen in 1905.
We are at such a point now. Basically 50 years after the 1976 projection of the famous "fusion never" graph, we know that there was no amount of money that would have achieved fusion in the era of Cray-1.
Mr Fusion is almost 10 years late, according to Back to the Future. I’m still waiting for my hoverboard.
[0]https://www.lockheedmartin.com/en-us/products/compact-fusion...
Notably, what is required for this is superconductors with a higher critical field strength. A higher critical temperature eases cooling requirements somewhat but does not in itself make fusion easier.
Also, quite a few other advances are required before home appliance-sized fusion reactors become feasible. After all, the largest part of most fission plants has to do with generating power from steam, not so much the nuclear reaction itself.
> If anyone could get highly detailed MRI scans whenever they wanted, a huge amount of disease could be prevented.
I think this is oversold. Regular MRI screening without any indication is generally regarded as unproductive not because of the cost of the scan, but because of the high number of false positives. Any normal human body is bound to contain some number of benign growths.
There's also a recent trend in some medical diagnostics of having a lighter touch, instead of running all the tests and potentially drawing the wrong conclusion from heaps of data.
How come then that outside of Western Europe and US it costs $100-200? Or is it because the MRI machines were amortized in EU/US first?
RTGs are not very large, so the technology already exists.
Find out cheaply and quickly if something's a break, a sprain, a connective tissue injury, or something else.
obviously that's untrue, since fusion for power use at the moment isn't "too expensive", it's "too hard". literally hundreds of billions of dollars have been spent and we're still not close.
I can't tell why there are so many uninformed posters on this topic on HN. is astroturfing by the nuclear/fossil fuel industry to delay doing the hard work to eliminate fossil fuel use? is it just that people read too much scifi as a kid and then didn't read any real science on the topic for thirty years?
The reactor will use around 600mw of electricity to generated 500mw of heat, in 10 minute runs. Not break even by any means, but getting much closer than ever before.
ITER uses - I'm estimating - about 5 billion dollars of cryogenic superconductors - 10,000 tons of superconducting tape. These require let's guess an extra 3 billion dollars in cooling capacity and cryonics. Cooling these magnets costs around 150mw of energy, reducing reactor efficiency significantly.
We also use much bigger magnets than theoretically necessary, because cryogenic superconductors quench - instantly lose superconductivity - if they take too much power.
Now imagine we have some 10x cheaper room temp superconductor that is 10x to 100x harder to quench:
+ Magnet mass drops 10-100x (to 1000 tons or 100 tons)
+ Magnet cost drops 100x-1000x (to 50 million or 5 million)
+ Cryonics system is 1.5 billion cheaper
+ We cut the reactor's electricity budget by 150mw, to 450mw
Overall we saved over 6 billion dollars, thanks to a non-fusion innovation.
Keep in mind these numbers are guesses and wishful thinking. But it suggests that a miracle superconductor could push ITER like devices closer to break even, perhaps even achieving it.
No, because fusion reactors (of any currently realistic design) produce extremely harmful radiation in the form of neutrons. Since neutrons are neutral, they can only be stopped by a large shielding mass when they directly collide with atoms in that mass. The mass itself then turns highly radioactive (with a half-life of a hundred years or so, so much more radioactive than spent fission fuel).
Plus, if the magnets fail while the fusion reaction is happening, then the superheated plasma will violently explode in all directions, killing anyone nearby, and spreading the radioactive remnants of the vessel all around.
Finally, these fusion reactors need some quantity of tritium, which is an extremely rare and extremely radioactive form of hydrogen (half life of only a few years) , that is never going to be easily available to sell on a consumer market.
No. Just...no. This is entirely wrong. The density of any proposed fusion plasma is 250,000 times less then the earth's atmosphere[1]. Fusion plasma would be crushed the surrounding air rushing in, not "explode".
Fusion plasma's are incredibly light, and incredibly thin. In the event of a full magnet quench, the only significant damage would be from magnetic quench boil off of coolant...which is a designed for failure mode, and would vent either liquid nitrogen (in HTS designs) or liquid helium in LTS designs like ITER.
Yes It would initially collapse, but still a lot of heat to dissipate, and I would expect an explosion
The thermal flux in the wall is high when the reaction is happening (multiple megawatts per square metre), but the temperature where the beryllium face is bonded to the copper-steel backing structure is below 300 Celsis, so there's not a huge heat storage of immense temperature in the structure.
The plasma itself is 1 gram of mass at 150 million degrees. The tokamak full of sea-level air would be 1 million grams of air, more or less. So it won't superheat 800 cubic metres of air into an explosion, even if there was a huge leak and pressure equalised very fast.
The bigger explosion hazard is the 800 cubic metre vacuum chamber imploding (unlikely, it's only 1 atmosphere difference, it's not a submersible), a magnet quench (which does store a lot of energy that can be released at once, just ask CERN, though the result isn't exactly catastrophic because it's part of the design, but it might not be good for the magnet) or some mundane but dramatic electrical, hydraulic, pneumatic or steam system failure that could happen in a coal power station. Future reactors may have a cooling blanket that could leak liquid lithium.
The problem is getting it hot enough.
Also, I question the explosion claim. Without confinement, the energy drops below the threshold necessary for fusion and the reaction stops. An explosion would require a self sustaining reaction of some kind and I'm just not sure where that would come from. It's not like there's much inertia in it.
But you're right about the neutrons. Maybe you can get away with a rice maker sized reaction chamber, but you're going to want it shielded in something the size of a couch. Probably best to just put it in a deep hole and pump water down there. Then you can provide the neighborhood with steam for heating and electricity for everything else.
I am not a physicist
I expect the sudden failure of magnetic confinement would lead to a chemical explosion because of super heated plasma
What would scare me more is an asymmetric loss of the magnetic field, which would produce a huge and not-designed-for mechanical stress, potentially leading to pieces of the superstructure breaking off and flying at high speeds, pushed by the very strong magnetic field.
But yeah, magnets throwing around chunks of each other, and chunks of radioactive shielding, would be an unpleasant thing.
Startup hype notwithstanding, the other fusion reactions are only incremental improvements over D-T, in terms of emitting less neutron radiation. From the standpoint of radiation protection it's all utterly moot—you'll never have a fusion reactor in a "rice-maker" to "power homes and vehicles". The amounts of radiation are absurdly extreme, and of an exotic kind that's really, really, unshieldable. (Infamously, the whole point [0] of the neutron bomb was a Cold-War tactical antitank weapon. Something that heavy steel tanks, envisioned to be in a nuclear land war in the Fulda Gap or whereever, would have no chance of surviving—not even with all the efforts the superpowers threw at the military challenge).
A sobering fact for those unfamiliar with nuclear physics. The point of less-neutronic nuclear fusion reactions isn't to render them mild enough not to disintegrate humans. It's to make them mild enough to not disintegrate steel [1].
We’re already seeing very promising improvements in batteries (some without even needing lithium), so probably sooner we’ll get wildly high battery energy densities
But the reasons are that its complex and expensive, not that it's accidentally a bomb.
core power volumentric density: 276 w/m^3
core density: 150000 kg/m^3
alternative units:
1.8 mw/kg .276 w/L
hello practical says a typical powered push mower might have about 1.7 KW and might weigh ball park 30 kilos, so ~60 W/kg or tens of thousands more power dense than the sun's core
useful fusion is hard because we're not trying to recreate our sun, we're trying to recreate fusion environments so intense I suspect they don't show up outside of supernovas or shortly after the big bang.
Also, that MRI machine would appreciate a uniform magnetic field, and everything else in the room would appreciate a rapid fall-off outside the machine, and a hula hoop won’t achieve either one.
Besides being a cool physics problem is not actually an economically viable source of electricity.
https://en.m.wikipedia.org/wiki/Superconducting_magnetic_ene...
Tangential, high density energy storage also means things that go kaboom by intent.
Then we could have really thin wires transferring huge amounts of power?
Also politically speaking this may just be a non-starter: on paper, we can already build a conventional low-resistance[0] ground-level planetary power grid for very reasonable prices and material requirements, but much smaller projects connecting Europe to the Sahara have stalled.
[0] 1 Ω at 40,000 km
[1] material costs of a few hundred billion USD, a little over a year of current global aluminium production, more to actually install it
[1] Goldene: A single atom layer of gold:
Would be the most game changing level of efficiency for our world we'd ever see.
I still think we're pretty dang close to it actually happening.