back

by ortusdux·5y ago·view on hn ↗
To the best of my understanding, one of the main reason that MSRs have not gone to market is that the salts are so corrosive that containment over the long term is not currently possible. Note the questions about materials, alloys, and flow sensors in the video.

ORNL had fully functional reactors in the 60's, but those reactors were only safe to operate for a few years, and they were at lower temperatures than what we are targeting today. Liquid fluoride thorium salts at 700+ C will readily dissolve chromium, which makes working with stainless very difficult. Other common alloying agents are susceptible to radiation (Co & Ni transmute when irradiated) which further shortens the lifespan. There is also the issue of tritium, which can permeate stainless steels, cause embrittlement, and escape into the environment.

ORNL developed Hastelloy N to help address these issues, and there is an effort to certify other structural steels to for use in reactors (316H, 800H, inco 617). None of the studies that I have seen indicate that any of these metals will survive for more that 5 years or so.

11 comments
> To the best of my understanding, one of the main reason that MSRs have not gone to market is that the salts are so corrosive that containment over the long term is not currently possible.

I've wondered why "long-term" is such a huge criteria. One of the biggest nuclear issues today is using them way past their lifespan because they are so expensive to build.

Let's assume molten-salt is cheaper to build and less dangerous (from a radioactivity point of view). Why not make expendable designs? Make them smaller so the building requirements are easier. Make them easy to remove and replace. Ease up on the restrictions a bit so they can be replaced every 5-10 years instead of decades.

Long-term mass production has tons of advantages. Costs per unit decrease. Defects per unit decrease (this is part of cost per unit usually, but not if a defect gets through inspection). Recycling used units should radically decrease material costs when getting close to peak reactor count (chain of custody paperwork from mine to installation in the nuclear plant is why an otherwise $0.15 screw winds up costing $50-100). Constant employment will build experience over many years and further decrease errors. Once the site design is finalized and enforced, the reactor design can be gradually improved and given the short lifespan, efficiency will increase a minimum of every decade. Likewise, design mistakes (once caught) will only be around a decade at most instead of a half-century like we see today.

Kibitzing outside of any expertise that I have, here, but thorium reactors are generally regarded as a significant nuclear weapons proliferation risk. This has been an inhibiting factor with regards to investment in research.

This design — cheap and transportable in a standard shipping container — significantly ups the ante on that risk, to put it mildly.

Deployed singly, small, low-power, short-lifespan reactors would increase the burden of maintaining regulatory control over those reactors — destined, by design, for catastrophic failure unless they are retired within a safety period — which in turn would the risk of a nuclear contamination incident. This could be mitigated by operating the reactors in banks at a containment/maintenance site.

I'm not sure what parts of the reactor could be usefully recycled. All the metals would be embrittled by neutron bombardment.

I'm just recalling from the Sorensen's LFTR's videos here, but I had the impression it was the exact opposite in terms of weapon-grade nuclear material.
Anything you take out of a Thorium-fueled reactor is going to contain Uranium-232 and its decay products, which emit easily-detectable and dangerous to nearby living things gamma radiation. U-232 is much harder to separate out from the useful U-233 (which is the fissile fuel that Thorium is converted into) than Uranium 235 is from Uranium 238, so it is very hard to extract clean safe fissile material from a Thorium reactor.

This means that a Thorium reactor is really hard to get material from for a nuclear explosion. Any material you do extract is going to be screaming "I'm over here" to any nearby gamma ray detector, and will be dangerous to handle.

But if you just wanted to make a dirty bomb by blowing up a small Thorium nuclear plant with conventional explosives, spreading highly radioactive material all over the place, that sounds much more feasible.

> Kibitzing outside of any expertise that I have, here, but thorium reactors are generally regarded as a significant nuclear weapons proliferation risk.

Also outside my expertise, but everything I read said the opposite. Wikipedia (which gets true/false questions wrong half the time) says (emphasis added):

Uranium-232 (232 U) is an isotope of uranium. It has a half-life of around 68.9 years and is a side product in the thorium cycle. It has been cited as an obstacle to nuclear proliferation using 233U as the fissile material, because the intense gamma radiation emitted by 208Tl (a daughter of 232U, produced relatively quickly) makes the 233U contaminated with it more difficult to handle.

https://en.m.wikipedia.org/wiki/Uranium-232

PS, HN, Y U NO blockquote?!?!

>PS, HN, Y U NO

IKR?

>so they can be replaced every 5-10 years

Recycling of highly radioactive materials, aka the entire reactor vessel every 5 years, is cost prohibitive.

That short of a lifespan will create massive amounts of waste to deal with.

Most of the by-products decay away much more rapidly in thorium reactors and the rest are super easy to detect. This should have recycling easier (let it set for a few years after powering down then recycle it necessary). If each unit fit into a slightly oversized shipping container, safely transporting it doesn't seem more difficult than transporting in nuclear material in the first place. Plus, since waste is going right back into a new containment, it probably doesn't have to be as clean provided you can keep workers safe.
Actually I did a Ph.d. related to chemical separations of nuclear materials. Granted my direct research was narrow on some aspects of solvent extraction related to actinide separation, but I did get a general impression of other areas from hearing stories and presentations at conferences and such. My impression is that you're vastly underestimating the complexities and challenges of those steps. Yes, this is just an appeal authority type argument, but your point is a lot of generalized high level statements without supporting evidence. What you're saying could be true, but from my impression it's more likely to be magical thinking. I'd love it to be true, but just because we want something to be true or it sounds nice doesn't make it more likely. In fact, we should double our due diligence for claims that tell us what we want to hear.
I agree. But, in a molten-salt reactor, the vessel and heat exchangers are equivalent to fuel assemblies. Dispose the vessel and heat exchangers similar to Zr-clad million dollar fuel assemblies.(PWR/BWR reload costs $60-100 million per 12-18 months.)

Material wise Zr clad is way more expensive than stainless 3XX & graphite combined. Even if Ni-alloy is used in disposable MSRs Ni-alloy+graphite will be slightly cheaper. (Base metal costs: Zr metal: ~$22,700/ton Ni metal: ~$13,100/ton)

Many actinides have volatile fluorides. Fluoride volatility is a proven process used to enrich uranium for LWRs. Fuel salt can be disposed after recovering U & Pu. Vacuum distillation is physical separation, which can also be used to further recover expensive base salt FLi7Be. No vacuum distillation is necessary if inexpensive salt is used.

My guess would be this: the cost of a reactor dominates the cost of electricity for fission, so making a reactor last longer reduces the time-averaged cost. Typically, reactors are expected to pay for themselves within 10-20 years. So, a reactor that self-destructs in 5 years needs to cost 2-4x less. It's not clear that MSRs have that kind of up-front cost advantage.
A recommended video to watch explaining corrosion protection:

https://youtu.be/jO37sMRqv8o

hard not to cringe at the safety standards in that video...
> One of the biggest nuclear issues today is using them way past their lifespan because they are so expensive to build.

Or they're trying to avoid the staggering and bankrupting cost of decommissioning them.

A similar issue exists with metal coolant reactors, which conceptually would be very simple, and would use proven technology, fuel rods and cladding and have similar or better operating characteristics to molten salt: low internal pressure, high temperature, inherent safety (fuel is physically separated from colant and never touches internal piping, pumps etc.)

It's very hard to formulate a metal coolant that has low melting point (pure lead is already too high), does not react with air or water (Sodium cooled reactors are notorious), has low cross-section (Hg fails), does not activate under the neutron flux to long lived chains (Bismuth and potassium in eutectic alloys) and does not dissolve the steel structure of the reactor (Tin).

If this problem could be fixed and have a safe, inert metal coolant with good thermal characteristics, fast breeders would become common place.

Aren't they already used in Russian submarines?
They were used and the design was modified into small portable nuclear reactor, the SVBR-100, a 100MW(e) reactor that can be sold containerised and delivered on a rail flatcar.

The problem is that a) the project owners have issues getting money to finish certification b) we don't mine enough bismuth.

The second problem is pretty major issue, because as far as I know, we don't actually mine bismuth - all the bismuth available worldwide is from processing of tailings in other mines, and isn't under any kind of high production rate.

Making just a dozen or so reactors for submarines was easy. Making mass-production line would actually mean a noticeable drain on world-wide supply of bismuth!

Then there are political problems involving buying russian tech.. :(

They are, but this is because the Soviet designers thought that the improvement in power density were worth the serious disadvantage of having coolant that freezes well above room temperature.
Yes, exactly.

Making a working reactor is as easy as disoling a critical mass of uranyl nitrate in a bucket of vodka.

Making a lasting reactor that will work non-stop for years is not so at all.

This is why I am sceptical about this. Solid fuel is by far more troble free, even if it doesn't last that long.

Why use stainless steel? I thought "stainless" is mainly useful because it doesn't rust (is there water & oxygen in a MSR?). Also, couldn't the steel have another layer inside (e.g. just iron or something else non-reactive) so that it never comes into contact with the salts?
For many of the same reasons that Spacex is using it for their starship. It's cheap and readily available, relatively easy to work with, and very well studied. There also really are not that many materials that have passed certification for use in reactors. I believe that "BPVC Section III-Rules for Construction of Nuclear Facility Components-Division 5-High Temperature Reactors" only covers half a dozen stainless steels.

As for cladding, the DOE's 2021 budget includes money for testing novel claddings. They are also looking for robust redox reference electrodes and additive manufacturing methods.

"Rust" is just oxidation and there are a lot of substances that will cause oxidation.
To add to this, temperature plays a really critical role in oxidation/corrosion of stainless steel (and other metals). The scale that forms at high temperatures has a different composition than what develops at lower temperatures. In certain environment, the scale remains on the surface, inhibiting further oxidation, but in other environments, the scale is continuously removed, eventually destroying it.
Iron is vastly more reactive than stainless steel, that's exactly why we have stainless steel.
The corrosion problem has been blown way out of corrosion, and I suspect, you're confusing it with the graphite swelling problem.

Pure nickel won't corrode in fuel salt, so to a first approximation, Hastelloy N works fine and lasts forever. Chromium is dissolved superficially, but this process is self-limiting. Two more corrosion mechanism were identified: some nickel transmutes and releases helium, which migrates to grain boundaries, and the fission product tellurium reacts with the metal. Solutions for both problems were identified by 1974, but never tested, because the MSR program was terminated.

What is more likely to limit core life, is the graphite. Graphite, when irradiated, first shrinks, then swells. MSBR had a very complicated core made of precisely manufactured graphite pipes. These would crack and deform, and then they need to be replaced. Everyone who is considering a graphite moderated core seems to plan for a long life of the pot and a four year life of the graphite inside it. Copenhagen Atomics plans to use heavy water for moderation, so there is no graphite to worry about.

You're correct. The vessel & primary heat exchanger of a MSR which are in contact with liquid-fuel is equivalent to fuel assemblies of many solid-fuel reactors. In solid-fuel reactors the clad which is designed to contain fuel doesn't last long. Similarly, in MSRs the materials in contact with fuel doesn't last long.

So, we need to dispose the vessel and primary heat exchangers similar to Zr-clad million dollar fuel assemblies.(PWR/BWR reload costs $60-100 million per 12-18 months.) Material wise Zr clad is way more expensive than stainless 3XX & graphite combined. Even if Ni-alloy is used in disposable MSRs Ni-alloy+graphite will be slightly cheaper. (Base metal costs: Zr metal: ~$22,700/ton Ni metal: ~$13,100/ton)

Why not use a stone type of material. Like a furnace?
There are a few reasons, but the biggest one is that ceramics and composites also react with the molten salts. One study from 2015 (1) showed that even testing the corrosion resistance of metals was difficult because the type of crucible used had a large effect on the results.

These systems need pumps at a minimum. Sensors (temp, ph, flow rate, chemical composition, etc.) are also very important for prolonged, low maintenance use. These cannot be made of ceramics (that I know of)

The container will also need to safely hold a dense pool of molten salt. The linked $88k test reactor has a capacity of 350 liters, which at 1.94 g/cc is ~680kg/1500lbs of Li2BeF4.

Another important reason is that heat transfer is the whole point of the reactor, and most metal alternatives have poor thermal properties.

1. https://www.researchgate.net/publication/282433901_Impact_of...

I know on boats they use a piece of sacrificial metal (Anodes) that corrodes more readily than what the boat is made of. This prolongs the life of the boat. Maybe they can use something like that in this situation too.
Furnaces use stone because it is not very sensitive to heat. Stone generally remains sensitive to acid.
Plastics would be great to withstand the acid, but they're sensitive to heat.
Granite, for instance, has a 1200c melting point which sounds good. Its large grainy structure might cause porosity issues. And it's main constituents are silicon and aluminum, both of which do transmute under neutron bombardment. (An expert could say if they transmute to anything you really don't want around, and how long it'll be there.)

I think ceramics are an answer to the porosity and similar issues. Consistent structure, no inclusions or other weaknesses. But they would face the same transmutation and induced radioactivity problems as stone of the same material.

You mean ceramics?

There are so many to chose from, and most are unusable for one reason or another. The leading candidate being investigated is carborundum (silicon carbide). It doesn't corrode in molten salt, it low appetite for neutrons, is reasonably easy to manufacture, conducts heat fairly well. It's even mentioned in the video.

Is there a reason except price that titanium isn’t used? It has a higher melting point than steel and resists acids.
I think you have to have oxigen present for it to be corrosive. Can't they vacuum it?
From an engineering standpoint any chemical reaction that changes the properties of a metal in an unwanted way will be called corrosion. It is something to prevent and/or monitor that will inevitably shorten the lifespan of the metal part. (Non-metals like fiberglass can also "corrode".)
You need an oxydizer whatever it is. Oxygen is only one oxidizer. Chlorine, fluorine and the other halogens are oxidizers too.
Fluoride (with a 'd') is not fluorine (with a 'n'). The difference is as big as that between sodium and sodium chloride. You can eat one, but not the other.

The most powerful oxidizer in MSR fuel salt is the U(IV) ion.

IIRC fluorine is/can be a more potent oxidiser than oxygen.
"Corrosion is a natural process that converts a refined metal into a more chemically stable form".

Oxidation is the most well known form of corrosion, but there are many other corrosive reactions out there.

https://en.wikipedia.org/wiki/Sulfide#Corrosion_induced_by_s...

To contain molten salt use solid salt.
Would diamond do better?
Diamond will burn like super-expensive coal. Probably not ideal for those temperatures.
Umm, that's doubtful. This isn't oxygen that generates a gas on reaction. You would need to look at the actual conditions at moderate temperature (800-1000F is only ~400-500C) for metal fluorides in contact with diamond films and carbon solubility. Even fluorocarbons tend to actually be moderately stable in these temperature conditions.

Diamondlike film coatings should last longer than carbon or graphite crucibles (since those materials are far softer). Those have been tested, although it would seem that high nickel (rather than floridizing Cr) steel is probably the prefered material. SiC also works and they discuss this, but there are concerns about Tritium absorption for the larger crystal structure and porous sintered components.

For some comments and measurements, see this thesis on metal fluoride corrosion control for reactors: http://fhr.nuc.berkeley.edu/wp-content/uploads/2014/10/06-00...

research on SiC also continues: https://www.osti.gov/servlets/purl/1531263