Metallium's Flash Joule Bet: A Promising Processing Platform-or an Industrial Step Change?

The company structure has two core business units:

• Under their Urban Mining Unit (Waste Recycling), a ‘Build-Own-Operate’ model, Metallium will purchase feedstock, own and operate the processing facility, and retain full economic interest in the recovered metals. Their first processing hub in Houston Texas is operational, and they already have additional hub locations earmarked in the US, and further expansion ambitions for Asia-Pacific, Europe, and the Middle East. The company contends that even a smaller 1-to-10 ton per day operation can be profitable meaning that the modularity and scalability of their technology can produce positive cashflow with modest plant sizes while these same plants are simultaneously built out to higher capacities.
• Under their Mineral Processing Unit (Mining Projects), a ‘Processing-as-a-Service’ (PaaS) model, Metallium partners with miners or processing plant owners by supplying equipment and services, and then monetizes its technology via ongoing per-ton licensing fees and royalties linked to production. This approach enables the company to scale with minimal capital intensity by leveraging their partners' existing infrastructure while maintaining high-margin technology exposure.

Reply Like

I want to raise something carefully, because I may not be understanding the process correctly, and I would rather ask than assume.

I went back to the Rice paper behind the FJH-Cl2 claims (Xu et al., PNAS, Sept 2025). Reading the methods, the sequence looks like this:

1. Calcine the magnet waste at 800 C in air for 30 minutes
2. Grind and sieve to 53 microns
3. Flash under chlorine, which volatilizes Fe, Co, Cu and Ni as chlorides
4. Left with a solid rare earth oxychloride residue at around 94% purity

Step 3 is the second-long step, and it is impressive. But steps 1 and 2 are not seconds, and step 4 is a solid.

What I do not follow is what happens after step 4. To separate individual rare earths, you need ions in solution. So the residue still has to be dissolved, presumably in HCl. The Ucore collaboration describes RapidSX as chloride-compatible with FJH upgraded feedstock, which to me reads as confirmation that a chloride leach sits between the two.

If that is right, then what the flash step actually replaces is the bulk acid leach of the whole magnet plus the iron removal stage. That is genuinely worth having, because you then only dissolve a fraction of the incoming mass and you avoid the iron precipitation and the iron-bearing effluent. But it is not the same as removing acid from the process.

On the cost side, I am also struggling, even at small scale. You have added an 800 °C half-hour roast. You are consuming chlorine to convert the iron content of the feed into FeCl3, and iron is the majority of the mass in an NdFeB magnet, so I assume that is not a small reagent stream. The FeCl3 that comes back out is a low-value water treatment chemical. And you still need a condensing and scrubbing train for FeCl3, CoCl2, and unreacted chlorine, which is corrosive and not trivial to run continuously.

So my question to the group. Does anyone know the system boundary of the LCA and TEA? Is the comparison magnet in to separated oxide out, or is it only the block that the flash step substitutes for? And are the calcination and the chlorine inside that boundary or outside it?

I am not trying to knock the technology. The thermodynamic window they found is elegant, and the point in the article about integration into conventional separation circuits is exactly the one I keep coming back to. I just want to understand what is being compared with to better understand. Corrections very welcome.

Reply Like

I just want to understand what is being compared with to better understand. Corrections very welcome.

In my view, valid questions for almost every modern technology's claims. They are all ambiguous on the boundaries of what their statistics encompass when comparing to traditional SX. Practically all seem to nonchalantly exclude necessary parts of the full end-to-end flow when talking of efficiency, energy use, economics, speed, or environmental impact, etcetera.

Relative to magnets (only one of many potential FJH feedstocks), almost all long-loop processing technologies are going to require demagnetization prior to grinding. (Magnets or magnetic particles in metal machinery would be 'catastrophic'.) Some like HyProMag alternatively choose short-loop magnet-to-magnet route with something like Hydrogen Decrepitation.

youtu.be/AY_gnlqbT5I?si=vqyTetGoUUQvVrtf&t=1202
youtu.be/AY_gnlqbT5I?si=bl3fNBGM0TLVcO3p&t=1337

The temperature of the flash can be used to (1) form and directly evaporate metal chlorides, or (2) form and then selectively evaporate metal chlorides based on differences in their boiling points, or (3) selectively form metal chlorides based on their Gibbs FEoF and their Arrenius activation speed (amount of energy or kilojoules per mol needed), and then evaporate based on boiling point. As you might surmise, minerals that are more varied along these aspects are easier to distinguish and directly separate.

While the technology platform is very versatile and agnostic, specific flow sheets are developed for each type of feedstock and processing parameters are then adjusted for each run.

Yes, chlorine is consumed unless the chlorides are reduced to oxides and the chlorine recovered. It's a considerable OpEx expense. Self production with a chloro-alkali plant can reduce procurement cost by almost half.
In some cases Metallium produces individual metal chlorides, in some cases they produce MRECs. MRECs can be sold or can require further separation. Metallium can also go to oxides if it makes economic sense, and they are developing further separation alternatives as well. They also have agreements with UCore for fully polished separations when needed. Early plans have been to derive most initial revenue from metal chloride products in a brine format.

Cheers

Reply Like

Thank you, Deven; this is helpful.

You are right about demagnetization, and I should have had it on my list. And your point that magnets are only one feedstock is fair. I was reading the magnet case in the PNAS paper and generalizing further than the paper allows.

The part I want to hold onto is your chlorine answer. You confirm it is consumed unless the chlorides are reduced back to oxides and the chlorine recovered, and you call it a considerable opex expense. That is exactly what I was unsure about, so thank you for being direct.

Which leaves one thing. The Rice paper reports 87% less energy and 54% lower opex against conventional hydrometallurgy. Given what you have just said about chlorine, does anyone know whether the chlorine and the 800 C calcination sit inside that comparison or outside it?

If inside, it is a very strong result. If outside, the numbers describe the block that the flash step replaces rather than the whole flowsheet.

Thanks,
Yehuda

Reply Like

Thank you, Deven; this is helpful.
...

If inside, it is a very strong result. If outside, the numbers describe the block that the flash step replaces rather than the whole flowsheet.

Yehuda,
I've read on over 30 different company technologies and you will find they are all somewhat ambiguous and unclear on their statistic boundaries, so it's hard to do relative comparisons. My impression is that most try to paint an attractive picture and most often supply their statistics only for their showcase technology steps. They are not all talking about the same feedstocks nor the same final products, so that also creates uncertainty. Secondly, many of the studies/statistics age fairly rapidly as the technologies and processes evolve and get tuned for commercialization. Early Rice papers are probably not representative anymore since Metallium has piloted Revision 1 and Revision 2 of their reactor already since the lab and bench and pilot work in some of the older studies. The company continues to express that expectations are being met and improvements are being made.
Metallium for example, just announced that they are now doing more beneficiation in front of their FJH reactors. They say: "The [updated] proprietary flowsheet is designed to recover saleable products from lower-value fractions through conventional processing while directing high-value material to the FJH platform. This integrated approach is expected to improve reactor utilisation, enhance overall project economics and create additional commercial opportunities across precious metals, gallium, germanium and other strategic products." In the past they have said it takes about $30-$50 electricity per ton of feedstock for FJH. But now they will actually process less volume of that ton through the FJH reactor and separate more out front with other machines consuming energy in different amounts. (Companies generally are probably not going to reveal a lot of their operational details.)

So what is the objective? To find a technology or company that produces the highest purity products? Have the lowest processing costs? Have the best recoveries? Create the least environmental impacts? Be the most profitable? ... They all matter, but it's going to be challenging getting accurate and comparable specifics anywhere. Are you looking for optimal technology, or optimal investment potential. (?) Those may not necessarily be the same. Companies are going to win or loose in the processing space based on many factors, not just being the 'best' at something.
I sympathize with your quest for comforting due diligence, but unfortunately it may end up being somewhat unfruitful or frustrating based on my experience seeking similar.

My belief is that the best solutions (and investments) will need to:

• Be capable of relatively rapid deployment
• Provide modularity and scalability
• Occupy reduced physical footprints
• Be flexible and feedstock-agnostic
• Be far more efficient and produce faster throughput
• Produce accelerated separations and higher yields
• Require lower energy and water consumption
• Require fewer inputs and inventories
• Reduce acids and reagent usage and waste byproducts
• Be more environmentally compatible
• Be constituted of non-restricted equipment and processes
• Have lower CapEx and OpEx requirements
• Demonstrate attractive and durable economics

I continue to be fairly impressed with Metallium and their potential, and as an investment opportunity. They are not 'the best' at everything. But management is sharp enough to compete in areas where they have advantages and can participate profitably. And I think the playing field is fairly open for companies who are swift and adept at getting their teams onto the pitch. What I like about Metallium is that the technology is very flexible and broadly utilizable. It has capability to be able to pivot when necessary and go after niches or more profitable channels as the market evolves, so in that sense I think they have durability and survival skills as we progress through unknowns and competition. With PCB feedstock alone they are basically their own masters, buying readily available feedstock and producing gold and silver for sale... It's hard to mess that up I would say, : ) and the economics are compelling. The multiple other ways they can make money are frosting on the cake.

Cheers. -D

Reply Like

D,

This is very helpful, thank you. Your distinction between optimal technology and optimal investment is the right frame, and I had been sliding between the two.

I agree comparison across companies is close to impossible. My hesitation is narrower than yours. In magnets, which is the area I know best, I cannot make the economics work. Chlorination breaks the alloy down to mixed oxides, so you then have to pay your way back up through separation, reduction, and alloying. The short-loop routes that keep the alloy intact avoid all of that. And FJH is only partially replacing the leach step, which is not where the cost sits.

So my question, and I hope you may know:

Now that they have moved more beneficiation in front, what share of inbound tonnage actually goes through an FJH reactor?

That number would tell me a lot about how to think about the whole thing, and I cannot find it disclosed anywhere.

Best,
Yehuda

Reply Like

Now that they have moved more beneficiation in front, what share of inbound tonnage actually goes through an FJH reactor?

That number would tell me a lot about how to think about the whole thing, and I cannot find it disclosed anywhere.

Yehuda,
The beneficiation referred to is for PCB feedstock, but the principle applies to others as well. Metallium will assess how to accomplish the processing goal in each case in the most efficient, economic, environmental manner, and design a flow sheet with that in mind. They are using the nameplate FJH technology where it is most effective, and more primitive (non-proprietary) approaches where they produce good results at lesser cost. The comments were made about minute 11:30 - 13:30 on the investor webcast:

The company will be publishing more details in the next 2-3 weeks.
But in general, flowsheets and processing parameters are going to vary from feedstock to feedstock.

If you are looking for technology with a focus or primary objective of processing magnets or their constituent materials, I think FJH would not be a top choice. It is something the tech can do, but there are more specialized technologies and larger companies that are dedicated to it. I think there are probably other more attractive opportunities for Metallium than NdPr.

Some ex-China Companies:

Ionic Rare Earths - Multifunctional Amide Ionic Liquids - high-purity separation with fewer stages and lower reagent consumption compared to traditional methods.
HyProMag (Mkango/CoTec) / HPMS Hydrogen Decrepitation (HD)
REEMAG - Powder-to-Powder (Direct)
Noveon Magnetics, Directly recycling magnets without full separation, claiming 90% energy reduction compared to traditional manufacturing.
REEcycle - Reclaimable Selective Solvent - This solvent is designed to be highly selective, meaning it "ignores" the iron, boron, and coatings (like nickel or copper) and only reacts with the rare earth elements
Iluka Resources -
USA Rare Earth - Continuous Ion Exchange (CIX)
MP Materials - SX +
REalloys Inc. - Saskatchewan Research Council (SRC),
Okon Recycling: focusing on the industrial-scale recovery, harvesting, and preparation of rare earth magnets for direct reuse.
Magreesource: A French company that has developed proprietary hydrogenation technology to re-manufacture magnets from recycled powders. Their (HPMS) dismantles magnets from end-of-life products turning them into a powder form while simultaneously demagnetizing them. Also developing 4D magnets printed by additive manufacturing,
TdVib LLC / RecycleForce / Terves - Acid-free Dissolution Recycling (ADR)

Reply Like

Some other fairly interesting Magnet info from Neo Performance Materials Plant and Process in Narva Estonia:

Stage 1 - Neo's Rare Earth Magnet Making

00:08 - Step 1 - Alloying
00:55 - Step 2 - Strip Casting
01:47 - Step 3 - Hydrogen Decrepitation

Stage 2 - Neo's Rare Earth Magnet Making

00:10 - Step 4 - Jet Milling
00:57 - Step 5 - Alignment Pressing
01:41 - Step 6 – Sintering

Stage 3 - Neo's Rare Earth Magnet Making

00:09 - Step 7 - Grinding
00:55 - Step 8 - Multi-wire Cutting
01:39 - Step 9 - Chamfering
02:11 - Step 10 - Phosphating
02:58 - Step 11 - Spray Epoxy

Reply Like