Overview
Ruthenium is the one metal whose supply cannot respond to price. World production is approximately 30 t/year, obtained exclusively as a by-product of platinum and nickel refining, roughly 90 % of it in South Africa, where PGM output is contracting. The mining decision is taken on the platinum, palladium and rhodium basket; ruthenium is recovered because it is already in the refinery feed. The supply curve is therefore vertical over any horizon shorter than a new mine, and a demand shock passes entirely into price.
That shock has arrived. Ruthenium traded at $1,750/oz in March 2026 against $560 a year earlier, and at $62.59/g — $62,590/kg — in 2026, a rise of 609.73 % since 2020. The forecast 2026 deficit of 203,000 oz is 6.3 t, about one fifth of world output. The entire world market is under $2 bn a year — less than one mid-sized data centre that cannot be built without it.
ARBOK-Ruthenium is the argument that the only expandable source is the fraction currently discharged, and the process that recovers it. Three aqueous streams carry dissolved ruthenium today and are managed as hazardous waste. In all three the dissolution step — the part that dominates the cost and the environmental burden of primary production — has already been performed and paid for by someone else.
Applications
Advanced-node interconnects. Ruthenium displaces copper and cobalt because its electron mean free path is roughly six times shorter, so surface and grain-boundary scattering degrade its resistivity far less at nanometre cross-sections. It also deposits barrierless, where copper needs a diffusion barrier that consumes a substantial fraction of a few-nanometre trench. Below 3 nm the two effects together make the substitution a requirement, not a preference.
Hard disk recording media. A ruthenium spacer under one nanometre thick sets the antiferromagnetic coupling between magnetic layers, and that coupling is what permits the areal densities of high-capacity nearline drives — the storage tier of every hyperscale data centre. Two decades of substitution effort have not displaced it.
Chlor-alkali. Ruthenium dioxide is the electrocatalytically active component of dimensionally stable anodes, a role established for half a century.
Catalysis. Ammonia synthesis; acetic acid carbonylation as promoter.
Passives. Ruthenium dioxide is the standard resistive phase in thick-film chip resistors, tens of billions of units a year.
None of these applications sets the price; all of them sit behind the same 30 t of supply.
Operating Principle
The inversion. Classical recovery extracts a solute from water. ARBOK-ZWD removes water from the solution. The stream is processed whole under deep vacuum at the temperature of the surroundings, where water separates from the dissolved load as a phase change rather than a thermal evaporation — which is where the low specific energy comes from. Clean water is returned at up to 100 % of intake by volume; the entire dissolved load leaves as dry separated fractions.
Why that matters here. Process cost is set by volume, not by concentration. A stream at 0.5 mg/L and a stream at 5 mg/L are processed identically and at identical cost; only the mass in the dry residue differs. This removes exactly the constraint that makes classical recovery uneconomic on dilute streams.
Why classical methods fail — economics, not chemistry. Sorption, solvent extraction, ion exchange and selective precipitation are configured around a target species at workable concentration. Their cost scales with volume processed and consumable replacement, while recovered mass scales with concentration. Below roughly single-digit mg/L the two curves cross, and they cross even at $62,590/kg, because reagent and media cost is incurred against the whole volume while the metal is present only in trace. The operator's rational response is the one observed: treat ruthenium as a contaminant, remove it to permit limits by the cheapest means, dispose of the sludge. The metal is paid for twice and recovered zero times.
Non-selective by design. The process does not choose between ruthenium and the rest of the dissolved load, so companion platinum-group metals at a refinery, or copper, cobalt and tungsten at a fab, are recovered in the same pass. No single fraction has to carry the economics alone.
Second stage. The module produces a concentrate, not a product. Conversion to individual metals requires a compact refining unit developed as part of the same platform, yielding ruthenium alongside platinum, palladium, rhodium and iridium. At tens of tonnes of world demand this is a modest industrial facility, not a national programme, and can be sited in any jurisdiction.
Key Parameters
| Parameter | Value |
|—|—|
| Working pressure | deep vacuum |
| Working temperature | ambient; no thermal input beyond the vacuum duty |
| Specific energy | under 1 kWh/m³ of stream |
| Water return | up to 100 % of intake by volume; zero discharge, no brine, no reinjection |
| Solids output | dry separated fractions |
| Consumables | none — no membranes, filters, reagents, coagulants or ion-exchange media |
| Cost dependence on feed concentration | none — cost is set by volume |
| Form factor | standard containerised modules, nominal 200 m³/day, single or cascade, continuous, open air |
| Module throughput | 73,000 m³/year |
| Second stage | compact refining unit yielding Ru, Pt, Pd, Rh, Ir |
Market context for the metal:
| Parameter | Value |
|—|—|
| World production | ~30 t/year, exclusively as by-product of Pt and Ni refining |
| Concentration of supply | ~90 % South Africa; output declining |
| Price 2026 | $62.59/g = $62,590/kg; +609.73 % since 2020 |
| Price March 2026 vs a year earlier | $1,750/oz against $560/oz |
| Forecast 2026 deficit | 203,000 oz = 6.3 t, about one fifth of world output |
| Total world market | under $2 bn/year |
Architecture and Components
Containerised deep-vacuum separation module at nominal 200 m³/day: deep-vacuum stage, vapour condensation and clean-water return, dry-fraction collection. Modules deploy singly or in cascade and run continuously in the open air. Downstream, a compact second-stage refining unit converts the concentrate into individual platinum-group metals.
Advantages
Against dilution. Cost independent of feed concentration is the property that makes these streams viable at all; everything else follows from it.
Against the treatment obligation. PGM refinery effluent and semiconductor etch baths are regulated hazardous waste, and their handling is a recurring operating cost with permitting exposure. A zero-discharge installation does not reduce that cost — it removes the regulated object, because there is nothing left to discharge.
Against water constraint. Leading-edge fabs are generally water-constrained by permit, so returning water to circulation has value independent of the metal.
Companion recovery. Platinum, palladium, rhodium and iridium come out of the same pass at a refinery; copper, cobalt and tungsten at a fab.
Operationally. Under 1 kWh/m³, no consumables of any kind, containerised, open-air siting.
Integrations
ARBOK-Rhodium · ARBOK-Iridium · ARBOK-Purification · ARBOK-CRYSTALLIZER
The same platform and the same containerised module serve the rhodium, iridium and other platinum-group metal cases; the streams overlap, and the second-stage refining unit is shared.
Deployment & Operation
Three source streams.
- *PGM refinery effluent.* Precious-metal refining is an aqueous process — dissolution, selective precipitation and electrowinning necessarily place metal into solution, and no separation train is perfectly efficient. A fraction of the ruthenium, with fractions of platinum, palladium, rhodium and iridium, reports to electrolysis bleed streams and wash waters. The metal is accounted as loss rather than inventory, and consequently is not assayed with the rigour applied to the main product stream.
- *Semiconductor fabrication effluent.* Ruthenium enters the fab as a deposition target or precursor and leaves in three places: the wafer, the exhaust and the water. Chemical-mechanical polishing removes deposited metal into a slurry; etch and clean steps dissolve it into spent baths; every process step is followed by rinses. The metal-bearing fraction is a concentrated bleed rather than the whole flow. This stream is unusual in two ways: the facility has rigorous analytical capability and complete process knowledge, so the operator knows exactly what is in the water; and the facility is water-constrained by permit.
- *Catalyst plant effluent.* Regeneration and washing operations generate solutions in which ruthenium is dilute and mixed with support residues. These are discharged.
Siting. Taiwan, Korea, Japan and the United States host the fabrication capacity and therefore the fab effluent. South Africa, Zimbabwe, Russia and Canada host the PGM refineries. The European Union hosts chlor-alkali production and assembly.
The strategic asymmetry is sharpest in the United States and the European Union: both are committing tens of billions to domestic semiconductor capacity, and neither holds a platinum-group metal mine. A fabrication plant can be built; the ruthenium to feed it cannot be mined domestically. The only ruthenium that can be made to appear inside those jurisdictions is the ruthenium already imported inside materials and currently discharged with the wash waters of the same plants.
TRL
TRL 8 (confirmed by Michael).
Preprint published 5 August 2026; the recovery platform underlying this application is industrially proven across other ARBOK Vacuum Cracking deployments.
Market Potential
Demand is coupled to the fastest-growing capital-expenditure programme in the world economy while supply is fixed and declining. The deficit arithmetic is the market: 6.3 t of shortfall against 30 t of supply, with no mechanism by which price can call forth additional primary output.
Recovery at the scale described does not flood a $2 bn market; it closes a gap that currently has no other means of closing.
Typical Project Economics
Single module at 200 m³/day, that is 73,000 m³/year, valuing the full recovered volume at 20 % below prevailing 2026 prices:
| Feed concentration | Ruthenium recovered | Value |
|—|—|—|
| 0.5 mg/L | ~37 kg/year | ~$1.8 M |
| 1 mg/L | ~73 kg/year | ~$3.7 M |
| 5 mg/L | ~365 kg/year | ~$18 M |
Companion platinum-group metals from the same stream, and the returned water, are additional.
Two remarks on method, stated in the source. The full recovered volume is valued rather than a notional absorbable fraction, because the binding constraint on ruthenium is availability rather than appetite; the 20 % discount is applied deliberately to acknowledge that a genuine alternative source would move the price. And these figures are one side of the ledger only — the removal of the treatment obligation sits on the other.
Risk Factors
Substitution in interconnects. Molybdenum is under active evaluation for the same physical reasons, and the outcome of that competition is not settled. The source states explicitly that it does not assume ruthenium wins every interconnect socket. Recording media carry no equivalent question.
Characterisation. Ruthenium concentrations in refinery bleed and fab effluent are not published and vary by facility and by process step. The ranges used are illustrative; site-specific assay is a precondition for any commercial commitment, and the source presents a range rather than a single figure by design.
Market thinness. Under $2 bn a year in total, so volume released without regard to price would move the market.
Grade. Concentrations are low in absolute terms. The counter-argument is that the relevant metric is cost per kilogram of contained metal delivered to a refining stage rather than percentage in the feed — valid only for as long as process cost genuinely remains independent of concentration.
Readiness. The second-stage refining unit is described at concept level and requires further engineering definition for this specific application.
Related Technologies
ARBOK-Rhodium · ARBOK-Iridium · ARBOK-Purification · ARBOK-CRYSTALLIZER · ARBOK-Brine