🔍ESC
navigateselectescclose
The LibraryRecycling & Refining Tech

Recycling & Refining Tech 2026 Investor Overview

Recycling & Refining Tech — 2026 Investor Overview

Scrap flows, e-waste recovery, Miller/Wohlwill chemistry, assaying. A complete 2026 overview of where the market sits and what to watch next.

Contents21 sections
  1. 01Global scrap flows
  2. 02Gold scrap
  3. 03Silver scrap
  4. 04PGM scrap
  5. 05E-waste PGM recovery
  6. 06Concentration economics
  7. 07Operational reality
  8. 08Miller chlorination vs Wohlwill electrolysis
  9. 09Miller process (chlorination)
  10. 10Wohlwill process (electrolysis)
  11. 11Operational stack
  12. 12Fire assay vs XRF assaying
  13. 13Fire assay
  14. 14XRF (X-ray fluorescence)
  15. 15When to use which
  16. 16Urban mining: Umicore Hoboken and Boliden Rönnskär
  17. 17Umicore Hoboken (Belgium)
  18. 18Boliden Rönnskär (Sweden)
  19. 19Other notable operators
  20. 20The marginal-supply question
  21. 21Read next

Recycling & Refining Tech 2026: A Precious Metals Investor Overview

Recycling refining technology is the silent half of the precious metals supply equation. In 2024, recycled gold supplied roughly 1,370 tonnes against mine production of approximately 3,660 tonnes — about 27% of total supply, per World Gold Council data. For platinum-group metals (PGMs), recycling typically supplies 25-35% of palladium and platinum, and over 80% of rhodium when including autocatalyst loops. This pillar overview covers global scrap flows, e-waste PGM recovery, the Miller chlorination versus Wohlwill electrolysis chemistries, fire assay versus X-ray fluorescence (XRF) assaying, and urban mining at flagship operators such as Umicore Hoboken and Boliden Rönnskär. The recycling refining stack is the source of price-sensitive marginal supply that institutional investors consistently underweight in their forecasts.

Refinery floor with gold dore bars
Recycled gold and PGMs supply 25-35% of annual demand, with sharp price elasticity above key thresholds.

Global scrap flows

Gold scrap

Gold scrap is dominated by jewellery returns, particularly from India, Turkey, the Middle East and East Asia. The price elasticity is well-established: a sustained 20% rise in spot gold typically delivers a 10-15% rise in scrap supply within two quarters. Indian scrap is unusually price-sensitive because households use gold as a savings instrument and monetise during distress.

Silver scrap

Silver scrap, per Silver Institute data, runs at 180-200 million ounces annually, dominated by industrial recovery (X-ray film, photovoltaic offcuts, electronic contacts) and jewellery. Silver recycling is structurally less responsive to price than gold because much industrial silver is uneconomic to recover at low concentrations.

PGM scrap

Platinum, palladium and rhodium scrap originates almost entirely from spent autocatalysts, with smaller streams from chemical-process catalysts and electronics. The autocatalyst loop has a 10-15 year lag, meaning 2026 scrap volumes reflect new-vehicle PGM loadings from 2011-2016.

Metal2024 scrap supply (tonnes)% of total supply
Gold1,37027%
Silver5,750 (≈185 Moz)18%
Platinum5025%
Palladium9532%
Rhodium1238%

See /categories/scrap-flows.

E-waste PGM recovery

Global e-waste topped 62 million tonnes in 2022 (UN Global E-waste Monitor), with formal collection rates of 22%. The metallic value embedded in e-waste exceeded USD 91 billion. PGM concentrations in e-waste vary materially.

Concentration economics

  • Smartphones: 0.024 g gold + 0.25 g silver + ~9 mg palladium per unit.
  • Desktop PCBs: 200-250 g/tonne gold equivalent (vs 1-5 g/tonne in primary ore).
  • Hard-disk drives: 5-10 g/tonne PGM.

Urban-mining grades for gold are typically 50-100x richer than primary ore, making e-waste a high-margin feed where collection logistics work.

Operational reality

The bottleneck is not metallurgy — it is collection. Informal-sector e-waste recycling in Agbogbloshie (Ghana), Guiyu (China, now significantly remediated) and Delhi recovers metal at lower cost but with severe environmental and health externalities. Formal-sector operators such as Umicore, Boliden, Aurubis and Glencore Horne refine to LBMA Good Delivery standards. See /categories/e-waste-pgm.

Miller chlorination vs Wohlwill electrolysis

The two dominant gold-refining chemistries deliver different purity outputs at different costs.

Miller process (chlorination)

Developed by Francis Bowyer Miller in 1867. Chlorine gas is bubbled through molten doré at ~1,100°C. Base metals and silver form chlorides that volatilise or float as slag. Output purity: ~99.5%, suitable for industrial gold but not Good Delivery.

Wohlwill process (electrolysis)

Developed by Emil Wohlwill in 1874. The Miller-output anode is dissolved in chloroauric acid (HAuCl₄), and pure gold deposits on a cathode. Output purity: 99.99-99.999%. Slower and more capital-intensive, but the only route to LBMA Good Delivery 99.5%+ bars in practice (modern refineries target 99.99%).

Operational stack

Most large refineries (Valcambi, Argor-Heraeus, Metalor, PAMP, Rand Refinery) run Miller as a first stage to remove silver and base metals, followed by Wohlwill for the final purity step.

ProcessPurity outputThroughputCapex
Miller99.5%HighLow
Wohlwill99.99%+Low-mediumHigh
Aqua regia (small-scale)99.9%LowLow
Solvent extraction (PGM)99.95%+LowHigh

See /categories/miller-wohlwill.

Fire assay vs XRF assaying

Fire assay cupellation in lab
Fire assay remains the LBMA-referee technique, with accuracy to ±0.01% for gold via cupellation.

Fire assay

Fire assay (cupellation) is the LBMA-referee method. The sample is fluxed with lead oxide, melted, then cupelled in a porous magnesium-oxide crucible at 950-1,000°C. Lead oxidises into the cupel, leaving a precious-metal bead. Accuracy: ±0.01% for gold. Destructive and slow (several hours), but definitive.

XRF (X-ray fluorescence)

Handheld and benchtop XRF analysers (Olympus Vanta, Thermo Niton, Bruker S1) measure characteristic X-ray emissions from a sample under primary X-ray excitation. Non-destructive, results in 30-60 seconds. Accuracy: ±0.1% for surface composition, but only the top 10-50 microns are interrogated, so gold-clad tungsten and similar fakes are missed unless the bar is sectioned.

When to use which

  • Refinery throughput grading: XRF.
  • Final settlement against Good Delivery: fire assay.
  • Counterfeit screening on coins: ultrasonic + XRF + Sigma.
  • Doré bar analysis at mine gate: fire assay (referee) + XRF (quick).

See /categories/assaying-fire-xrf and the cross-pillar /categories/storage-security.

Urban mining: Umicore Hoboken and Boliden Rönnskär

Umicore Hoboken (Belgium)

Umicore's Hoboken plant treats roughly 350,000 tonnes per year of complex feed — e-waste, autocatalysts, industrial residues — to recover 17 metals including gold, silver, platinum, palladium, rhodium, ruthenium, iridium, indium and selenium. The IsaSmelt furnace plus copper electrorefining circuit is the metallurgical core.

Boliden Rönnskär (Sweden)

Rönnskär, on Sweden's Bothnian coast, operates an electronic-scrap line (the Kaldo furnace) that is the world's largest dedicated e-scrap smelter, processing ~120,000 tonnes/year. Outputs include gold, silver, copper, zinc, lead and selenium. The plant's emission and energy-recovery profile is the EU benchmark.

Other notable operators

Aurubis Hamburg, Glencore Horne (Quebec), Mitsubishi Materials Naoshima (Japan), Sumitomo Toyo (Japan), Korea Zinc Onsan. See /categories/urban-mining and /categories/mining-equities.

The marginal-supply question

For an investor modelling gold or PGM supply over a five-year horizon, scrap is the most price-elastic component and the most frequently mis-modelled. Mine supply is contracted by capex cycles 7-10 years deep; scrap supply responds within 1-3 quarters. The implication: in a sustained price rally, scrap absorbs more of the demand response than mine production, capping the upside more aggressively than equity analysts typically assume.

Scrap supply elasticity vs price chart
Scrap is the price-elastic supply component; mine production responds on a 7-10 year capex cycle.
About the Author

Dr Abdur Rashid

Editor-in-Chief

Site admin since 2026.

View profile · all dispatches
Discussion

Reader Letters

The mailroom is empty.Be the first to write in.

All correspondence is read by an editor before publication.