Mining Recycling
and Circular Economy Approach

Green copper premiums exist bilaterally at around 2% over LME, at trivial volumes, and the LME has discussed sustainability-linked contracts without implementing any that distinguish primary from secondary, and the LBMA's responsible sourcing framework covers conflict risk and ESG without touching the recycled-versus-mined question, and the Aluminium Stewardship Initiative certifies production standards without generating a price signal at the exchange level, and so a mining company raising project finance against a forward curve and a recycler trying to finance a hydromet plant to process end-of-life batteries are competing for the same capital in a market where the output product, refined cobalt or nickel sulfate, trades identically regardless of whether it came from a laterite in Sulawesi or a battery collection program in Saxony, and the financial system structurally cannot see the difference. That is the foundation on which everything below rests, and it does not come up in most circular economy discussions about mining, which is odd.

This gets technical and there is no way around it.

Primary aluminum exits the Hall-Héroult cell at 99.7%+ purity. The iron limit for 5182 can end stock is 0.35%. For 6111 automotive closure sheet, 0.40%. For 2024 aerospace plate, 0.50%. These are not conservative engineering margins. They are set by intermetallic phase diagrams. Iron in aluminum forms Al₃Fe in slowly cooled melts, moving to Al₆Fe and metastable Al_mFe at higher cooling rates, and when silicon is also present above 0.5% the plate-like β-Al₅FeSi nucleates preferentially. β is the problem. It acts as a stress concentrator during deep drawing and stretch forming. A can end that cracks during tab riveting, an automotive fender that orange-peels during stamping: trace it back and you find β needles whose size and distribution are set by Fe and Si in the melt.

The Fe-Si interaction is where it gets genuinely complicated and where most recycling discussions lose the plot. In the ternary Al-Fe-Si system, the solidification path determines which intermetallic phases nucleate first. Below 0.3% Si, α-Al₁₅(Fe,Mn)₃Si₂ forms preferentially. α has a compact Chinese-script morphology that does comparatively little mechanical damage. Above 0.5% Si, β-Al₅FeSi dominates, and no heat treatment converts β back to α post-solidification. This is irreversible. Mn additions promote α over β, which is why 3004 can body stock at 0.8-1.3% Mn tolerates more iron than 5182 at 0.2-0.5% Mn. BCAST at Brunel, SINTEF in Trondheim, and the Light Metals Research Centre in Auckland have worked on this for decades, and what the work shows, over and over, is that the crystallography constrains recycling decisions and the constraints are hard. The constraints are hard, phase-diagram hard, and no amount of process optimization softens them.

And iron cannot be refined out of molten aluminum at industrial throughput. Pechiney piloted fractional crystallization at Voreppe in the 1970s-80s. 99.99% purity achieved. At kilograms per hour. At energy costs that were multiples of primary smelting. Alcan tried at Jonquière. The Japanese Society for the Promotion of Science funded three-layer electrolytic refining through the 1990s-2000s and the conclusion was the same: the metallurgy works, the economics do not, and the gap between what is achievable in a pilot cell and what is competitive at commodity scale has not closed in fifty years of trying. It may never close. Nobody in the aluminum refining research community seems optimistic on this point, though the work continues because the prize would be enormous if it ever worked commercially.

A modern car body uses 5xxx for inner panels, 6xxx for closures, 7xxx for structural reinforcement, plus cast in the engine block and transmission housing. At end of life the fragmentizer shreds everything together. Sink-float takes out the heaviest non-aluminum fractions, eddy current pulls the aluminum from the ferrous stream, and what comes out is a mixed-alloy bulk product heading for die-cast regardless of what it was in its previous life. LIBS sorting at conveyor speed can identify alloy grade on individual fragments and TOMRA and STEINERT sell these systems commercially for packaging waste, and pilot applications for auto scrap have been running at European recyclers since 2021, and the results are promising for packaging and frustrating for automotive because the pieces are smaller, dirtier, more geometrically varied, and the compositional gaps between 5xxx, 6xxx, and 7xxx are much narrower than can-body-versus-everything-else. One mis-sorted 7xxx fragment in a 6xxx melt pushes zinc above spec and downgrades the heat, which is the kind of consequence that makes a melt shop manager conservative about accepting sorted scrap regardless of what the LIBS vendor's brochure says.

Boliden's Rönnskär feeds shredded boards into a copper line at 1250°C. PGM and gold credits fund the operation. Circuit boards at 15-25% Cu by weight carry the smelting thermodynamics; lower-copper feeds dilute the matte and degrade performance. Boliden cherry-picks and so does every other e-waste smelter with emission controls. What gets rejected migrates to operations where labor is cheaper and environmental compliance is unenforced or nonexistent. The formal recycling system skims the high-grade fraction.

Anode slimes go through chlorination, silver electrolysis, gold refining, Se recovery, Te recovery. Progressively smaller value in progressively more specialized kit.

Seventeen elements out of seventy-plus recovered from a smartphone at a facility like this. Indium, gallium, germanium, tantalum, tungsten, rare earths go to slag. Umicore at Hoboken captures more via integrated pyro-hydro and Dowa and JX Nippon have sequential leach circuits for Ga and In from LCD panels but indium went from $800/kg to $150/kg between 2014 and 2016. Nobody builds a dedicated recovery circuit for a metal with that kind of price volatility unless they have a very specific view on where the price is going and a board willing to accept the risk of being wrong, and most boards are not.

DRDGOLD reprocessed 28.7 million tons of Witwatersrand gold tailings in FY2023 with no conventional mines, JSE and NYSE listed. Atalaya Mining: 56,085 tons Cu in concentrate from the Rio Tinto complex in 2023, LSE listed, Phoenician-era material. Freeport-McMoRan leaches old tailings at Morenci, Bagdad, Sierrita. These are publicly listed companies filing audited production reports.

Head grades at operating copper mines have dropped from 1.5%+ in the 1990s to below 0.6%. Mid-century tailings sit at 0.2-0.3% Cu. HPGR plus column flotation plus SX-EW: yesterday's waste crosses today's cutoff grade. Over 200 billion cumulative tons in global storage, classified as waste, unrecognizable to JORC or NI 43-101 because the deposit is anthropogenic, surveyed, directly sampable, already ground. No Measured/Indicated/Inferred category fits a tailings dam containing 800 million tons at 0.25% Cu where grade comes from assay and geometry comes from as-built surveys. Chile and South Africa have working groups. The regulatory architecture does not exist. Getting there requires amending mining law, environmental law, and financial reporting standards simultaneously, and the three stakeholder communities involved have different institutional cultures, different timelines, and different ideas about what "resource" means. A single large tailings facility might carry several billion dollars of contained metal at current prices as a closure liability on the balance sheet, and the difference between closure liability and contingent asset is the difference between sealing the impoundment and walking away versus maintaining access and eventually reprocessing.

Hideo Nanjyo coined "urban mine" at Tohoku in 1988. NIMS quantified Japan's above-ground metal stock: 6,800 tons Au, 60,000 tons Ag in electronics and industrial products, indium in LCDs rivaling geological reserves. Japan imports all significant primary metals, so the work was strategic and collection infrastructure and smelter investment followed the data. The US has not inventoried its stock. Hundreds of millions of vehicles, millions of buildings, a continental electrical grid, and nobody has systematically counted the metals or tracked the rate at which they enter end of life. Estimates exist. Their error bars span a factor of two depending on methodology.

Between 15 and 30% of the DRC's cobalt is artisanal, much of it from Gécamines tailings near Kolwezi and Likasi where decades of copper-focused processing left cobalt behind. Hand tools, no ground support, collapses. Buying houses, child labor documented by Amnesty International, heavy metal contamination in residential water measured by Lubumbashi toxicologists, BGR supply chain mapping. Blockchain provenance pilots have had limited uptake because the premium for "clean" cobalt does not cover the cost of keeping streams segregated through the refining chain. Material from dozens of sources gets blended at the aggregation point and traceability ends there. Cobalt from a tailings pit ends up in cathodes in Ningde or Komárom.

The economics of secondary recovery are more attractive wherever labor costs two dollars a day and environmental compliance costs nothing. Agbogbloshie, Seelampur, Guiyu. High recovery rates driven by desperation, achieved under conditions documented by human rights organizations whose reports the circular economy literature does not cite. Material flow analysis assigns a number to "tons recovered" and moves on.

Global copper in service: north of 500 million tons. The energy transition adds 6-8 million tons/year demand on top of existing 22 million in mine output, with secondary contributing 4.5 million. Copper installed in a building in 2005 is locked until demolition, maybe 2050, maybe 2070. A water main from 1990 might stay underground until 2090. Service lives for infrastructure run in generations. Mine production expands during the energy transition because the in-use stock releases on a timescale of decades and the demand surge is now, and no recycling rate changes this because the constraint is temporal, not technological.

Copper accumulates in steel scrap from embedded motors and wiring. Molten steel cannot shed it. 0.3% average in obsolete scrap, 0.10% max for automotive flat-rolled or hot shortness during rolling. Dilution with virgin DRI or pig iron is the only practical remedy. Daigo at the University of Tokyo modeled the accumulation trajectory from 2005 and showed that under 90% scrap recycling the dilution ratio for flat products climbs steeply. The European steel decarbonization push (BF-to-EAF) increases scrap fraction, which increases copper concentration. Every ton of scrap-based EAF steel for auto sheet needs near-zero-copper DRI, and DRI at that scale depends on natural gas or hydrogen supply chains with their own infrastructure and policy constraints. Decarbonization strategy and tramp element management are entangled in ways the policy discussion has not engaged with.

Zinc galvanizing disperses 40% of the coating as ZnO in roadside soil over 30 years. Phosphorus in fertilizer: 15% recovered via struvite at wastewater plants. Over half of zinc and nearly all barium enter dissipative applications annually. Recycling loses 4% Cu per cycle, 2-8% Al, 5-12% steel. After ten cycles at 80% collection, 60g remains from an original kilogram. Primary extraction floor somewhere near 65% of annual Cu demand.

Escondida: desalinated seawater pumped 170 km uphill, 3,100m elevation gain, $3.4 billion capital, $6+/m³ delivered. Collahuasi: 90% process water recycling. Cerro Verde: discharge quality above incoming, Arequipa municipality required it. British Columbia and Finland: $0.10-0.50/m³, lower recycling rates, same technology available. SART at Telfer, Yanacocha, Maricunga: adoption tracks cyanide cost and solution copper grade, sustainability commitments play no observable role. Circularity scales with the cost of linearity and the mechanism is the plant-level P&L.

A lithium-ion cell that snaps apart at module level can still contain NMC bonded to Al foil with PVDF. Separating active material requires NMP or 500-600°C calcination. Then acid dissolution (4M H₂SO₄, H₂O₂ reductant, 60-80°C, 2-4 hours) and sequential SX via D2EHPA or Cyanex 272 for Ni/Mn/Co separation, which is standard laterite hydromet. The chemistry is not the problem. The pre-treatment is. A PVDF-bonded coating on aluminum foil embedded in a pouch cell with electrolyte residue and separator film needs extensive preparation before the leach circuit can accept it, and pre-treatment cost dominates the recycling economics for lithium-ion batteries, and the reason it dominates is that the cell was designed by electrochemists optimizing for energy density and cycle life at beginning of life and nobody in the design process asked how this cathode would behave in a leaching tank fifteen years later.

Water-soluble binders could replace PVDF. Direct recycling could re-lithiate degraded cathode without dissolution. These require extractive metallurgy input: mineral liberation, co-precipitation in pH-Eh space, phase behavior in multi-component leach systems. The people with that knowledge spent careers on primary ore processing at concentrator design firms and mining R&D centers and university departments of mining and metallurgical engineering. The knowledge transfers directly to end-of-life recovery. The transfer does not happen because the professional distance between a concentrator design team and a battery R&D lab is enormous, measured in conferences that have no overlapping attendees and journals that have no shared citation pools, and nobody has built a bridge across it because the two communities do not know the bridge is missing. The metallurgist does not read the battery journals. The electrochemist does not read Minerals Engineering or Hydrometallurgy. The product design conversation and the end-of-life recovery conversation exist in parallel, within meters of each other at some universities and within different buildings at some companies, and they do not intersect.

If extractive metallurgy were consulted during cell chemistry selection, binder systems and cathode architectures would look different and recovery costs would drop. This particular intervention costs less than commodity exchange reform, mining code amendment, product passport regulation, tailings reclassification, or financial reporting standards revision, each of which requires its own legislative calendar and stakeholder negotiation and none of which has been coordinated across jurisdictions and all of which have been discussed in working groups for the better part of two decades without producing binding outcomes in any major mining jurisdiction.