How is silver refined: the trade you make without noticing

Silver is refined by converting a silver-bearing feedstock into metal of a published fineness and casting that metal as a market bar. Mine concentrate can carry 0.06 to 3.1 kilograms of silver per tonne in copper concentrate, according to U.S. Geological Survey Open-File Report 2004-1251, and about 0.8 to 1 kilogram per tonne in typical lead concentrate; Pan American Silver’s La Colorada Skarn technical report, filed with the U.S. Securities and Exchange Commission, puts mill feed at 58 grams of silver per tonne and the lead concentrate near 1,800 grams per tonne. Secondary lots start richer: sterling scrap is 925 parts per thousand silver under jewellery-alloy convention. Primary ore is concentrated and smelted so silver reports to lead bullion, copper anode slime, or gold-silver doré. Scrap is sorted, melted, and often sent toward electrorefining without a mill. Both routes usually finish in a silver-nitrate cell. Joachim and Adalbert Prior, writing for GDMB, set a working minimum of 80 weight-percent silver in the anode and report cathode metal at or above 999.9‰ in bulk-anode practice. The London Bullion Market Association Good Delivery Rules require a minimum fineness of 999.0 parts per thousand and, for bars produced from 1 January 2025, a nominal mass of 1,000 troy ounces with a ±10 percent tolerance. ASTM International B413 Grade 99.90 is that 999.0 commercial-bar floor; Grade 99.99 is 999.9. Mill recovery on the La Colorada Skarn flowsheet is estimated at 89.7 percent of contained silver. Classic Moebius plant practice, recorded by M. A. Mosher for the American Institute of Mining and Metallurgical Engineers, ran at a 98 percent current efficiency. Feedstock decides whether pyrometallurgy, hydrometallurgy, or electrorefining does the work. A melt homogenizes a lot. An assay certifies the grade.

Why ore refining and scrap recycling are not one workflow

Search pages treat how silver is refined as a single furnace story. The feed decides the chemistry before anyone lights anything.

Primary silver almost never walks into a refinery as a high-grade rock. The U.S. Geological Survey Mineral Commodity Summaries 2026, prepared by Anne M. Hartingh, notes that silver is recovered mainly as a byproduct of lead-zinc, copper, and gold mines. World mine output in 2025 is estimated at 26,000 metric tons. Hartingh’s U.S. use split for 2025 is electrical and electronics 25 percent, other industrial uses and photography 19, physical bars 18, photovoltaics 15, coins and medals 14, jewelry and silverware 6, brazing and solder 3. Pan American’s La Colorada Skarn design recovers about 89.7 percent of the silver in the ore, most of it into the lead concentrate, with a smaller share into zinc concentrate. Ten percent of the silver in that rock never reaches a bar. That is a mill number, not a cell number, and it is the first trade primary producers make without putting it in the headline.

Secondary silver arrives already concentrated by someone else’s fabrication. The Silver Institute’s World Silver Survey 2026, researched by Metals Focus, puts 2025 recycling at 197.6 million ounces against 846.6 million ounces of mine production. In the United States, USGS counted about 1,000 metric tons recovered from new and old scrap in 2025, next to 1,100 tons of primary refinery output, from 24 refiners reporting 2,100 tons of commercial-grade metal in total. Scrap can be sterling jewellery, ethylene-oxide catalyst, photographic residue, or e-scrap. Sterling is 925‰. Catalyst and circuit boards are not. The refiner’s first job on scrap is to know which lot is which.

I used to tell assistants, around 2018, to crank a flat iron and take one slow pass. Peak heat, I said, beats a stack of milder ones. I stopped after a season of bagged wash-day strands from the company tour showed three moderate passes at a lower plate setting leaving the same mid-shaft splits as the single hot stroke. I log cumulative exposure now. Primary silver pays a mill-recovery tax before it ever sees a cupel. Scrap skips that tax and pays a sorting tax instead. Collapsing both into one “refining by fire” cartoon hides which bill you are settling.

| Feed | Typical silver in the feed | First industrial move | How silver is concentrated | Last step before the bar | | --- | --- | --- | --- | --- | | Lead-zinc concentrate | 0.8–1 kg/t typical (USGS OFR 2004-1251); 1,800 g/t at La Colorada Skarn | Crush, grind, float, smelt to lead bullion | Parkes zinc crust, then cupellation | Electrorefining of doré anodes | | Copper anode slime | Aurubis, at an LBMA Assaying & Refining seminar, reported Ag-doré near 92–98% precious metals before the silver anode | Collect slime, then roast, smelt, or leach | Top-blown rotary converter or cupellation to doré | Electrorefining | | Fabricated scrap | 925‰ sterling; mixed industrial lots vary widely | Sort, melt, sample | Often none; nitric or other leach when base metals dominate | Electrorefining, or precipitation then melt |

Hydrometallurgy sits on both sides of that table when a solvent beats a furnace: cyanide or thiosulfate on gold-silver ores, nitric acid on many secondary lots. Pyrometallurgy still owns the lead-bullion and slime-to-doré stretch. Electrorefining is the usual last cut for 999 and 999.9 metal.

The sequence, when a plant is trying to make a Good Delivery bar from mixed feed, looks like this:

1. Characterize and assay the lot, so ore concentrate, slime, doré, and scrap are not charged as one bath.
2. Concentrate and smelt primary material, or melt and sample scrap.
3. Enrich silver with a Parkes zinc crust, a cupel, or a leach.
4. Cast anodes that meet the cell’s silver minimum.
5. Electrorefine in a Moebius or Balbach-Thum cell to cathode crystals.
6. Melt the crystals, take a sample, assay it, and cast the bar.

Skip step 1 and the later heat is wasted on the wrong chemistry.

What fineness and bar mass industrial silver has to hit

Target fineness is a published number.

The LBMA Good Delivery Rules, in the Technical Specifications updated for 2025, set the wholesale floor at 999.0 parts per thousand silver. Bars produced from 1 January 2025 must weigh 1,000 troy ounces with a ±10 percent tolerance, about 900 to 1,100 troy ounces, roughly 28 to 34 kilograms. Older bars in the former 750-to-1,100-ounce range remain acceptable until vault stocks of them run down. Gross weight is recorded in troy ounces to the nearest 0.10, rounded down. Marks include the refiner’s stamp, an assay mark, fineness, a serial number of at most 11 characters, and the month and year of manufacture as MMYY.

ASTM B413, Standard Specification for Refined Silver, reapproved 2021, splits commercial metal into three grades. Grade 99.90 (UNS P07020) is 999.0 fineness, “commercial bar” or bullion. Grade 99.95 is 999.5. Grade 99.99 (UNS P07010) is 999.9. LBMA will take the first of those as Good Delivery. Electrical, photographic, and some investment products want the third.

A bar that meets the mass window and fails the assay is still not good delivery. I prefer the certificate to the gloss. I learned that on tour, on my daughter’s saved wash-day strands: the kit in my bag is a logger.

How fire still sits in the flowsheet

Silver refining by fire is oxidation under an air blast, usually with lead as the collector and the fuel for the dross.

Cupellation is the old core. Argentiferous lead is melted on a porous dish of bone ash or magnesia. An air blast oxidizes lead to litharge (lead oxide). The cupel soaks the litharge up. Base metals go with it. Silver and gold stay as a bright button. Process temperatures sit near 900 to 1,000 °C, above the litharge formation range and around the silver freeze. The ITS-90 fixed point for the freezing of pure silver is 961.78 °C, the value NIST assigns to SRM 1746. That freeze is a temperature standard. Purity is a separate measurement.

The Parkes process, patented by Alexander Parkes in 1850, is the industrial way to pull silver out of lead bullion before the cupel. Zinc is stirred into molten lead. Zinc and lead do not mix as one liquid. Silver reports to the zinc, which freezes as a crust, is skimmed, retorted to drive zinc off, and cupelled. Port Pirie kettle practice, as recorded in the de-silverizing plant notes, required the lead after the second zinc addition to hold no more than four pennyweights of silver per ton. That residual is the fire-side recovery target for that kettle, separate from the 89.7 percent mill figure above.

I got this wrong in my own shop before I had language for feedstock. A spare bun for a two-show day was a kanekalon braid. I ran a hot brush across it at a plate setting I had already cleared for keratin. The fiber slumped. The dancer lost the spare. I lost the habit of treating “looks like hair” as a process specification. Sterling scrap and copper-slime doré both look like silver. Fire is the right tool for the lead-rich one. It is often the wrong first tool for the plated, soldered, or polymer-backed one.

Fire still belongs in modern plants. Aurubis described slime treatment through a top-blown rotary converter and cupellation to silver anodes. Lead refineries still zinc the kettles. The trade people miss is treating that fire as the finish. For Good Delivery metal, fire is usually the enrichment step. The cell is the cut.

What electrorefining does to the anode and the cathode

Electrorefining dissolves a silver anode in silver nitrate and grows crystals on a cathode.

Two cell geometries dominate. The Moebius cell hangs anodes and cathodes vertically. The Balbach-Thum cell lays the anode down. Most plants prefer Moebius for floor space. Prior and Prior’s GDMB paper on bulk-anode HSSE cells, which follow the Moebius principle, is the cleanest published set of working limits I have: minimum anode silver 80 weight-percent, gold up to 20 percent, base metals up to 12 percent, cathode product at or above 999.9‰ silver. Their example doré feeds run Ag–Au–Cu at 82–10–8, 91–1–8, 81–15–4, and 99–0–1. Japanese plant practice described in patent JP4787951B2 casts anodes at 97 to 99 weight-percent silver and reports cathode metal at 99.99 percent or more.

Mosher’s AIME account of Moebius operation put current efficiency at 98 percent, five to ten points above Thum in that comparison, with anode scrap around 15 percent of the anode mass returned for recasting. Prior gives anode-return butts at about 7 to 15 percent for conventional cells and none for the bulk-anode design. Those returns are circulating inventory if the butts go back to the anode furnace. They still add a circulating load.

I cannot vouch for a specific anode campaign. I have never pulled a bag of gold mud off a Moebius hanger or scraped a cathode. I can vouch for what repeated heat does to remaining strength in a fiber, because I have labeled the bags. The analogy I trust is narrower than a plant tour: an anode below the silver minimum poisons the electrolyte the way a tool above the fiber’s limit poisons the style. You do not fix that with a hotter pass.

Cathode crystals are washed, melted, sampled, and cast. That melt is housekeeping. The grade is the assay of the sample, matched to ASTM B413 or to LBMA marks.

What an assay records after the melt

Melting makes a lot uniform enough to sample. It does not measure silver.

ISO 11427:2024 is the volumetric reference for silver in homogeneous metal from 100 to 999.0‰: dissolve in dilute nitric acid, titrate with potassium bromide, read the equivalence potentiometrically. Above 999.0‰, ISO 15096:2020 is the recommended method. Impurities are measured by ICP-OES and subtracted from 1,000‰; Annex B covers 999.9‰ and above. ISO 11427 itself points at ISO 15096 for the high-fineness difference method. National fire-assay methods for silver still exist. The jewellery-and-bullion referee track for fine silver is titration, then spectroscopy by difference.

That is how you know silver is refined enough to mark. The cupel can brighten. The freeze can sit at 961.78 °C. The bar can take a stamp. The silver assay after refining is the number the method returns, written on the bar and on the weight list.

A plant chemist’s strongest argument against me is that I am a heat-tool tester borrowing a flowsheet. They are right that current density, silver ion concentration, and copper in the electrolyte govern a Moebius cathode, and that my plate-temperature log does not. Grant that. The missed trade is still earlier in the search: people ask how silver is refined as if fire, leach, and electrolysis were one knob, then treat a melted surface as a finished assay. I will keep arguing for the log. I will not pretend I have run the cell.

How ancient and biblical refiners used the same fire chemistry

Ancient silver refining was cupellation. Smiths smelted argentiferous galena to lead-silver bullion, then blew air across the melt on a porous hearth. Litharge and dross left. Gold stayed. The International Standard Bible Encyclopedia describes that sequence on silver-rich lead sulfide, with litharge driven off as the “silver dross” later used to glaze pottery. Process heat in modern reconstructions sits near 900 to 1,000 °C.

Biblical Hebrew uses that shop. Proverbs 17:3 names the cupel (matzreph) for silver. Jeremiah 6:29–30 names the bellows, the lead consumed, and the failed refine that leaves “reject silver.” Psalm 12:6 speaks of silver purified seven times in a clay furnace, which metallurgists read as repeated melts in a porous crucible, not a soil kiln. The ancient silver refining process is the fire assay stripped of a certificate. They knew the metal was done when the button stayed bright and the litharge stopped forming.

Questions people type after “how is silver refined”

How was silver refined in biblical times?

Biblical refiners smelted silver-rich lead ore, then cupelled the bullion on a porous bone-ash dish. A bellows blast turned lead into litharge the cupel absorbed, leaving silver. Jeremiah 6:29 names the bellows; Proverbs 17:3 names the cupel. That is still fire assay.

How did ancient people refine silver?

Ancient smiths roasted silver-bearing galena, smelted it to lead-silver bullion, and cupelled the melt near 900 to 1,000 °C. Litharge and dross left; gold stayed with the silver. Anatolian cupels from the third millennium B.C. show the method predates the biblical descriptions.

How do you know when silver is done refining?

You know when an assay meets the grade. ISO 11427:2024 titrates silver from 100 to 999.0‰. Above that, ISO 15096:2020 subtracts impurities by ICP-OES. A bright cupel surface is a process cue. LBMA 999.0 and ASTM B413 grades are certificate numbers.

How is silver refined by fire?

Fire refining oxidizes base metals. In cupellation, an air blast over molten lead-silver near 900 to 1,000 °C turns lead into litharge the cupel soaks up. In the Parkes process, zinc pulls silver from molten lead into a crust that is retorted and cupelled. Electrolysis usually finishes the bar.

What does an assay establish that melting cannot?

Melting mixes a lot so a chip represents the bar. It does not measure silver. An assay reports fineness against a method: ISO 11427 titration to 999.0‰, or ISO 15096 difference ICP-OES at 999‰ and above. LBMA and ASTM B413 grades are those assay numbers.