Gold Extraction Methods and
Processing Techniques Explained

The way gold occurs in ore dictates the entire extraction approach.

Free gold is the easiest to deal with. Native gold sitting in fractures or gaps between mineral grains, not wrapped up in anything else, recoverable by gravity or even amalgamation. Sulfide-locked gold is an entirely different order of problem, sub-micron gold particles locked inside the crystal lattice of pyrite and arsenopyrite, cyanide solution cannot touch them. The third type is preg-robbing ore, where naturally occurring organic carbon in the rock adsorbs gold cyanide complexes back out of solution during leaching, meaning the gold dissolves first and then gets stolen back.

These three types frequently coexist. The oxide zone of a single mine might be predominantly free gold, dig deeper into the primary sulfide zone and it becomes locked gold, cyanide recovery drops from the nineties straight down to the thirties or forties. This is called ore transition in the industry, and how well it is managed directly determines whether mid-life cash flow stays stable. Projects where the feasibility study glosses over this section almost inevitably run into trouble three to five years after commissioning.

The reliability of preg-robbing tests deserves separate discussion. Under laboratory conditions slurry density, agitation intensity, and carbon concentration are all standardized. In industrial leach tanks the residence time distribution is far from ideal plug flow, dead zones and short-circuiting mean the contact pattern between organic carbon and gold cyanide complexes is completely different. Ore that the lab calls mildly preg-robbing can show moderate or even severe preg-robbing behavior at industrial scale. Some ores also have strong particle-size dependence in their preg-robbing behavior, organic carbon is sparse in the coarse fractions and highly concentrated in the ultrafine fractions. If the grind size in the metallurgical test doesn’t match the plant design grind size, the test conclusions become useless.

Geometallurgy ties geological models to metallurgical response data, predicting ore characteristics block by block, period by period. Globally, fewer than twenty percent of projects invest enough geometallurgy work at the feasibility stage. The reason is straightforward: the discipline requires individual metallurgical tests on large numbers of drill core intervals rather than composite tests, it is expensive and time-consuming, and the conclusions it tends to produce are along the lines of “the ore is more complex than you thought,” which is not helpful for financing. No project owner wants to pay money to prove their ore is difficult.

A well-designed physical separation front end can cut the scale of downstream chemical processing by more than half.

Gravity separation exploits the nearly sevenfold density difference between gold and gangue. Knelson and Falcon centrifugal concentrators are currently the main equipment for coarse gold recovery. The key is not equipment selection but installation position, it must be embedded inside the grinding circuit for mid-stream recovery. Gold has the highest malleability of any metal, particles in a ball mill get hammered and stretched into flakes, and those flakes settle like fine gangue rather than heavy minerals, gravity cannot recover them. Every additional grinding cycle increases the probability of free gold turning into unrecoverable smeared gold.

Flotation in gold processing is a concentration tool. Head grade might be 3 g/t, flotation concentrate grade can reach 30 g/t or higher, eighty-five percent of the gangue is rejected upfront, downstream oxidative pretreatment only has to handle fifteen percent of the original ore mass. The regrind fineness of flotation concentrate has a nonlinear inflection point in its effect on subsequent oxidation. At P80 of 10 to 15 microns decomposition rates rise noticeably. Below about 8 microns the slurry rheology changes qualitatively, yield stress shoots up, agitation and pumping power consumption spikes, and in high-viscosity slurry bubble dispersion becomes difficult, directly throttling mass transfer efficiency in the oxidation reactor. The location of this inflection point varies widely between ores and can only be determined through testwork.

The MacArthur-Forrest process came into existence in 1887. Since then over eighty percent of global gold production has gone through cyanidation. The Elsner equation:

4Au + 8NaCN + O₂ + 2H₂O → 4Na[Au(CN)₂] + 4NaOH

The chemistry of cyanidation is simple. The parameters that make it work well or poorly at industrial scale have been in a state of neglect for a long time. This section goes into detail because cyanidation is the absolute backbone of gold metallurgy and the operational details here carry far more weight on the overall flowsheet than any other single stage.

Dissolved oxygen. Oxygen is a reactant in the Elsner equation, everyone knows the theory, the practice is a different matter. Water at standard atmospheric pressure saturates at about 8 ppm DO. In industrial slurries where solids displace volume and reducing minerals continuously consume oxygen, DO typically hangs around 2 to 3 ppm. Leach kinetics under these conditions are severely limited by oxygen mass transfer. DO needs to be above 8 ppm, difficult-to-leach ores need 12 to 15 ppm, requiring pure oxygen injection or Venturi sparging systems. Many plants have DO monitors installed on the leach tanks, the operations team never looks at the numbers, never manages DO as a process parameter. In operations audits this is the most frequently identified improvement item, and one of the simplest to fix: increase blower capacity or change the sparging configuration, and recovery goes up 1 to 3 percentage points. For a plant processing 2 million tonnes per year at 2 g/t head grade, one percentage point is about 40 kilograms of gold per year. The return on investment from a single blower would look absurd in any industry.

The gap between locked cycle test results and industrial performance. The locked cycle test returns intermediate products from one cycle to the next to approximate steady state. Its steady state is built on perfect mass balance, precise time control, and zero equipment failure. An industrial plant faces hourly ore grade fluctuations, unplanned shutdowns, reagent supply delays, and shift changes. That attractive recovery number in the feasibility report takes a 5 to 8 percentage point haircut at industrial scale. Experienced metallurgical engineers and bank independent technical advisors both apply this mental discount when reading feasibility reports, and everybody knows everybody else is doing it.

Heap leaching suits oxide ore at 0.5 to 1.5 g/t, ore is crushed and stacked on HDPE liner, cyanide solution drip-irrigated from the top. Recovery 60 to 75 percent. Per-tonne ore processing cost roughly a third of agitated leaching.

What actually happens inside a heap is still a black box. Tracer tests reveal some flow paths. Internal local temperature, pH gradients, cyanide concentration distribution, and oxygen penetration depth are essentially unmeasurable at industrial scale. Channeling is the most obvious problem, fines form low-permeability zones inside the heap, solution follows preferential pathways, large areas of ore never get effectively contacted. Agglomeration with cement or lime binder is the standard countermeasure.

High-clay ore heaps are a different story altogether. Permeability looks fine right after crushing and agglomeration. As soaking time extends, clay minerals swell on contact with water, permeability decays to a tenth of its initial value within weeks, the entire heap turns into a giant mud cake. Irreversible. All the money spent on crushing, hauling, and agglomeration is sunk. Smectite and kaolinite respond to water completely differently, distinguishing the type and proportion of clay minerals in the ore is the single step in heap leach feasibility assessment where cutting corners has the worst consequences, and also the step most likely to be dismissed with “it’s all clay anyway.”

Long-running heaps also have an operational nuisance: microbial communities form biofilms on ore particle surfaces and in pore channels, progressively blocking flow paths. Heap leach pads are chronically wet, near-neutral to mildly alkaline pH, low in organic nutrients, superficially unfavorable for microbial growth. The micro-environment inside a heap is far more complex than macro-scale measurements suggest, localized anaerobic pockets and trace nutrients released by mineral dissolution are enough to sustain slow expansion of specific microbial populations. Some operations periodically dose the leach solution with oxidizing biocides, results vary with heap conditions.

Thiosulfate leaching has a natural advantage on preg-robbing carbonaceous ore because organic carbon does not adsorb gold thiosulfate complexes, eliminating the preg-robbing mechanism at its root. The fatal weakness is reagent self-decomposition: thiosulfate ions in solution continuously oxidize into polythionates and tetrathionate, reagent consumption is enormous and recirculation is difficult. Barrick’s industrial application at Goldstrike is widely cited. The internal industry assessment is considerably more nuanced than public reporting suggests: that project’s economics rest on a very specific ore type and production scale. On any deposit where cyanidation can achieve even a passing recovery rate, thiosulfate cannot compete on economics.

Glycine leaching has had high academic interest in recent years, low environmental toxicity, potential for simultaneous copper and gold leaching from Cu-Au ores, slow leach kinetics, limited gold dissolution capacity under alkaline conditions. Every few years a new “green lixiviant” generates excitement at academic conferences. Brought back to the logic of engineering economics, the combined advantages of cyanidation (speed, reagent cost, mature recovery circuits, accumulated scale-up experience) keep alternative processes confined to the specific ore types where cyanide genuinely cannot do the job.

When gold is tightly locked inside sulfide minerals, pretreatment has one purpose: destroy the crystal lattice and expose the gold.

Roasting treats sulfide concentrate at 600 to 700°C in an oxidizing atmosphere. Fluidized bed design is the current standard, coupled with a downstream sulfuric acid plant to capture SO₂. On high-arsenopyrite ore, temperature control is the central difficulty. Above a critical temperature, arsenic combines with iron to form iron arsenate glass, which re-encapsulates gold particles and is harder to break open than the original sulfide. This is over-roasting. Inside a fluidized bed there is a temperature difference between the bubble phase and the emulsion phase, local hotspots can exceed the average temperature by 50 to 80°C, so localized over-roasting keeps happening even when average temperature is under control. The response for high-arsenopyrite ores is two-stage roasting: stage one at lower temperature for selective arsenic removal, stage two at higher temperature to oxidize remaining pyritic sulfur. The temperature windows and residence time ratios for both stages have to be calibrated individually for each ore, and adjusted dynamically when ore composition fluctuates.

POX runs in a sealed autoclave, 190 to 225°C, total pressure around 2000 to 3200 kPa. Sulfides are completely oxidized to sulfates and iron oxides, no SO₂ emissions. Sulfide decomposition above 95 percent on most refractory gold ores. Titanium-lined autoclaves, a single line costs hundreds of millions of dollars.

BIOX uses Acidithiobacillus ferrooxidans and related acidophilic bacteria to decompose sulfides at ambient temperature and pressure, 40 to 45°C, pH 1.2 to 1.8. Capital and operating costs far below POX, suitable for medium and small scale projects. Residence time 4 to 6 days, significant footprint requirement.

More space on BIOX here, because this process has a large number of operational issues that cannot be found in textbooks and are only known to people who have run one.

Temperature control is the highest-probability failure point. Sulfide biooxidation is exothermic. When cooling capacity in a large reactor is insufficient or cooling equipment fails, slurry temperature can climb from 42°C to over 50°C within hours. Mesophilic bacteria start dying in large numbers above 48°C. After a population crash, recovery takes weeks to months, the plant is semi-paralyzed during that period. Whether the cooling system has enough redundancy at the design stage directly determines operational stability. Quite a few plants experienced their first bacterial crash during the first summer after commissioning, because the cooling calculations at the design stage used annual average temperature rather than peak temperature during the hottest month.

Foam. Microbial metabolic products include proteinaceous surface-active substances. Under continuous aeration these form a stable foam layer on the reactor surface. Foam overflow causes material loss and environmental contamination. A thick foam blanket on the surface impedes CO₂ release and O₂ transfer. Mechanical defoamers or chemical antifoam agents are used for control. Antifoam type and dosage require careful calibration because some organic antifoams are toxic to the bacteria at high concentrations. This kind of thing is invisible in papers and feasibility reports. In daily operations it never goes away.

BIOX reactor inoculation and startup also deserve mention. Bacterial culture for a new BIOX system typically starts from small seed reactors and is progressively scaled up to full-size reactors, this process normally takes three to six months. Inoculum source selection, culture medium formulation, arsenic concentration gradient control in the feed concentrate, any misstep at any point can delay startup. Startup delay means the plant is sitting there with equipment in place, bacteria not yet established, unable to produce, and every extra month is a large sum of fixed costs running idle. Some projects start the seed reactor cultivation in parallel with construction so that the bacteria preparation timeline runs alongside the civil works timeline, avoiding equipment waiting for bacteria. This arrangement sounds simple. In project management practice few projects pull it off because bacterial cultivation falls under the operations department’s scope of work, and during the construction phase the operations team is often not yet on site, or is on site without budget.

Ultrafine grinding uses IsaMill or SMD to grind concentrate to P80 below 10 microns. No chemical reagent consumption, no gas emissions. When gold is dispersed in the sulfide crystal lattice as solid solution (lattice-bound gold), ultrafine grinding is ineffective regardless of fineness, chemical oxidation is required. The ideal application is ore where gold exists as discrete micro-inclusions within sulfide particles. Some projects combine ultrafine grinding with BIOX, grinding first to partially open the sulfide encapsulation, then BIOX to oxidize the remaining sulfide matrix.

This is the part of the entire gold metallurgical flowsheet most directly connected to money and most sensitive from an operational management standpoint, so it gets detailed treatment.

Gold after cyanide leaching is in solution as Au(CN)₂⁻, concentration typically 1 to 5 mg/L.

Activated carbon has exceptionally strong adsorption affinity for gold cyanide complexes. CIL runs leaching and carbon adsorption simultaneously in the same tank train. CIP completes leaching first then adsorbs. CIC passes clarified solution through fixed-bed carbon columns. CIL has particular value on carbonaceous ore because activated carbon’s adsorption kinetics and capacity far exceed those of natural organic carbon, competitively pulling gold away from the native carbon.

AARL elution. Loaded carbon is first acid-washed with 3% dilute hydrochloric acid to dissolve calcium carbonate and iron oxide scale deposited in the pores, restoring pore accessibility, then hot caustic elution follows. Eluate gold concentration typically 100 to 300 mg/L, electrowinning deposits gold sludge on stainless steel wool cathodes. Gold sludge is acid-washed, dried, then smelted in an induction furnace with borax and soda ash as flux, poured into Doré Bars, gold content 70 to 90 percent.

Every plant has a metal balance. Gold entering the plant should equal gold produced plus gold lost to tailings. Almost every plant has an unreconciled difference of 1 to 5 percent. The sources are scattered across the entire flowsheet: gold carried in fine carbon particles, gold unrecovered from smelting slag, deposits on pipe walls and equipment internals, slurry sampling errors propagating through calculations, and the human factor. Each individual loss point is negligible, the aggregate is an uncomfortable number. Major mining companies invest in gold room security and auditing at a level comparable to bank vaults.

Gold lock-up in the plant piping system. Over years of operation gold deposits in various forms at pipe elbows, valves, pump casings, and dead spots at the bottom of tanks. Gradual, invisible, not reflected in daily metal accounting. When a plant does a major shutdown or removes old piping, the amount of deposited gold recovered from the piping system is sometimes surprisingly large. Some plants include a dedicated “pipe flushing and residual gold recovery” step in their shutdown plans, built into the annual gold production forecast. A plant that has been running for over ten years can have tens of kilograms of gold accumulated in its piping system. How this hidden inventory is treated in the accounts and to which audit period it is assigned is a standalone topic in mining company internal financial management.

When a plant restarts after a maintenance shutdown, recovery in the first few days is usually abnormally high. The reason is that gold-bearing solution and deposits that accumulated in piping and equipment during the shutdown get flushed into the recovery circuit, creating a brief pulse. Experienced operations teams know this and factor the pulse effect into monthly gold production reporting to avoid misjudging steady-state recovery.

One more operational detail related to unaccounted loss that rarely gets written up: gold enrichment on smelting room floors and in floor drains. The gold room and smelting area floors are in long-term contact with gold-bearing materials. Wash-down and sweeping water contains non-negligible gold concentrations. Design standards require smelting area floors to have impermeable coatings with all floor drainage collected in a dedicated sump, sump sediments periodically recovered and processed. Plants that do this well recover a few kilograms of gold per year from this source alone. Plants that do it carelessly lose that gold down the drain into tailings.

The Merrill-Crowe process requires highly clarified pregnant solution with DO below 1 ppm, zinc dust is added for cementation. Recovery exceeds 99 percent, precipitate grade higher than carbon adsorption. Most widely used in heap leach operations because heap leach produces clarified solution rather than slurry, naturally meeting the clarity requirement. Silver recovery efficiency is significantly better than carbon adsorption, which provides an additional economic argument for choosing the process on ore with a high silver-to-gold ratio.

Doré Bars go to the refinery. Miller chlorination passes chlorine gas through molten crude gold, base metals and silver are preferentially chlorinated and skimmed off as dross, gold purity reaches 99.5 percent. The Wohlwill process uses crude gold as anode, pure gold sheet as cathode, electrodeposition in chloroauric acid electrolyte, producing 99.99 percent purity refined gold meeting the LBMA Good Delivery standard.

Cyanide detoxification typically uses the SO₂/air process (INCO process), oxidizing free cyanide and WAD Cyanide to low-toxicity cyanate. Tailings WAD cyanide concentration must be brought below 50 mg/L. The regulatory divergence between WAD and Total Cyanide is a long-running debate. Most mining regulations use WAD as the discharge control metric. Iron cyanide complexes are not captured by WAD testing but can slowly decompose to release free cyanide under strong ultraviolet radiation, and at high-altitude low-latitude mine sites this release pathway is not negligible. Adopting total cyanide as the discharge standard would multiply detoxification reagent costs several times over.

AMD. Sulfide-bearing tailings exposed to air and water generate sulfuric acid, pH drops to 2 to 3, leaching arsenic, lead, cadmium, mercury. Once started, AMD can persist for centuries without human intervention.

Dry stack tailings. Tailings slurry is filtered to 15 to 20 percent moisture content then placed and compacted in lifts, virtually eliminating dam failure risk. Filtration equipment capital and operating costs are 30 to 50 percent higher than conventional wet tailings storage. Under tightening ESG standards, new projects seeking approval for conventional wet tailings facilities face increasing difficulty.

Process water balance is a survival prerequisite in water-stressed mining regions. Plants in the Australian interior, the West African Sahel, and the Andean altiplano are designed with process water recirculation rates above 95 percent. Dissolved salt accumulation in recirculating water progressively shifts the slurry chemistry over the years. Sulfate and thiocyanate buildup deserves particular attention: high thiocyanate concentration suppresses gold leach kinetics and competes with gold cyanide complexes for adsorption sites on activated carbon. A plant that has been running for years with steadily climbing recirculating water salt levels can be 2 to 4 percentage points below its commissioning-era cyanide circuit performance. The operations reports show recovery slowly declining, process parameters check out normal, reagent dosing unchanged, equipment condition unchanged. Three months of troubleshooting and the answer finally surfaces: thiocyanate concentration in the recirculating water has doubled compared to three years ago. This kind of slow water quality deterioration causing gradual performance decay cannot be detected or attributed without a dedicated water quality monitoring and trend analysis program.