Gold Ore Processing
and Mineral Beneficiation

Laboratory metallurgical test results almost invariably overestimate plant performance. The particle size distribution from laboratory grinding is more uniform than industrial hydrocyclone overflow. The aeration and agitation conditions in a laboratory flotation cell are more ideal than in an industrial flotation machine. Laboratory cyanide leaching uses deionized water rather than industrial process water loaded with accumulated impurities. These differences stacked together mean the recovery figure in a feasibility study typically loses 2 to 5 percentage points by the time the plant is commissioned. The first thing an experienced metallurgical engineer does when reviewing a feasibility study is mentally apply a downward correction factor to the laboratory data, the magnitude of which depends on the scale of the laboratory tests, the representativeness of the samples, and the degree of alignment between the test protocol and the industrial flowsheet.

A significant proportion of ores labeled “refractory” are misdiagnosed. In some ores, low cyanidation recovery is not caused by gold being locked in sulfides. The dissolved oxygen in the slurry is being excessively consumed by reducing minerals, or copper, zinc, and other impurity ions accumulated in process water have formed stable complexes with cyanide, displacing effective cyanide concentration. These problems can be substantially improved by adjusting leaching conditions (intensifying aeration, pre-treating process water, or simply extending leaching time) without the need for expensive pre-oxidation. Once the “refractory” label is applied, subsequent process design tends to follow the pre-oxidation path without anyone going back to re-examine whether the fundamental diagnosis was correct. A POX system costing hundreds of millions of dollars built for an ore that did not need it has precedent in the industry.

This is why the starting point of the entire workflow has to be Process Mineralogy, not crushing, not grinding, not leaching. The mode of gold occurrence in the ore directly determines the entire process route. When native gold (Free Gold) is coarser than 75 microns, gravity separation alone can achieve high recovery. When gold exists as ultra-fine particles locked within the crystal lattice of sulfide minerals, known as Refractory Gold Ore, the sulfide crystal structure must be destroyed first before gold can be accessed by cyanide or other leaching agents. Getting this classification wrong cascades through every downstream decision.

The relationship between gold liberation and grind size is not linear. An economic fineness inflection point exists, beyond which the energy and grinding media costs required to improve recovery by one percentage point rise exponentially. Determining this inflection point requires systematic Liberation Analysis by Size Fraction, combined with automated mineralogy systems (such as MLA or QEMSCAN) to quantitatively characterize the mode of gold occurrence in each size class. Many feasibility studies do not achieve this level of precision when recommending grind size, relying instead on analogy.

Insufficient sample representativeness during the feasibility stage is the number one technical cause of project failure across the entire gold mining industry. Many projects complete all metallurgical testing with just tens of kilograms or even a few kilograms of drill core and proceed to issue a feasibility study, when the mineralogical variability within the orebody requires systematic sampling at hundreds of points to properly characterize. When the plant encounters a batch of ore with carbon or arsenic content far exceeding the test samples, the entire process route can shift from profitable to loss-making within weeks. Projects that invested in extensive Variability Sample testing during the feasibility stage show markedly better ramp-up speed and operational stability. In the eyes of management, this expenditure falls into the category of “money spent before a single gram of gold is mined,” and cutting it carries no psychological burden.

Gold ore classification and its impact on process selection:

Oxide ore, where sulfides have been naturally weathered and gold mostly exists in a liberated state, can typically go directly to cyanide leaching or heap leaching. Many oxide ores contain residual goethite that entraps ultra-fine gold, forming so-called “false oxide ore” with cyanidation leach rates far below expectations. Assessing the degree of oxidation cannot rely solely on sulfur content; it also requires iron valence state analysis and phase identification.

Primary sulfide ore, where gold is locked within pyrite, arsenopyrite, and other sulfide minerals, requires fine grinding or pre-oxidation to expose gold. Gold distribution within arsenopyrite crystals is often non-uniform, with gold content at crystal margins differing from the core by several fold, making grind size selection extremely delicate.

Carbonaceous ore (Preg-robbing Ore), where natural organic carbon in the ore acts like activated carbon and re-adsorbs dissolved gold from cyanide solution, known as the Preg-robbing effect. The activity of carbonaceous matter varies enormously. Graphitic carbon and amorphous organic carbon can differ in preg-robbing capacity by an order of magnitude. Measuring total organic carbon content alone is far from sufficient; a dedicated Preg-robbing index test is necessary.

Copper-gold ore, where copper minerals consume large quantities of cyanide during leaching, causing reagent costs to soar. The conditions most favorable for gold leaching are also the conditions under which copper dissolution is maximized. Every iteration of the process design is essentially a search for the balance point between gold recovery and cyanide consumption.

The inter-particle compression mechanism of HPGR (High Pressure Grinding Rolls) causes ore particles to crush against each other. The abundant micro-cracks in the product significantly increase the rate at which leaching agents penetrate into the ore interior, and unit energy consumption is notably reduced. Replacing conventional three-stage crushing with HPGR in heap leach projects can improve leach recovery by 3 to 8 percentage points. HPGR has an Edge Effect: material at the two ends of the rolls receives far less pressure than the central zone, resulting in a product size distribution that is finer in the middle and coarser at the edges. A mature HPGR circuit design must include effective screening in closed circuit and edge material recirculation systems.

A typical gold ore grinding circuit is a SAB or SABC circuit. SAG Mill efficiency is extremely sensitive to feed size distribution and ore competency indices. As mining advances into different parts of the orebody, ore hardness changes, causing SAG mill throughput to fluctuate sharply. Mature processing plants establish an Ore Blending Strategy to maintain stable grinding circuit operation. SAG mill liner lifter bar height and the liner wear profile have an outsized impact on mill power draw and throughput. In the first two to three weeks after new liner installation, lifter bar height is at its maximum, the ore trajectory inside the shell is optimal, and mill power utilization is at its peak. As liners wear, lifter bar height decreases, ore motion shifts from cataracting to cascading, and grinding efficiency declines noticeably. In the late stage of liner life, the mill is turning but much of the work is unproductive. Good grinding operations track the liner wear curve and proactively adjust mill speed and feed rate at mid-liner-life to compensate for efficiency decay.

Hydrocyclone operators, in pursuit of finer overflow particle size, sometimes blindly reduce the spigot diameter, causing underflow density to rise too high and produce “roping”, at which point the hydrocyclone has lost its classification function. Many processing plants have no video monitoring of the underflow discharge condition in the control room. The cost of installing a camera is trivial. The losses it can prevent are substantial. The reason this is often not done comes down to a lack of communication willingness between the control room and the plant floor.

On the process flowsheet of most modern gold processing plants, the gravity recovery circuit is only a small branch. Its economic value has little to do with its visual footprint on the flowsheet.

The GRG (Gravity Recoverable Gold) test, developed by Professor André Laplante, is used to assess what proportion of gold in the ore can be recovered directly by gravity separation. Even with a GRG value of only 30%, installing a centrifugal concentrator in the grinding circuit is worthwhile. Gravity-recovered gold can be directly smelted into Doré, completely bypassing the subsequent cyanidation and elution-electrowinning circuits. In a high gold price environment, the time value of this fast-track recovery channel tends to be overlooked by financial models, because most financial models do not apply differentiated discounting to gold recovery turnaround days.

The batch discharge interval for centrifugal concentrators has no textbook answer. Too short an interval reduces concentrate grade; too long and bed compaction may reduce recovery. Many processing plants spend months adjusting this parameter after initial commissioning before finding the optimum.

When the ionic concentration of tailings dam return water fluctuates severalfold between wet and dry seasons, froth properties change fundamentally. The accumulation of sulfate, calcium and magnesium ions, and residual xanthate decomposition products in return water alters the electrochemical environment at mineral surfaces, rendering the reagent regime calibrated under laboratory conditions completely ineffective. Some processing plants need entirely different flotation reagent schemes for wet and dry seasons, yet this water-quality-driven process switching is rarely written into the design report, because feasibility-stage testing is typically conducted with clean water. At technical conferences, when flotation optimization is discussed, there are plenty of papers on reagent regimes. Papers that systematically study the impact of process water quality are scarce. Process water composition changes with time and climate, data collection is difficult, clean reproducible experiments are hard to produce, and publishing papers is therefore difficult. So a variable that is critical in engineering becomes a cold topic in academia and a blind spot in the design phase.

Now, the conventional topics.

Gold itself has no natural floatability. The objects of flotation are the carrier minerals that host the gold: pyrite, arsenopyrite, chalcopyrite, and others. If the batch of pyrite with the highest gold content happens to have the slowest flotation kinetics (for example, due to partial surface oxidation reducing floatability), pursuing higher gold recovery means extending flotation time and adding more scavenger stages, with both concentrate grade and throughput being constrained. This is the fundamental contradiction of gold flotation, and it cannot be eliminated, only negotiated to an acceptable compromise in each specific project.

Potassium amyl xanthate is the most commonly used sulfide mineral collector. For carbonaceous ores, xanthate may simultaneously float organic carbon, severely contaminating the concentrate. Dithiophosphate-type collectors or mercaptobenzothiazole (MBT) can be used as alternatives in such cases. Adding collector within the grinding circuit (“in-mill addition”) exploits the surface activity of freshly exposed mineral surfaces to improve adsorption efficiency, but if the Eh of the grinding environment is unsuitable for collector adsorption on the target mineral surface, it instead causes reagent waste and unnecessary gangue entrainment. When to use in-mill addition and when not to depends on electrochemical environment measurements of the grinding circuit, not on “another plant did it and got good results.” Slurry electrochemical environments vary enormously with ore type and grinding media. Blindly copying another operation has a high failure rate.

Precise control of lime dosage is a challenge. The dissolution rate of lime in slurry is affected by particle size, water temperature, and slurry ionic strength, and pH meter readings in the high-alkalinity range tend to exhibit lag and drift.

Flash Flotation is installed directly on the hydrocyclone underflow in the grinding circuit, recovering coarse gold-bearing sulfide particles before they are over-ground. There is an inherent trade-off between unit-area throughput and concentrate grade in the flash flotation cell.

When gold is locked in sulfides, direct cyanidation recovery can fall below 30%. Pre-oxidation breaks down the sulfide encapsulation layer. Three methods exist. The choice between them is not a technology beauty contest. It is driven by five variables: sulfur-to-arsenic ratio in the concentrate, project scale, geographic remoteness, the operating team’s skill level, and the financing structure.

Start with scale. The autoclave for Pressure Oxidation (POX) has a largely fixed design, fabrication, and installation cost regardless of whether it processes 200 tonnes per day or 2000 tonnes per day of concentrate. The difference in core equipment investment is far smaller than the difference in capacity. Small and medium-scale gold projects that need pre-oxidation but whose concentrate volumes are insufficient to justify a POX autoclave find themselves in an awkward position. Some project teams are inclined toward POX, partly because it sounds more like “a proven industrial solution” in the financing narrative and passes technical due diligence from banks more easily. This decision driver can be misaligned with what the ore actually requires. BIOX or ultra-fine grinding combined with direct cyanidation may be the more pragmatic choice in such cases.

POX itself operates at 190 to 230 degrees Celsius and approximately 3000 kPa total pressure. It generates sulfuric acid in situ during oxidation, and when the ore contains carbonate gangue, the neutralization action of the acid effectively controls acidity in the reaction system. The autoclave interior simultaneously contains high-temperature strong acid and a strongly oxidizing atmosphere. Titanium linings and lead linings each have applicable temperature and acidity ranges, and a wrong choice leading to lining corrosion perforation is one of the most serious causes of unplanned shutdown at POX plants. Residues from pressure oxidation often contain basic iron sulfate precipitates, and if neutralization washing is insufficient, residual sulfate will continue to release acidity during cyanidation, consuming lime and degrading the leaching environment.

Roasting subjects flotation concentrate to oxidative treatment at 600 to 700 degrees Celsius. Temperature window control is critical: too low and sulfide decomposition is incomplete; too high and low-melting-point phases such as fayalite form, re-encapsulating gold in glassy slag and causing secondary locking losses. Over-roasting making leaching harder has occurred more than once in production practice. Roasting of arseniferous ores requires addressing arsenic volatilization and fixation, and modern roasting processes must be equipped with complete sulfuric acid off-gas recovery systems.

Bio-oxidation (BIOX) uses acidophilic iron- and sulfur-oxidizing bacteria to slowly oxidize sulfide minerals at ambient temperature and pressure. Capital costs are lower than POX, typical reaction cycle is 4 to 6 days. The optimal temperature for mesophilic strains is between 35 and 45 degrees Celsius, and reactor heating costs in cold regions can undermine project economic viability. When concentrate arsenic content exceeds a certain threshold, bacterial activity is significantly inhibited; this is BIOX’s hard ceiling.

Cyanide Leaching remains the core process of the global gold mining industry. Gold dissolution in cyanide solution requires simultaneously satisfying sufficient free cyanide concentration and adequate dissolved oxygen, while reducing substances in the slurry consume large quantities of oxygen, slowing the leach rate. The spatial distribution of dissolved oxygen within the slurry is highly non-uniform. The bottom and central areas of agitated leach tanks tend to be oxygen-depleted zones. Simply increasing agitator speed to improve aeration efficiency is not always feasible, because excessive agitation accelerates physical attrition of activated carbon, which in a CIL circuit means direct metal loss.

The AARL elution method offers drastically reduced elution time compared to the Zadra method, with higher water quality requirements and stricter temperature uniformity control in the elution column. If temperature dead zones exist within the column, incompletely eluted carbon will show reduced gold loading capacity in the next adsorption cycle.

Traditional electrowinning cells use stainless steel wool cathodes. Sludge stripping requires manual handling, making it the highest-concentration safety risk position in the entire processing plant and a weak point in metal security management. New-generation insoluble cathode electrowinning technology and automated sludge stripping systems are already in use at some mines, reducing metal losses during sludge processing and closing off the hardest-to-audit metal loss risk points in manual operations. Personnel control and material balance auditing in the smelting room still have significant management gaps at many mines. Because this involves internal security, it is essentially never discussed at technical conferences. Its impact on the plant’s Mine Call Factor should be quantified.

Heap Leaching is suitable for low-grade oxide ores, typically gold grades of 0.3 to 1.0 grams per tonne. Ore is crushed, stacked on an impermeable liner, and sprinkled with cyanide solution from the top, with gold-bearing pregnant solution collected from the bottom and fed into an ADR (Adsorption-Desorption-Recovery) system. Heap leach recovery typically ranges from 60% to 80%. Its extremely low unit processing cost makes it irreplaceable in large low-grade projects. Permeability management is the core challenge. Agglomeration can improve heap permeability. There is a fine balance in the amount of cement or lime added in the agglomeration drum: too much cement gives good pellet strength but reduces porosity and worsens permeability; too little cement and pellets disintegrate under gravity and vibration during stacking. Ores containing swelling clay minerals such as montmorillonite need pre-treatment with lime to passivate clay surfaces before adding binder, otherwise clay swelling will rupture already-formed pellets from the inside. The lime dosage and mixing time for this pre-treatment step must be determined by testing; no universal formula exists. Low atmospheric pressure at high altitude reduces dissolved oxygen concentration in solution, directly slowing the gold dissolution rate. In heap leach projects on the Andean Plateau above 4000 meters elevation, this effect can cause recovery losses of 5 to 10 percentage points. Mitigation measures include forced pre-oxygenation of the irrigation solution and additional ventilation piping within the heap.

In an increasing number of gold mining projects, water balance rather than ore processing capacity is the throughput ceiling. The fresh water makeup rate, process water recovery rate, tailings dewatering efficiency, and evaporation losses form a closed water cycle, and any imbalance can force the entire production line to reduce throughput. In arid regions, water rights permits and water source security can be more difficult to obtain than the mining rights themselves. Cyanide tailings must undergo cyanide destruction treatment before discharge to the tailings facility. Dry Stack Tailings uses high-pressure filtration to dewater tailings to below 15% moisture content before stacking, eliminating the risk of conventional tailings dam failure, while the recovered filtrate is of great significance in water-scarce regions.

The Mine Call Factor being chronically below 100% is a universal phenomenon across the global gold mining industry, with the shortfall ranging from 2% to 8%. The composition of this shortfall is extremely complex: sampling bias, slurry leaks, ultra-fine gold carried away in thickener overflow, fine carbon from activated carbon attrition lost with tailings, and gold losses in fumes and slag during smelting all contribute. Soluble Gold Loss in thickener overflow is easy to overlook. Clear overflow liquor appears to contain no solid particles. Nano-scale gold colloids and gold-cyanide complexes readily pass through thickener and filtration systems, accumulating to considerable losses over time.