Mining in South America
Chile Peru and Argentina

Chilean copper concentrators were designed to process supergene chalcocite at grades above 1%. They are now processing primary chalcopyrite at 0.6%.

Chalcocite is Cu₂S, roughly 80% copper by weight, and it formed in supergene enrichment blankets where descending groundwater reconcentrated copper at the water table. Coarse-grained. Soft. Responded predictably to the xanthate-MIBC flotation recipes the industry had refined over decades. Cochilco recorded a national average grade of 1.03% in 2004. By 2022, 0.64%. The enrichment blankets at Escondida, Chuquicamata, Collahuasi, Los Pelambres are thin or gone.

Chalcopyrite underneath is CuFeS₂, only 34.6% copper by weight, finer-grained, harder, and interlocked with gangue minerals at a scale that demands finer grinding to achieve liberation. Grinding finer through the SAG and ball mill circuit, already 40% to 60% of total plant electricity, generates ultrafine particles below roughly 10 microns. Ultrafines do not attach reliably to air bubbles in flotation cells. They follow water instead, reporting to concentrate as entrained gangue, diluting grade. They also increase slurry viscosity, changing air dispersion patterns and bubble-particle collision frequency inside the cells.

Escondida, Los Pelambres, Collahuasi, Chuquicamata all hit versions of this problem within a few years of each other as their respective mine plans descended through the supergene-primary boundary. The solutions varied by site because the mineralogy varies. Chalcopyrite grain size at Escondida is different from chalcopyrite grain size at Los Bronces. Gangue mineralogy matters: clays, talc, and micas are naturally hydrophobic and report to concentrate as contaminants without even needing collector reagent. Some operations added regrind mills on rougher concentrate to improve liberation before cleaning stages. Some installed staged flotation reactors separating coarse and fine particle treatment. Some reformulated collector chemistry entirely, moving from standard potassium amyl xanthate to alternative collectors with better selectivity on fine chalcopyrite. Each of these changes took pilot testing, months of full-plant trials, and substantial capital.

Geometallurgy was supposed to help with this. Block models now carry predicted flotation response and grind characteristics alongside grade and rock type for each mineable block. The idea is to forecast concentrator performance based on scheduled ore delivery. In practice, the drill spacing that calibrates these models samples a fraction of one percent of the ore body. Geometallurgical surprises still happen. A block predicted at 88% recovery comes through at 84% because of a clay seam the drilling missed. Four points of recovery on 100,000 tonnes of ore is a lot of lost copper. The models improve with production data fed back over time, but they remain approximations, and the gap between predicted and actual recovery is a persistent source of variance between planned and actual output at Chilean operations. Geometallurgy is useful. It is not the solution its proponents sometimes present it as.

Selective rejection of enargite from chalcopyrite concentrate is technically feasible. It exploits the fact that enargite has different electrochemical properties from chalcopyrite, so at controlled pH and pulp potential, enargite can be depressed while chalcopyrite floats. Some operations have implemented enargite rejection circuits. These add reagent cost, require tight process control, and reduce overall throughput because the circuit now has an additional separation task competing for flotation capacity.

Desalination added another chemical variable.

Atacama. No rainfall. Groundwater overallocated. Escondida's desal plant processes over 2,500 liters per second pumped uphill more than 3,000 meters at roughly 60 MW of continuous power demand for pumping alone. When desalinated water entered flotation circuits, copper recovery dropped 1% to 2% at multiple operations.

Chloride ions remaining after desalination changed the electrochemical environment at the mineral surface. Xanthate collectors adsorb onto chalcopyrite through a mechanism mediated by the redox potential at the mineral-solution interface. Chloride shifts that potential. MIBC frother behavior also changed because chloride affects bubble coalescence dynamics. Published research from the University of Chile and the University of Queensland, among others, has investigated the mechanism, and the magnitude of the effect varies with chloride concentration, pH, and the specific collector being used. Some operations were hit harder than others depending on their particular combination of water chemistry, ore mineralogy, and reagent recipe.

Codelco produced above 1.8 million tonnes at peak. Roughly 1.3 million by 2023. Chuquicamata Underground and Nuevo Nivel Mina at El Teniente together carry over $20 billion in capital costs with repeated overruns. Chuquicamata Underground is block caving beneath a pit mined since 1915 through rock destressed by a century of excavation. The company's governance has been described by its own executives to Chilean press as structurally undercapitalized because of annual surplus transfers to the national treasury. The 2023 royalty reform, price-linked progressive taxation, prompted BHP among others to flag reduced fiscal competitiveness in public filings.

Conga in Cajamarca holds approximately 11.6 billion pounds of copper and 11.6 million ounces of gold per Newmont's 2010 resource estimate. National EIA approved. Construction started. Four high-altitude lakes to be drained and replaced with engineered reservoirs.

The regional governor organized mass opposition. Protests in 2011 and 2012. National state of emergency, then retreat. Frozen for over a decade. Newmont wrote down the asset.

Las Bambas reached production. Over 400 cumulative days of road blockades on the 300-kilometer concentrate route to Matarani per Defensoría del Pueblo records 2016-2024. Quellaveco in Moquegua cost Anglo American $5.5 billion with over twenty years from exploration to production, sustained by pre-construction engagement and participatory water monitoring. Whether Quellaveco is a replicable model or an outlier enabled by Anglo American's specific financial capacity and Moquegua's relatively favorable politics is an open question that the industry prefers not to examine too closely.

Peruvian silver, 3,500 to 4,000 tonnes per year per USGS, is almost entirely byproduct from polymetallic base-metal mines. Supply follows copper and zinc economics.

Mg:Li ratio determines whether an Argentine lithium project makes money or loses it.

Atacama: below 7:1. Most Puna salars: 3:1 to 20:1, some above 30:1. Magnesium co-precipitates with lithium during evaporation and is hygroscopic. A 600 ppm lithium salar at 6:1 Mg:Li beats an 800 ppm salar at 20:1 on operating cost. The lithium grade goes in investor presentations. The Mg:Li ratio sits in the NI 43-101 appendix. Junior companies on the TSX-V and ASX acquired Puna claims during the 2021-2022 price spike. Some have brine chemistry incompatible with economic evaporative processing.

Provincial governance: San Juan permits mining, Mendoza bans open-pit methods by law, Chubut has reversed across elections. Mining rights sit with provinces.

Undeveloped copper: Josemaría (Lundin Mining), Los Azules (McEwen Copper), MARA (Glencore/Pan American Silver), combined potential above 500,000 tonnes per year, none in production. Federal export retention taxes, currency controls, and inflation above 100% in 2023 per INDEC.

Bolivia's Salar de Uyuni, largest lithium resource, Mg:Li above 20:1, state-led development, over a decade, negligible output.

Cochilco data show Chilean industry average copper grade roughly halving over twenty years. Peruvian grades at mature operations are on the same trajectory with a lag. Lower grade increases grinding energy, water, reagent consumption, and tailings volume per tonne of metal. Each carries its own cost and permitting burden. Copper price increases compensate imperfectly because costs rise concurrently.