Chapter 02

How ore forms

Enough geology to reason about the problem instead of pattern-matching on it. The organising idea is simple: an ore deposit is a place where a natural process concentrated an element by a factor of hundreds or thousands, and then the result survived. Everything else is detail about which process.

Start with the enrichment factor

Metals are already everywhere — just uselessly dilute. Ore is a concentration anomaly, and the size of the anomaly required varies enormously by metal. This table is the most quantitatively useful thing in the chapter.

MetalCrustal abundanceTypical ore gradeEnrichment needed
Iron~5%30–65%~10×
Aluminium~8%~25% (bauxite)~3×
Copper~25–60 ppm0.4–1.5%~100–300×
Nickel~50–80 ppm0.5–2%~100–300×
Zinc~70 ppm3–10%~500–1000×
Lead~10–15 ppm2–8%~1500–5000×
Gold~1–4 ppb1–10 g/t (= 1000–10000 ppb)~10³–10⁴×
Why this table predicts the difficulty of your problem

Iron needs a 10× enrichment, so iron deposits are huge, common and easy to find — often visible from orbit. Gold needs a factor of ten thousand, so gold deposits are tiny, rare, and the concentrating process must have been extraordinarily efficient and extraordinarily localised.

That maps directly onto your statistics. High-enrichment metals produce short variogram ranges, huge nuggets and violently skewed grade distributions — gold is the canonical nightmare for Chapter 8. Low-enrichment metals produce smooth, well-behaved, large-scale fields. Before you model anything, look up the enrichment factor; it tells you roughly what kind of random field you are about to meet.

Plate tectonics is the master control

Deposits are not scattered at random. They cluster in metallogenic provinces — belts where a particular tectonic setting ran a particular ore-forming process for millions of years. Knowing the setting narrows the search space enormously, and it is the strongest prior any targeting exercise has.

Tectonic settingWhat happensDeposits produced
Subduction arcs Oceanic plate dives, dehydrates, melts the mantle wedge; water-rich magmas rise and vent their metals Porphyry Cu-Mo-Au, epithermal Au-Ag, VMS. The Andes and the western Americas.
Continental rifts & basins Crust stretches, basins fill with sediment, hot brines circulate along a redox front Sediment-hosted Cu-Co (Zambian Copperbelt), SEDEX and MVT Pb-Zn.
Cratons & greenstone belts Ancient stable crust; deep crustal faults tap metamorphic fluids during orogeny Orogenic gold (Abitibi, Yilgarn), komatiite-hosted Ni.
Large layered intrusions Enormous magma chambers cool slowly; dense sulphide and chromite layers settle out Ni-Cu-PGE and chromite (Bushveld, Sudbury, Norilsk).
Deep weathering profiles Tropical weathering leaches everything soluble, leaving the insoluble behind Bauxite, nickel laterite, supergene copper enrichment.

Three engines

1 · Magmatic — the melt does the sorting

As magma cools, crystals form in sequence and elements partition between melt, crystals and any separate fluid or sulphide phase. Two mechanisms matter. Sulphide saturation: if a mafic magma becomes saturated in sulphur, an immiscible sulphide liquid separates, and nickel, copper and platinum-group elements partition into it with extreme preference — then it sinks and accumulates. Fractional crystallisation: elements too large or oddly-charged to fit into common minerals (the incompatible elements — lithium, tantalum, rare earths) concentrate in the last dregs of melt, producing pegmatites and carbonatites.

2 · Hydrothermal — hot water is the great concentrator

The majority of the metals you will care about were moved and deposited by hot aqueous fluids. This is the engine to understand properly, because it has four separable components — and each one leaves a different, mappable trace.

ComponentQuestion it answersMappable evidence
SourceWhere did the metal and the fluid come from?A fertile intrusion; a thick metal-bearing sedimentary pile; a particular magma chemistry
TransportHow did the metal stay dissolved?Chloride- or bisulphide-rich brines; salinity and temperature from fluid inclusions
PlumbingWhat path did the fluid take?Structure — faults, shear zones, breccias, permeable beds. Often the single best predictor.
TrapWhy did the metal drop out here?A cause of sudden change: cooling, boiling, fluid mixing, a redox boundary, reaction with wall rock

Metals travel as dissolved complexes — gold as a bisulphide complex, base metals typically as chloride complexes. They stay in solution as long as conditions stay stable, and precipitate when something changes abruptly. That is why ore bodies sit at boundaries: a fault intersection where pressure drops and the fluid boils, a reactive limestone horizon, a redox front where oxidised brine meets reduced carbon-bearing rock.

The feature-engineering translation

"Ore forms at abrupt boundaries" is a statement about gradients and intersections, not values. The predictive features are rarely "magnetic intensity here" — they are distance to a fault intersection, gradient of the magnetic field, proximity to a lithological contact, local structural complexity. If your prospectivity model is fed raw layers and no derived structural features, you have withheld the actual physics from it.

3 · Sedimentary & surficial — gravity and weather

Some deposits need no heat at all. Placers concentrate dense, durable minerals — gold, tin, diamonds — by fluid sorting in rivers and beaches. Laterites form where tropical weathering strips out soluble elements over millions of years, leaving aluminium or nickel behind. Banded iron formations, the source of nearly all the world's iron, record the chemistry of an ocean that had not yet been oxygenated.

Alteration: the target is bigger than the ore

This is the most operationally important idea in the chapter. Hydrothermal fluid does not only deposit metal — it chemically rebuilds the rock it passes through, over a volume far larger than the ore body itself. An orebody might be 500 m across; its alteration halo can be several kilometres, arranged in a predictable zoned pattern.

In a porphyry copper system the zoning runs outward from a potassic core (K-feldspar, biotite, magnetite) through phyllic (quartz-sericite-pyrite) and argillic (clays) to a peripheral propylitic zone (chlorite-epidote-carbonate). Each of those assemblages has distinct physical properties — magnetite makes the core magnetic, pyrite makes the phyllic zone chargeable, clays have diagnostic infrared absorption features.

Why exploration works at all

A 500 m orebody under 300 m of cover is a needle. A 4 km zoned alteration system with a magnetic core, a chargeable shell and a clay signature is a target you can actually detect from the air — and the zoning tells you which direction the centre lies. Most successful exploration is vectoring up an alteration gradient toward a centre you cannot yet see. This is also exactly why hyperspectral matters (Chapter 5): it maps alteration mineralogy directly.

Weathering, cover, and the reason this got hard

Near surface, sulphides oxidise. This produces an oxide cap and often a gossan — a rusty, iron-stained outcrop that is the classic prospector's signpost. Below it, leached metal can re-precipitate as a supergene enrichment blanket at the water table, sometimes at several times the primary grade.

Everything with a gossan in accessible country was found long ago. The frontier is ground buried under transported cover — younger sediment, glacial till, desert sand — that has no chemical relationship to the bedrock beneath. Surface geochemistry over transported cover samples the wrong rock. That is the central difficulty of modern exploration, and the reason geophysics and inference matter so much more than they used to.

The mineral systems framework

Modern targeting organises all of the above into one checklist. A deposit requires every ingredient; missing one means no deposit, regardless of how good the others look.

IngredientRequirementProxy you can map
1 · SourceA reservoir of metal and fluidFertile intrusions, metal-rich stratigraphy, isotopic signatures
2 · EnergySomething to drive circulationMagmatic heat, tectonic stress, burial-driven fluid flow
3 · PathwayPermeable structure connecting source to trapCrustal-scale faults, shear zones, breccia bodies, unconformities
4 · TrapA gradient that forces precipitationRedox boundaries, reactive lithologies, structural dilation sites
5 · PreservationThe deposit must survive to today at a depth you can mineExhumation history, cover thickness, post-ore deformation
This is a structured prior, and you should use it as one

The mineral systems framework is the field's own answer to "what features should the model see." Two consequences worth taking seriously. First, it is conjunctive — all five must be present — which argues for multiplicative or logical combination rather than a purely additive score, and makes a plain linear model structurally wrong. Second, preservation is routinely forgotten: a model trained on outcropping deposits learns the preservation history of the region it was trained on, and transfers badly to a region with a different exhumation history. That is a concrete, geologically-named mechanism for the covariate shift you will measure in Chapter 9.

Sources for this chapter

  • Robb, Introduction to Ore-Forming Processes (Wiley, 2nd ed. 2020). The best single textbook for exactly this chapter's material, and readable by a non-geologist. If you buy one geology book, buy this one.
  • Ridley, Ore Deposit Geology (Cambridge, 2013). Deposit-model-by-deposit-model reference; use it to look up whichever system your project actually is.
  • Robert et al., "Models and exploration methods for major gold deposit types", in Proceedings of Exploration 07 — a widely used practical bridge from deposit model to exploration method. Freely available via DMEC / Decennial Mineral Exploration Conferences.
  • McCuaig & Hronsky, "The mineral system concept: the key to exploration targeting", SEG Special Publication 18 (2014). The canonical statement of the source–energy–pathway–trap–preservation framework used above.
  • Rudnick & Gao, "Composition of the Continental Crust", Treatise on Geochemistry. Where crustal abundance numbers come from — worth knowing that published values differ by reference and by whether "upper" or "bulk" crust is meant, which is why the table above gives ranges.
  • Society of Economic Geologists — SEG Special Publications and Economic Geology are where deposit models are established and revised.
Crustal abundances and ore grades are order-of-magnitude figures compiled from standard references; both vary substantially by source and by deposit.