Ore Mineralogy and Liberation in Mineral Processing: The Complete Guide

In mineral processing, no single factor shapes plant performance more fundamentally than the mineralogy of the ore being treated. Before a single crusher turns or a flotation cell bubbles, the mineralogical character of the rock — which minerals are present, how they are arranged, how finely they are intergrown — has already set the boundaries of what is achievable. Grade and recovery targets, grind size requirements, reagent selection, and concentrate quality are all ultimately expressions of mineralogy. Engineers who understand their ore at the mineral level make better decisions at every stage of the processing flowsheet. Those who do not are perpetually surprised by the results they get. This guide covers the complete story of ore mineralogy and liberation: from the formation of ore deposits and the distinction between valuable minerals and gangue, through the textural controls on liberation, to the powerful automated analytical tools that now let engineers quantify mineralogy in ways that were impossible a generation ago. It concludes with the implications of mineralogy for comminution strategy, flotation selectivity, and the challenge of refractory ores that refuse to yield their values by conventional means.

Why Ore Mineralogy Drives Everything in Mineral Processing

Mineral processing has two fundamental objectives: liberation and concentration. Liberation means breaking an ore apart until the valuable minerals are free — or at least sufficiently exposed — to be separated from the waste rock. Concentration means using the physical or chemical differences between liberated mineral particles to sort the valuable ones into a product stream and reject the rest as tailings. Both objectives are governed, from the outset, by mineralogy.

The mineralogy of an ore determines which separation methods are even feasible. A magnetite iron ore lends itself to low-intensity magnetic separation. A chalcopyrite copper ore requires flotation. A cassiterite tin ore may call for gravity concentration. A gold ore where the gold is locked inside arsenopyrite will need pressure oxidation or bacterial leaching before conventional cyanidation can work. None of these choices can be made rationally without knowing what minerals are present and how they behave.

Mineralogy also sets hard limits on what is achievable. The grade of concentrate that can theoretically be produced is capped by the metal content of the mineral host: chalcopyrite (CuFeS₂) contains at most 34.6% copper; galena (PbS) at most 86.6% lead. No amount of process optimization can push concentrate grades beyond these mineralogical ceilings. Similarly, the maximum achievable recovery is constrained by the degree of mineral liberation that is practically attainable at a given grind size. These mineralogical limits define what engineers call the liberation-limited grade-recovery curve, which represents the theoretical best performance for a given ore sample, and against which actual plant performance is measured.

Beyond the fundamental process choice and theoretical limits, mineralogy influences day-to-day operations: reagent consumption in flotation, wear rates on grinding media, settling behavior in thickeners, filtration rates, and concentrate moisture. A thorough mineralogical understanding is not an academic luxury — it is the foundation of effective process engineering (Wills & Finch, 2016).

Valuable Minerals and Gangue: Understanding Your Ore

An ore is defined as a rock from which one or more metals or minerals can be profitably extracted. It consists of two broad components: the valuable minerals (sometimes called pay minerals) that contain the commodity of interest, and the gangue — the remaining minerals that have no economic value in the current processing context, or whose value is insufficient to justify separate recovery.

Valuable minerals take a wide variety of forms depending on the geological environment of ore formation. In native ores, metals occur in their elemental state: native gold, native silver, and native copper are classical examples. In sulfide ores — the most economically important class of metallic ores — metals are bound to sulfur in minerals such as chalcopyrite (CuFeS₂), galena (PbS), sphalerite (ZnS), pentlandite ((Fe,Ni)₉S₈), molybdenite (MoS₂), and arsenopyrite (FeAsS). Oxide ores contain metals bound to oxygen, often through secondary weathering processes: examples include malachite and azurite (copper carbonates), hematite and magnetite (iron oxides), cassiterite (tin oxide, SnO₂), and rutile (titanium oxide, TiO₂). Complex ores contain profitable amounts of more than one valuable mineral, presenting additional challenges in producing separate, clean concentrates.

Gangue minerals are dominated by the rock-forming silicates and carbonates that make up the bulk of the Earth’s crust: quartz (SiO₂), feldspars, micas, amphiboles, pyroxenes, chlorite, talc, calcite, and dolomite are among the most common. While gangue is broadly defined as valueless, context matters: fluorite and baryte, sometimes found with galena deposits, are themselves commercially valuable nonmetallic minerals. Iron sulfide gangue, particularly pyrite, is frequently associated with copper, lead, and zinc sulfides and must be selectively depressed during flotation to produce clean, penalty-free concentrates.

The ratio of valuable mineral to gangue in an ore body — the grade — varies enormously across commodity types. Porphyry copper deposits, which represent the world’s dominant source of copper, may be mined at grades as low as 0.3% Cu. In contrast, iron ore deposits may require grades of 30% Fe or more to be economic after beneficiation, while a high-grade gold deposit might be defined at 5 g/t Au. Diamond ores operate at extraordinarily low grades, sometimes below 0.1 carats per tonne, yet remain viable because of the extreme value of the product. These grade differences drive radically different processing strategies and place very different demands on the precision of separation (Wills & Finch, 2016).

Understanding the full mineralogical inventory of an ore — not just the dominant valuable mineral but the complete suite including trace minerals, minor impurity phases, and gangue types — is essential for anticipating processing difficulties and designing reagent schemes that achieve the selectivity required.

Ore Texture and Structure: How They Affect Processing

Beyond the simple question of which minerals are present, the spatial arrangement of those minerals within the rock — the ore texture — is one of the most important controls on how an ore behaves during processing. Texture encompasses grain size, grain shape, the nature of grain boundaries, the degree of intergrowth between adjacent minerals, the spatial distribution of mineral phases, and the presence of structural features such as veins, banding, or foliation.

Grain size is the most obvious textural parameter and directly controls the liberation size — the particle size to which ore must be ground before valuable minerals are sufficiently free for separation. A coarse-grained, free-milling ore where sulfide grains are several hundred micrometres in diameter may liberate adequately at a grind size of 150–200 µm. A fine-grained ore where the same sulfides are intimately intergrown with silicates at grain sizes of 10–20 µm may require grinding to below 20 µm to achieve comparable liberation, at vastly greater energy cost.

The pattern of intergrowth is also critical. In some ores, valuable minerals occur as discrete, well-defined grains with clean boundaries against the gangue. Breakage preferentially follows these boundaries — a phenomenon known as liberation by detachment — and liberation is relatively easy to achieve. In other ores, the valuable mineral is finely disseminated through the gangue matrix in an irregular, patchy intergrowth with no clean planes of weakness. Here, random breakage produces a large proportion of middling particles, and achieving high liberation demands very fine grinding.

Mineralogists recognize a spectrum of ore textures from simple to complex. At the simple end sit the so-called free-milling ores, typified by coarse-grained galena-sphalerite ores where the sulfide grains are large and well-defined, as seen in examples such as the Pine Point deposit in Canada. At the complex end are massively intergrown textures where multiple sulfide species form fine, interlocking aggregates, as at the McArthur River zinc-lead deposit in Australia, where grain sizes below 10 µm defeated processing attempts for decades before fine-grinding technology caught up with the mineralogy (Wills & Finch, 2016).

Structural features also influence processing. Vein textures can mean that valuable minerals are concentrated in narrow bands, facilitating selective breakage along vein margins. Banded or laminated textures in iron formations and sedimentary manganese ores create predictable liberation behaviour if breakage follows the banding. Foliated textures in metamorphic host rocks can generate platy gangue particles — particularly micas — that cause problems in flotation through slime coating and bubble loading.

The practical implication is that a mineralogical characterization program must go beyond modal analysis (how much of each mineral is present) to include textural characterization. An ore body containing 2% copper in coarse, well-liberated chalcopyrite grains is a fundamentally different processing challenge from one containing 2% copper in fine, intimately intergrown chalcopyrite — even though both have the same head grade. Recognizing textural complexity early, ideally during the exploration and feasibility phase, allows metallurgists to design appropriate test programs and select the right grinding technologies (Wills & Finch, 2016; Hoal et al., 2009).

Liberation: Definition, Liberation Size, and Locked Particles

Liberation is the process by which valuable minerals are separated from gangue by comminution — crushing and grinding — so that they exist as individual free particles available for physical concentration. It is perhaps the single most important concept in mineral processing, because no separation process can distinguish between a valuable mineral and gangue if they are locked together in the same particle.

Complete liberation occurs when every particle in the ground ore consists entirely of either valuable mineral or gangue, with no mixed or locked particles. In practice, complete liberation is never achieved. Even when ore is ground to particle sizes smaller than the mineral grain size, some locked or composite particles — called middlings — will always be present. This is because comminution is a random process: breakage planes do not follow mineral grain boundaries perfectly, and some grain boundaries will always be intersected by the breaking planes rather than opened along them.

The degree of liberation is defined as the percentage of the valuable mineral occurring as free particles in relation to the total mineral content in both free and locked particles. It is a key metric for assessing whether the comminution circuit is delivering ore in a condition suitable for the downstream separation process. Quantification of liberation is now routine using automated scanning electron microscopy-based systems, which scan polished cross-sections of sample and report liberation class distributions (Wills & Finch, 2016).

The liberation size is the particle size at which a satisfactory degree of liberation is achieved for a given ore and a given separation process. It is not a fixed property of the ore but a practical target determined by the balance between the benefits of finer grinding (higher liberation, better concentrate grade, lower tailings loss) and the costs (energy consumption, generation of fine particles that are difficult to treat). For base metal sulfide ores, target grind sizes of around 100 µm were common in the 1960s when treating relatively coarse-grained ores. Modern, finer-grained deposits may require target sizes of 20–30 µm or even finer, driving the development of stirred mill technologies such as the IsaMill and the Stirred Media Detritor.

The Gaudin liberation model, one of the earliest mathematical treatments of liberation, predicts that to achieve 75% liberation of a mineral, the particle size must be reduced to roughly one-tenth of the mineral grain diameter. While this model relies on simplifying assumptions, it captures the essential insight that fine grinding is disproportionately expensive relative to the marginal improvement in liberation it delivers at very fine sizes, and helps explain why liberation-limited grade-recovery curves look the way they do.

Locked particles are characterized by their liberation class — typically expressed as the fraction or percentage of the particle area occupied by the valuable mineral. Particles with greater than 95% valuable mineral content are generally classified as fully liberated (free). Particles with 80–95% valuable mineral are liberated to a degree sufficient for most practical separation processes. Below 80%, particles are non-liberated, with sub-classification into binary (two-mineral) composites and complex (multi-mineral) composites. The behavior of these particle classes in separation processes differs markedly: fully liberated chalcopyrite particles will respond vigorously to copper collectors in flotation, while complex locked particles containing only 10–20% chalcopyrite may contribute little to concentrate mass while diluting concentrate grade if they report to the froth (Wills & Finch, 2016).

Not all processing situations require high liberation. Gravity and magnetic separation can recover value from locked particles, provided there is a sufficient density or magnetic susceptibility contrast between the locked composite particle and the free gangue particles. Flotation, however, is more demanding: it requires at least a surface exposure of the valuable mineral for reagent adsorption to occur. This makes liberation analysis particularly critical for flotation circuit design.

Process Mineralogy: QEMSCAN, MLA, and Automated Techniques

Process mineralogy — the application of mineralogical knowledge to mineral processing — has been transformed by the development of automated scanning electron microscope (SEM)-based analysis systems over the past three decades. Where mineralogists once relied on time-consuming manual optical microscopy to characterize ore samples, modern automated systems can analyze thousands of particles per sample in a few hours, delivering quantitative data on mineral identification, grain size, liberation, and mineral associations with a statistical robustness that manual methods cannot match.

Three principal commercial systems have become industry standards. QEMSCAN (Quantitative Evaluation of Minerals by Scanning Electron Microscopy), developed by CSIRO Australia and commercialized through FEI (now Thermo Fisher Scientific), uses a focused electron beam to generate backscattered electron (BSE) images and energy-dispersive X-ray (EDS) spectra from polished sections. The BSE image contrast reflects average atomic number, allowing different mineral phases to be distinguished; the EDS spectrum provides elemental composition for each measurement point, enabling mineral identification against a library of reference spectra. MLA (Mineral Liberation Analyser), developed at the Julius Kruttschnitt Mineral Research Centre (JKMRC) at the University of Queensland and also commercialized by FEI, operates on similar principles with different acquisition and processing algorithms. The more recently introduced TIMA (Tescan Integrated Mineral Analyser) provides comparable capability from a different vendor. All three systems produce pseudo-colour maps of polished cross-sections, from which modal mineralogy, grain size distributions, liberation spectra, and mineral association matrices can be extracted (Wills & Finch, 2016; Gottlieb et al., 2000).

The key outputs of automated mineralogy for mineral processing include: modal mineralogy (the weight percentage of each mineral phase in the sample); grain size distribution for each mineral phase; mineral liberation data, typically reported as liberation class distributions showing the proportion of each mineral in each liberation bin; mineral association data, showing the statistical probability of each mineral being associated with each other mineral in composite particles; and elemental deportment, showing how each element of interest is distributed among the various mineral phases that carry it.

These data inform practically every aspect of process design and optimization. Modal mineralogy reveals the full gangue mineralogy, identifying problematic phases such as talc, chlorite, or micas that can cause reagent overconsumption or froth stability problems in flotation. Grain size data define the liberation size target for comminution. Liberation class distributions allow the construction of liberation-limited grade-recovery curves, providing a theoretical benchmark against which actual flotation performance can be compared. Mineral association data highlight which gangue phases are most intimately associated with valuable minerals and therefore hardest to separate. Elemental deportment studies reveal whether a metal of interest is distributed among multiple mineral carriers, some of which may be poorly amenable to the chosen separation process.

Beyond polished cross-sections, X-ray micro-computed tomography (micro-CT) has emerged as a powerful complementary tool, enabling three-dimensional visualization of mineral distributions within particles and drill core without destructive preparation. Micro-CT provides true 3D liberation data, avoiding the systematic overestimation of liberation that arises from taking 2D sections through locked particles — a problem that motivated decades of stereological correction research (Wills & Finch, 2016; Lin et al., 2013).

Automated mineralogy has also become a routine monitoring tool in operating concentrators. Regular analysis of mill feed, flotation feed, concentrates, and tailings can reveal shifts in ore mineralogy as the mine advances through different ore zones, enabling rapid adjustment of operating parameters before significant losses occur. This application of process mineralogy to plant control is increasingly referred to as on-line mineralogy or mineralogical process monitoring (Hoal et al., 2009; Evans et al., 2011).

Deportment Studies: Where Does the Value Go?

A deportment study is a systematic mineralogical investigation into how the valuable metals in an ore are distributed — or deported — among the various mineral phases present. The question is deceptively simple but profoundly important: when we say an ore contains, for example, 1.5% Ni, in which minerals does that nickel actually reside? The answer determines what processing methods are viable and what recovery levels are realistically achievable.

In a nickel sulfide ore, the nickel of interest is typically hosted primarily by pentlandite, which is recoverable by flotation. However, nickel can also substitute into the crystal lattice of pyrrhotite, olivine, serpentine, and other silicate or oxide gangue minerals at concentrations of a few hundred to a few thousand parts per million. This nickel, locked in gangue mineral lattices, cannot be recovered by flotation regardless of how fine the ore is ground or how optimized the reagent scheme is. A deportment study that identifies, say, 15% of the nickel budget residing in non-recoverable silicate lattices fundamentally changes the recovery target: the theoretical maximum recovery is not 100% but 85%, and any process design that targets higher recovery is chasing an impossibility (Wills & Finch, 2016).

Similar deportment questions arise in virtually every ore type. In gold ores, gold may be present as coarse free-milling particles recoverable by gravity concentration, as fine gold associated with sulfide minerals, as sub-microscopic gold in solid solution within pyrite or arsenopyrite lattices (the so-called invisible gold or refractory gold), and as gold adsorbed on carbonaceous material. Each of these gold forms requires a different recovery approach, and the proportion of gold in each form defines the processing pathway required for maximum extraction. In copper ores, copper may be deported among chalcopyrite, bornite, chalcocite, covellite, malachite, azurite, and chrysocolla — sulfides and oxides with very different flotation responses. In lead-zinc ores, the deportment of silver among galena, sphalerite, and tetrahedrite phases influences both the economic value of each concentrate stream and the selection of reagents.

Deportment studies are performed using a combination of automated mineralogy (QEMSCAN/MLA for major mineral phases) and microanalytical techniques such as electron probe microanalysis (EPMA), laser ablation ICP-MS, and dynamic secondary ion mass spectrometry (SIMS) for trace element distributions within individual mineral grains. The combination of bulk mineralogy with microanalysis allows the complete metal balance to be closed across all mineral phases, expressed as a deportment table showing the weight percentage of each target metal carried by each mineral host (Wills & Finch, 2016).

The practical value of deportment data extends through the entire life of a mine. At the feasibility stage, deportment studies on drill core composites define the recoverable metal budget and the processing approach. During operations, periodic deportment studies on mill feed and tailings identify changing ore mineralogy that may be causing recovery losses, distinguishing between losses due to liberation failure (locked particles reporting to tailings) and losses due to deportment to refractory mineral phases. In tailings reprocessing investigations, deportment studies on historic tailings reveal whether unrecovered value is accessible with modern technology or remains permanently locked in non-recoverable phases.

The Link Between Mineralogy and Comminution Strategy

Comminution — the reduction of ore particle size through crushing and grinding — is the most energy-intensive step in mineral processing, accounting for up to 50% of a concentrator’s total energy consumption. The primary purpose of comminution is liberation: breaking the ore until valuable minerals are free from gangue. Mineralogy therefore dictates not only the target grind size but also the most efficient comminution strategy for achieving that target (Wills & Finch, 2016).

The starting point is always the liberation size: how fine must the ore be ground to achieve adequate liberation of the valuable minerals? This is determined by the grain size of the valuable minerals and the textural complexity of their intergrowths with gangue. A coarse-grained porphyry copper ore may liberate chalcopyrite adequately at 150–200 µm. A fine-grained epithermal gold ore may need grinding to 30–50 µm. An ultra-fine-grained refractory gold ore where gold occurs as sub-micron inclusions in arsenopyrite may require grinding to below 10 µm — a task that cannot be accomplished economically in conventional ball mills and demands stirred mill technology.

Beyond the target size, mineralogy influences the selection and design of the comminution circuit. The relative hardness of the valuable mineral and gangue phases can lead to differential breakage: softer sulfide minerals tend to break preferentially at grain boundaries, promoting liberation, while hard silicate gangue remains as coarser fragments. The presence of clay minerals or talc can cause grinding circuit inefficiency through slurry rheology problems. The presence of weak, cleavable minerals such as mica or graphite can generate fine flake-shaped particles that interfere with subsequent classification and separation.

Research into non-random breakage seeks to exploit mineralogy to improve liberation efficiency. Microwave heating induces differential thermal expansion among mineral phases, generating stress concentrations preferentially at grain boundaries and pre-fracturing the ore along inter-mineral contacts before conventional comminution. High-voltage pulse disintegration applies electrical pulses that fracture selectively along grain boundaries. Both approaches aim to increase liberation at a coarser particle size, reducing the energy and media cost of fine grinding while improving separation efficiency (Wills & Finch, 2016; Kingman et al., 2004).

The connection between mineralogy and grinding circuit optimization is also expressed through Bond work index testing. Different ore types within the same deposit — defined by their lithology, alteration, and mineralogy — frequently exhibit significantly different grinding hardness values. Understanding this relationship, and mapping it through the ore body by geological domain, is a core element of geometallurgical practice and is discussed further in the companion article on geometallurgy.

For complex or fine-grained ores, a multi-stage comminution strategy may be employed. In two-stage separation flowsheets, ore is first ground to a size where the bulk of the coarse gangue is liberated — even if the valuable mineral is not yet fully free — and a rough separation is made to discard most of the gangue at a coarse size. The resulting rougher concentrate is then reground to achieve higher liberation before final cleaning. This staged approach minimizes the total amount of fine grinding required, reducing energy costs while still achieving good final liberation in the concentrate stream.

Mineralogy and Flotation: Surface Chemistry and Selectivity

Froth flotation is the most widely used separation process in mineral processing, and its performance depends critically on the surface chemistry of the minerals being separated — which is itself a function of mineralogy. The fundamental principle of flotation is selectivity: making the surfaces of valuable mineral particles hydrophobic (water-repellent) so that they attach to air bubbles and float to the froth, while keeping gangue mineral surfaces hydrophilic (water-loving) so that they remain in the pulp and report to the tailings.

The mineralogy of the ore determines the natural surface chemistry of the particles and therefore the reagent scheme required to achieve selectivity. Sulfide minerals such as chalcopyrite, galena, and sphalerite are naturally weakly hydrophobic — a consequence of their covalent surface bonding — but this natural hydrophobicity is insufficient for practical flotation. Selective collectors such as xanthates, dithiophosphates, and dithiocarbamates adsorb at the surface of target sulfide minerals through metal-sulfur bonding interactions, greatly enhancing hydrophobicity and bubble attachment probability. The selectivity of this adsorption between different sulfide minerals depends on the electrochemical and surface chemical differences among them, which are inherent mineralogical properties.

The gangue mineralogy is equally important for flotation design. Talc and chlorite are naturally hydrophobic minerals that will float without any collector addition, causing severe problems in the flotation of sulfide ores hosted in talcose or chloritic rocks — particularly in some nickel and platinum group metal ores. Calcite, a common carbonate gangue, competes with collector adsorption on sulfide surfaces through calcium ion activation and requires careful pH control and depressants to minimize its effect on concentrate grade. Clay minerals adsorb on bubble surfaces and on mineral surfaces, degrading froth stability and reducing flotation kinetics.

Surface oxidation of sulfide minerals — caused by weathering at surface or in transition zones, or by oxidation during grinding in the presence of dissolved oxygen — changes the surface chemistry of the mineral and its response to collectors. Chalcopyrite with an oxidized, hydroxide-rich surface responds poorly to xanthate collectors. The mineralogy of the oxidized zone, including the distribution and type of secondary copper minerals, must be assessed when designing flotation circuits for ores with significant oxide or transitional mineralogy (Wills & Finch, 2016).

Mineral associations identified through process mineralogy data are directly relevant to flotation selectivity. If chalcopyrite grains are intimately associated with pyrite — as is common in many porphyry copper ores — then achieving a clean copper concentrate without excessive pyrite contamination requires either finer grinding to unlock the sulfides from each other, or careful reagent selection that depresses pyrite selectively while floating chalcopyrite. The degree to which pyrite is associated with chalcopyrite in locked particles, versus occurring as free pyrite particles, determines how much of the pyrite problem is a liberation issue (solvable by regrinding) versus a surface chemistry issue (requiring reagent solutions). This distinction is only accessible through quantitative mineralogy.

The intechopen review on mineralogy in flotation highlights the importance of integrating mineralogical characterization with flotation test work in a feedback loop: mineralogy explains why flotation performs as it does, and flotation results guide further mineralogical investigation to pinpoint specific causes of poor recovery or selectivity. This integrated approach is now considered best practice in process mineralogy.

Refractory Ores: When Standard Processing Fails

Not all ores yield their values to conventional mineral processing methods. Ores that resist standard flotation, gravity, or cyanidation approaches are described as refractory. Refractoriness can arise from several distinct mineralogical causes, and understanding the specific cause is essential for selecting an effective treatment strategy.

The most widespread form of refractoriness in gold processing arises when gold is physically encapsulated within sulfide mineral grains — typically pyrite or arsenopyrite — at grain sizes below the practical liberation limit of grinding. The gold may be present as tiny metallic inclusions of a few microns or less, or as sub-microscopic gold atoms substituted into the sulfide crystal lattice (invisible gold or solid solution gold). In both cases, the gold is inaccessible to cyanide leaching even after fine grinding because it is never exposed at particle surfaces. The sulfide matrix must first be destroyed by oxidation before the gold is accessible. Pre-treatment methods include pressure oxidation (POX), roasting, and bacterial (bio)oxidation, all of which selectively oxidize the sulfide matrix and liberate the gold for subsequent cyanidation. The choice among these methods depends on the ore mineralogy: the sulfide content, the arsenic content, the presence of carbonaceous material that can re-adsorb dissolved gold (double refractory ores), and the nature of the clay mineralogy (Wills & Finch, 2016).

Carbonaceous gold ores present a double form of refractoriness. The carbonaceous material in these ores — graphite, kerogen, or humic substances — not only cannot be floated efficiently but actively adsorbs gold-cyanide complexes from solution during leaching, a process known as preg-robbing. Gold dissolved by cyanide is resorbed onto carbonaceous surfaces before it can be recovered by carbon adsorption or zinc precipitation. Treatment requires either oxidation of the carbonaceous material (by roasting or oxidative leaching) or the addition of competing adsorbents (such as activated carbon or resins) to compete with the preg-robbing material for gold-cyanide complexes — a technique known as carbon-in-leach (CIL) with higher carbon ratios.

Fine mineral dissemination creates refractoriness in base metal ores by making complete liberation economically impractical. The McArthur River zinc-lead deposit is the classic example: the ore contains abundant sphalerite and galena, but grain sizes below 10 µm made conventional flotation recovery poor for decades. The development of the IsaMill, a horizontal stirred mill capable of efficiently grinding to 7 µm product size, combined with appropriate flotation reagent development, finally unlocked the deposit economically in 1995. The ore produces a bulk zinc-lead concentrate with a product size of 80% passing 7 µm (Wills & Finch, 2016).

Complex mineralogy creates refractoriness when valuable metals are distributed across multiple mineral phases with conflicting flotation responses. In some copper ores, both sulfide copper minerals (easily floated) and oxide copper minerals (poorly floated by conventional xanthate collectors) are present. The oxide minerals require sulfidization — treatment with sodium sulfide to create a pseudo-sulfide surface — before they respond to conventional flotation. In polymetallic ores containing copper, lead, zinc, and iron sulfides, achieving selective separation into individual clean concentrates is technically demanding because the minerals have overlapping flotation responses, and the selectivity achievable is limited by mineralogical factors such as the degree of intergrowth between the sulfide phases.

Process mineralogy is indispensable in diagnosing refractoriness. The combination of QEMSCAN/MLA liberation analysis with microanalytical gold deportment mapping, for example, can quantify exactly what fraction of the gold in a refractory gold ore is in solid solution, in micron-scale inclusions, in coarser grains amenable to fine grinding, and associated with carbonaceous material. This allows a rational choice of pre-treatment method and provides the data needed to model expected recovery improvements from different processing routes.