Cumulates and the evolution of layered intrusions

Introduction

The geology of layered intrusions is tightly tied in with phase relations. Most layered intrusions formed from basaltic magmas that undergo significant fractionation during their crystallisation. The crystallisation is commonly illustrated by Bowen's reaction series, which is shown below:

Bowen's reaction series is an expression of the changes that a magma will undergo by removal of the crystallising minerals (in a normal basaltic suite). A basalt will crystallise olivine and Ca-rich plagioclase at high temperature and pyroxene an lower. By removal of these minerals, the magma will gradually change its composition to an andesite (typically when 70 percent of the magma remains). Further crystallisation of pyroxene, plagioclase, and amphibole will change the composition further to a dacite, and finally to a rhyolite.
Basaltic magmas commonly show pronounced differentiation in intrusions, and strong evidence of crystal fractionation is expressed in the layered appearance of the rocks. Granites, on the other hand, are already close to their minimum crystallisation temperature, and have only minor possibilities for differentiation. Therefore it is rare to see much evidence of fractional crystallisation.

Cumulates

The crystallisation of igneous rocks can largely be described as the transition from a largely homogeneous magma to a crystalline assemblage of structured mineral phases. The crystallisation is a complex function of the magma composition, the depth of crystallisation, the cooling rate, and the availability of volatile constituents (H₂O and CO₂). Petrologists use phase diagrams to illustrate these effects, and despite the complex interactions, even simplified systems appear to match natural systems remarkably closely. Phase diagrams illustrate the compositional changes that magmas undergo as a result of magmatic differentiation processes, and as such they can be used to interpret the changes that are observed in natural rocks. You can see how the phase relations link to the formation of a cumulate succession in the animation [here].

Cumulates are simply rocks formed by crystal accumulation (Wager et al., 1960). The process of accumulation is unspecified in the original definition and may be the result of widely different processes in different rocks. The differential settling of crystals from the magma according to Stoke's law was initially regarded to be the major process, but more recent studies indicate that other processes are more significant. Different researchers favour different mechanisms according to their experience on different intrusions (or experimental work), but it is probably safe to say that the different structures cannot universally be assigned to a single process. The most prominent models promoted for the Skaergaard intrusion include:

Deposition of crystals by convective current activity and slumping of crystal-loaded suspension currents features strongly amongst the explanation of the abundant and diverse magmatic structures (Irvine, 1987; Irvine et al., 1998).
In-situ nucleation and crystal growth in a solidification front at the cumulate-magma interface has been used to explain unusual features of certain types of layering (McBirney and Noyes, 1979).
Solid-state reorganisation during cooling in the cumulate piles has been used to explain small-scale layering (e.g., Boudreau, 1987).

There is probably no one method that can universally explain all layering features in layered intrusions. Regardless of the exact mechanism of crystal-liquid separation, it is important to keep in mind that cumulates have concentrated certain constituents and excluded others from their primary magma. As such, they probably nowhere represent the original composition of the magma. In some intrusions, some of the magma froze to the walls when the magma intruded, and as a first approximation this “chilled margin” can be assumed to represent the original magma composition. For more detailed studies however, the magma compositions can only be determined by indirect means (numerical modelling, experimental). Cumulates have distinct textures that reflect their accumulative nature, but care should be taken in their interpretation, as the textures resemble other rock types formed in widely different ways (cf. McBirney and Hunter, 1995; Irvine et al., 1998). The textures constitute only a small part of the evidence that we are collecting in our detective story on the formation of the rocks.

Cumulates appear to be well-described as mixtures of two distinct mineral assemblages: The cumulus assemblage comprises the accumulated crystals that form a self-supporting framework of mineral grains in the rock. The postcumulus (or intercumulus) assemblage comprises the material that crystallised in the interstitial pore spaces between the accumulated crystals.

The cumulus material formed during normal fractional crystallisation can be recognised by forming a self-supporting interlocking framework of crystals commonly having well-developed crystal shapes and occurring in roughly constant (cotectic) mineral proportions. Once a mineral reaches its liquidus, it will be part of the fractionating assemblage throughout the remaining crystallisation – unless it gets replaced by another mineral in a peritectic reaction with the melt. The classical example of such a peritectic reaction is the change from olivine to pigeonite (olivine + SiO₂ = pigeonite) and back to olivine in the Skaergaard intrusion.

Cotectic cumulates evolve gradually towards lower temperature mineral assemblages and compositions during crystallisation, and a typical uninterrupted cumulate succession will evolve from monomineralic cumulates at high temperature to more and more complex cumulates with progressive cooling. Noncotectic cumulates occur where the crystallisation has been disrupted or in other ways is far from equilibrium (notably during magma mixing and assimilation). In such successions, the mixing may cause the formation of monomineralic cumulates – for example of chromite or plagioclase. Furthermore, magma mixing occasionally leads to the formation of intraplutonic chilled margins, thermal erosion and reaction features, and reversals to higher temperature cumulus mineral assemblages and compositions. Noncotectic cumulates can also form during more extensive periods from stratified magmas, where different magma layers are out of equilibrium (for example if one magma layer has plagioclase and another chromite on the liquidus).

The dominant processes acting during the postcumulus stage are the crystallisation of the intercumulus magma and the compaction of the cumulus pile. Crystallisation partly forms overgrowth on the primary cumulus crystals (adcumulus growth), and partly discrete intercumulus minerals. Compaction occurs in response to the increasing pressure acting on the individual crystals as more material is accumulated. This leads to a rotation of cumulus crystals, local pressure dissolution (resorption), and a filter pressing of the intercumulus liquid. The efficiency of these processes governs to a large extent the texture of the rocks and leads to a classification of cumulates according to the amount of discrete intercumulus material: Adcumulates have 0-7%; mesocumulates 7-25%; and orthocumulates more than 25% discrete intercumulus material (Irvine, 1982). In rare cases, compaction and adcumulus growth can completely expel the remaining intercumulus liquid, with the result that no discrete intercumulus minerals occur.

Various reactions and readjustments occur in the rocks during cooling at the postmagmatic (subsolidus) stage. At high temperatures (greater than 450-500 ºC), diffusion may occur within and between mineral phases causing modifications of the mineral compositions and parageneses to lower temperature conditions. Notably in the Skaergaard intrusion, this involves the formation of ilmenite from titanomagnetite; the inversion of pigeonite to an intergrowth of orthopyroxene and augite; the formation of orthopyroxene lamellae in augite; and the homogenisation (elimination of compositional zoning) of olivine and pyroxenes.

The evolution of layered intrusions

Layered intrusions display evidence of complex crystallisation and cooling histories that are reflected in their diverse field relations. Most characteristic are their often well developed stratification that closely resembles layering in sedimentary rocks. The processes that the rocks undergo at different stages of their evolution will roughly follow the history below.

The cumulus stage is dominated by the processes of crystal fractionation and accumulation. The accumulated rocks commonly display well developed modal and/or textural stratification that in many cases resemble that seen in clastic sedimentary rocks. Cross bedding structures, unconformities, graded bedding, slumping, and the occurrence of xenoliths and conglomerate beds indicate that most intrusions hosted a dynamic depositional environment; and the action of various types of magmatic currents has traditionally been implied. However, layering occurs in a variety of very different styles and scales, and it is likely that a number of very different processes were involved in their formation. The layering in the Marginal series of the Skaergaard intrusion, for example, has very different characteristics than the layering in the Layered series, which again is very different from the thin “inch-scale” layering in the Stillwater complex, Montana (Hess, 1960).

Assimilation preferentially incorporates low-melting constituents of the host rocks and can have a large effect on the crystallisation. Assimiation occurs primarily during the early stages of differentiation and decreases with time as the magma chamber gets isolated from the host rocks by crystallisation and crystal accumulation. During fractional crystallisation, the effect of assimilation will appear very similar to the magmatic evolution because low melting constituents are concentrated in the residual magma. The effects of assimilation can be difficult to recognise in the field, but locally partly digested xenoliths are exposed in the marginal parts of the intrusions. The extent of assimilation in a layered intrusion cannot be assessed without detailed geochemical and isotope geochemical studies.

Marked reversals in the cumulus mineral assemblages and compositions are mostly interpreted as the consequence of magma mixing. The reversals usually signify a return to higher temperature conditions – which is most satisfactory explained by a new batch of hot magma entering the magma chamber. Large layered intrusions typically display several episodes of magma mixing; for example, the Eastern Layered Series of the Rum complex displays 15 successive stratigraphic units separated by distinct reversals in cumulate mineral assemblages and compositions. These reversals are commonly associated with small amounts of noncotectic cumulates (Wager, 1968), and locally resorption structures are exposed at the contacts (cf., Irvine, 1983).

Magma unmixing (liquid immiscibility) may be a consequence of several processes. Most notably, a local oversaturation in sulphur is a common consequence of magma mixing or assimilation, which results in the formation of immiscible sulphide (Fe-Cu-Ni-S-O) magmas. Such magmas provide efficient collectors of Ni, Cu, Co, and the platinum-group elements and are therefore important for the formation of mineral deposits. Another possible situation is if convection isn’t vigorous enough to sustain a well mixed magma, fractional crystallisation or assimilation may lead to magma stratification with separately convecting layers of different density.
Volatile unmixing could potentially lead to changes in the cotectic mineral proportions, which may change the crystallisation slightly. Volatile unmixing is most likely to occur in shallow magma chambers, where it may generate an overpressure in the magma chamber and lead to the fracturing of the host rocks. The fracturing leads to the formation of dikes (regional dike swarms and cone sheets) and perhaps eruptions if the fractures reach the Earth’s surface. In the cumulate succession, an episode of volatile unmixing may lead to a shift towards more plagioclase rich cumulates.

The postcumulus stage involves primarily compaction and the crystallisation of the intercumulus liquid. Compaction occurs to a variable extent as a result of the accumulation of cumulus crystals. The compaction leads to a filter pressing of the intercumulus liquid, the rotation of cumulus crystals to form a planar lamination (or enhancing a pre-existing fabric), local deformation, fracturing, and pressure dissolution (resorption). Whereas crystallisation at the cumulus stage involves large amounts of magma, and consequently forms unzoned cumulus crystals, crystallisation of the interstitial magma occurs in a confined space from small amounts of magma. Therefore, the postcumulus overgrowth develops strong compositional zoning. Discrete intercumulus minerals reach their liquidus at different times, and the resulting texture depends on differences in the time of nucleation, the crystal growth rates and grain boundary energies. In many rocks “crystallisation successions” can be made out from the textural relations, but care should be taken in their interpretation, as they can be influenced by grain-boundary adjustments. If the cooling is slow, grain-boundary adjustments may lead to extensive recrystallisation, by which crystals in favourable orientations grow preferentially at the expence of crystals in less favourable orientations. This leads to a "coarsening" of the cumulate texture.

At the subsolidus stage, cooling leads to the formation of contraction fractures that can be invaded by hydrothermal fluids. These fluids can alter the original rocks in a variety of different ways. An addition of fluids lowers the solidus of the rocks, and at high temperature local remelting may occur. This can result in the formation of pegmatites and a variety of replacement phenomena. Below the minimum melting temperature, the interactions are governed by elemental interchange between the rocks and the fluids. This may lead to localised breakdown of the primary minerals (feldspars to mica and clays; olivine and pyroxene to hornblende, serpentine and talc) and a chemical leaching of certain elements (notably the alkali metals, chlorine, and sulphur).