For a layered intrusion whose magma is supposedly sourced from a mantle plume, it is plausible to assume that parental melts directly derived from within the plume are of basaltic to komatiitic composition Campbell et al. This range reflects to a large extent mineral accumulation and crystal fractionation during igneous differentiation of isotopically light olivine. The large degree of melting within a mantle plume e.
One component in the Windimurra Upper Zone, being directly sourced from the mantle, can thus be constrained to a primitive mantle component Sossi et al. In addition, progressive crystallization of the Middle Zone will concentrate these residual, isotopically heavy liquids toward the top of the intrusion through compaction; compression of the cumulate mass is likely to trigger filter pressing McBirney, ; Mathez et al.
There, it forms interstitial liquid or a melt pool within a crystal mush. The origin of the extreme Hf isotope signature of this melt is uncertain, but is not unique and may be related to similar sources of extreme Hf isotopes observed in komatiites elsewhere cf.
Nebel et al. Pulses of hot and buoyant replenishing melt from the mantle plume source will preferentially intrude into these existing crystal mush zones in an evolving Middle Zone magma chamber. Mixing of liquids-in-residence was probably filter-pressed and pooled from underlying sequences in an evolving Middle Zone heavy in Fe isotopes with a radiogenic Hf isotope signature in a homogeneous Hf-Fe isotope melt pond, yet with various antecrysts and possibly also partly layered.
Given the sheer volume of melt required to form the Upper Zone, represented here by ca. Inevitably, however, crystals from the previous mush zone would be present in this blend of melts. It remains unclear as to what extent a melt injection can disturb existing layering, but the perturbation within the melt pool is expected to be severe with respect to temperature and viscosity difference between new melt and liquid-in-residence.
If magma cools below its liquidus temperature, rheologic properties change significantly in an interlocking crystal network Philpotts and Carroll, However, reheating through magmatic recharge can cause remobilization of crystals Irvine et al.
We thus propose that the replenishment forged a new Upper Zone in the intrusion through a massive inflation in volume consequent to a melt surge. Despite the apparent difference in isotope compositions, both the replenished and residential liquids will be although not isotopically but compositionally similar , yet not identical. It is expected that the residual liquids have similar Si but higher Fe concentrations following a tholeiitic fractionation trend, whereas newly injected melts are higher in MgO with a possible komatiitic composition.
The new melt blend must thus be in disequilibrium with existing crystals that formed from earlier liquids, causing some remelting. Overgrowth of crystals and zonation is likely common during subsequent cooling. This would lead to moderate chemical and strong isotope heterogeneity in individual minerals, a feature observed for Fe isotopes in some Chen et al.
In the mixing scenario envisaged here, Fe isotope systematics of mineral pairs or within minerals are thus not useful to decipher the history of LMI petrogenesis, or would require high resolution isotopic analyses that is currently not achievable.
Strong and non-systematic variability is expected if melts are very different in Fe isotope compositions, as observed in the Beiba intrusion Chen et al. If they are identical, no variation is observed at all, as is the case for the Upper Zone of the Bushveld Complex Bilenker et al. Because blends of melts are expected to have a homogeneous isotope signature, the only way this can occur is through isolation of melt batches or through freezing of Hf isotopes in antecrysts from a previous melt batch.
Isolating melts is difficult to reconcile with a turbulent melt injection, so we consider the latter scenario more plausible. Indeed, Nebel et al. It is argued here that this melt batch is the former Middle Zone. In the proposed scenario of melt injection into an existing crystal mush zone, the systematic increase in initial Hf isotopes toward the roof Nebel et al. This initial Middle Zone crystallization is referred to here as stage 1 crystallization and is illustrated in Figure 5.
Stage 2 crystallization in the newly established Upper Zone will then follow with overgrowth on existing grains and growth of new minerals with a different Hf-Fe isotope composition.
Scatter is expected in any trend resulting out of this mix, as interstitial liquid will have Hf isotope signatures of the new melt mix, whereas Fe isotopes can be further subject to fractionation through crystal formation, two independent factors captured in a single analysis.
Figure 5. Schematic sketch of the temporal two-stage evolution of the Upper Zone magma-reservoir. Prior to melt replenishment, a melt pool, possibly with initial layering, was established in the roof zone of the now Middle Zone of the intrusion, which formed the top part of the intrusion. During melt replenishment, a perturbation in the magmatic pool allowed melt and crystal re-organization.
During this second stage, the new Upper Zone formed, which constitutes a mixture of newly injected melt and liquid-in-residence and its crystal cargo. Crystals that form during this process would be in disequilibrium with any hybrid melt pool.
A single parental liquid did not exist. Crystal settling in the chamber must have occurred through convection and a density contrast within the magma chamber, with the majority of the crystal cargo accumulated at the roof of the intrusion. It remains unclear why melt with the lowest crystal fraction is concentrated toward the base of the intrusion, but possibly because of an injections into an already existing, dense melt pond.
Notable is that the lower density of komatiite liquid compared to existing crystals requires a physical process stopping the liquid to progress further towards the roof. A second effect can be imperfect and local limited perturbation of existing crystals during replenishment toward the roof, such that the chamber is filled and expanded like a balloon through an existing melt pond at the base of the newly established Upper Zone.
The roof section is less affected by the melt injections. Melt infiltration, however, must have been sufficient to infiltrate all layers, albeit to different degrees, and most likely followed by rapid establishment of rhythmic layering in individual layers Bons et al. The Hf isotopes demand that these layers have formed from a blend of melts, the replenished liquid and the liquid-in-residence.
At least some Fe-Ti oxides must thus postdate the replenishment, and were only then accumulated close to the base of the newly formed Upper Zone. These either form instantly and with this crystal chemical control on Fe isotopes overrides the Fe isotope signatures of the liquids incorporating preferentially heavy isotopes, Shahar et al. Alternatively, a redox-reaction forced instant magnetite saturation of the liquid-in-residence with later magnetite formation being dominated by Fe of replenished liquid, and with this a lighter Fe isotope composition.
An aspect that needs to be accounted for in our study is, however, that we report whole rock analyses. The Hf isotope composition in these rocks, composed of almost entirely magnetite, can then be dominated by interstitial liquid, which is compatible with regular cumulate behavior, whilst the Fe is driven by magnetite. The Hf-Fe isotope co-variation in the magnetitites is then only an artifact of two independent factors, combined in a whole rock analysis.
The scenario promoted here can explain a range of geochemical oddities in the Upper Zone of the Windimurra Igneous Complex and possibly other layered intrusions. Cryptic melt replenishment with incomplete mixing and variable melt-crystal ratios may be responsible for: 1 apparent chemical disequilibrium among crystals, as to overgrowth, remelting and formation of a second generation of crystals from a different magma batch; and 2 , cryptic isotope layering among phases, which relate to magma replenishment of melts that appear similar in major elements but are distinct in their trace isotope systematics.
Notable is that this is the case for major and trace elements. This sequence of events can explain reported Fe isotope disequilibrium among some mineral separates Chen et al. For a cyclic layering to develop, however, smaller zones must have subsequently developed into density horizons with possible internal small-scale convection. This process of layering remains enigmatic and is unrelated to the replenishment, but likely continues as proposed for other layered intrusions e.
Petrologic investigations of individual mineral phases may, in general, fail in tracking the replenishment process, because of the mix of crystals and melts in disequilibrium, perturbation of crystals, coupled with subsequent convective reorganizations of different generations of crystals. This will likely result in large variations both chemically and isotopically on a mineral scale that, in addition to subsequent layering, create a pseudo-systematic agglomeration of minerals with only some chemical memory of their turbulent past.
Whilst mineral studies and whole rock analyses are often complementary, individual minerals would have been affected by various stages of the mixing process and to different degrees, so may only show a snap shot of what is a complex mechanism of recharge, mixing and physical sorting. Under these circumstances, and counterintuitive to most recent studies, whole rock composition possibly yield a more detailed picture of the history of the intrusion. Noteworthy is, though, that this requires a special sequence of intrusive events and mantle sources, as is the case for Windimurra.
In the context of melt replenishment into magma chambers of any kind, a two-component mixing as is observed here may well, at least in parts, explain the widely observed Hf isotope heterogeneity in zircons in plutonic bodies. ON and RA designed the study and the model. All authors commented on the manuscript. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The Melbourne TIE team is acknowledged for support. Maximus Resources is kindly acknowledged for providing core material and significant contribution to an associated ARC linkage project. Reviewers are kindly acknowledged for their comments and MA for his editorial handling.
Bergantz, G. Open-system dynamics and mixing in magma mushes. Berger, J. Deformation-driven differentiation during in situ crystallization of the 2 center dot 7 Ga iguilid mafic intrusion West African Craton, Mauritania. Bilenker, L. Iron isotopic evolution during fractional crystallization of the uppermost bushveld complex layered mafic intrusion.
Bons, P. Layered intrusions and traffic jams. Geology 43, 71— Campbell, I. Melting in an Archaean mantle plume: head its basalts, tail its komatiites. Nature , — Cawthorn, R. Layered intrusions. Amsterdam: Elsevier. Google Scholar. Variations in Cr-content of magnetite from the upper zone of the Bushveld complex - evidence for heterogeneity and convection currents from magma chambers. Earth Planet. Charlier, B. Compositional and kinetic controls on liquid immiscibility in ferrobasalt—rhyolite volcanic and plutonic series.
Acta , 79— Chen, L. Iron isotope fractionation during crystallization and sub-solidus re-equilibration: constraints from the Baima mafic layered intrusion, SW China. Magnesium isotopic evidence for chemical disequilibrium among cumulus minerals in layered mafic intrusion. Cheng, T. Methodsx 1, — Czamanske, G. This makes it difficult to determine if simple shear flow or pure shear compaction is dominant. However, if pure shear is important then it must be accompanied by significant pressure-solution to produce the strength of the observed foliations.
Correlation of fabric strength and overall crystal shape suggest that growth and shear were concurrent. But by far the most important textural process is equilibration coarsening , testified by crystal size distributions.
Almost all rocks in layered mafic intrusions have equilibrated, but the degree of progress towards equilibrium is very variable, even within layered mafic intrusions. It appears that the process is controlled by local variations in liquid content, mineral mixture and temperature.
The thermal history of igneous rocks is encoded in their microstructures via the control by cooling rate on crystal growth. If temperatures remain close to the liquidus, with small undercoolings, growth rates are low and microstructures are strongly affected by constraints imposed by interfacial energies.
The minimisation of interfacial energy results in the melt topology becoming a function of dihedral angle and porosity. Further effects include the loss of the smallest crystals Ostwald ripening , though this process is restricted to very small grainsizes. For large crystals, interfacial energy controls the microstructure by determining the shapes, but not the size or number of crystals.
In mafic-ultramafic magmas, minerals with a larger structural anisotropy e. Microstructures in partially solidified material from lava lakes and entrained glassy enclaves suggest that interfacial energies play only a minor role in the supersolidus of rapidly-cooled or coarse-grained rocks. They are most important for slowly-cooled bodies of hot magma crystallizing as crystallographically simple minerals, especially adcumulates in layered complexes.
Interfacial energies may also play a role in the sub-solidus, leading to the development of a granular microstructure in some adcumulates. This is, however, only important for very fine-grained rocks such as chill zones, or for monomineralic rocks. Elsewhere there is little or no evidence for significant grain growth in the sub-solidus, demonstrating that the end-stage of microstructural equilibration generally is not reached in crustal layered intrusions.
More and more evidence for the development of silicate liquid immiscibility during cooling of magmas in layered intrusions have been presented. Here, we review some theoretical principles with a focus on the separation of two silicate melts, i. We discuss the role of melt structure and present phase equilibria relevant to stable and metastable immiscibility. The understanding of immiscibility in magmas has strongly benefited from recent progress in experimental approaches.
Kinetics studies evidence the importance of nucleation barriers in producing unmixing, coarsening and potential separation of equilibrium melts.
Improvement of analytical tools has also enabled detailed study of major and trace element partitioning. The study of immiscible emulsion in volcanic rocks also brings important information on the evolution of plutonic systems and on the potential formation of compositional gap along liquid lines of descent.
We then present the most recent evidence for immiscibility in some major layered intrusions, i. Paired melts are identified as contrasted melt inclusions trapped in apatite and their segregation can be responsible for the formation of Fe—Ti—P-rich rocks.
We finally discuss more broadly the potential effect of immiscibility in interstitial melt and the implications on the evolution of the crystal mush. Fractional crystallization of ponded basaltic magma results in rocks that become increasingly more evolved in composition inwards. However, almost all mafic sills and layered intrusions show basal reversals tens to hundreds of meters thick in which minerals and rocks become compositionally more primitive upwards.
Two main types of basal reversals can be distinguished with different relationships with the overlying Layered Series. Most common are fully-developed reversals that gradually pass into the overlying Layered Series through a crossover horizon with the most primitive rock and mineral compositions. Above the crossover, the Layered Series becomes progressively more evolved. This type of reversal presents a mirror image of the overlying Layered Series. In contrast, aborted reversals are separated from a more primitive or evolved Layered Series by a sharp textural and compositional break.
Both types of reversal are attributed to three principal processes: 1 inflowing magma becomes progressively more primitive, 2 magma undercooling gradually decreases and 3 adcumulus growth becomes increasingly more effective. The major difference between these types is that fully-developed reversals result from a continuous evolution and therefore show a gradual transition into the overlying Layered Series.
In contrast, aborted reversals form when the evolution is abruptly interrupted by a pulse of primitive magma that resets the crystallization history in the chamber.
This generally results in basal reversals in sharp contact with a more primitive Layered Series. In some lopolithic layered intrusions, aborted basal reversals are discordant with the Layered Series and therefore can be in contact with more evolved rocks along the upper sections of the inward-dipping floor. Anisotropy of magnetic susceptibility AMS has been recognised as a well-established fabric analysis tool for intrusive igneous rocks since the s.
The AMS technique provides directional information for magnetic foliation and magnetic lineation fabric components of the AMS ellipsoid, potentially coupled with a quantification of the overall fabric strength and geometry.
The magnetic susceptibility and therefore the AMS of igneous rocks is often dominated by ferromagnetic mineral phases such as magnetite or low-Ti titanomagnetite, even where present in very minor amounts e.
Fe-bearing silicates exhibit subordinate paramagnetic behaviour but are volumetrically much more important constituents of igneous rocks than Fe-Ti oxides, so may also contribute considerably to the AMS. A significant application of AMS is in the characterisation, constraint and quantification of very weak or subtle mineral fabrics related to flow or tectonic deformation.
In particular, studies of magnetic fabrics in sheet intrusions and in granite plutons have enormously enhanced our understanding of the magma flow regimes and emplacement kinematics in these settings. Studies of AMS in layered mafic-ultramafic intrusions have been comparatively sparse. A wide array of rock magnetic and complementary quantitative fabric analysis techniques can be employed to support an AMS dataset in this regard.
With studies of layered mafic-ultramafic intrusions currently proceeding at unprecedented micro- scales of textural and geochemical detail, AMS offers petrologists a unique approach to investigating the microstructure of cumulates and the textural complexity they exhibit. Stevenson, Craig Magee. The interstitial melt in partially molten cumulate piles in layered intrusions must at some point reach saturation with a volatile phase such as water vapor or hydrosaline melt.
A number of models have been proposed in which orthomagmatic fluids migrate through partially solidified cumulates and participate in the formation of ore deposits. Here I examine the topology of the crystal—melt—vapor system in a cumulate and relate this to the role of capillary forces in governing the size and mobility of individual vapor bubbles.
Capillary forces will play a dominant role in setting the number density and sizes of bubbles. In any cumulate rock with crystals smaller than several cm in diameter, bubbles of the postcumulus aqueous phase will be unable to migrate away from their sites of nucleation and growth. The major deposits of platinum-group elements PGE of the Earth are associated with ultramafic and mafic igneous rocks. The bulk of current PGE production is extracted from narrow stratiform horizons referred to as reefs located in the lower to central portions of large layered intrusions and is dominated by the Bushveld Complex in South Africa.
The PGE-mineralized horizons occur in most case as laterally continuous and uniform layers that can extend over hundreds of kilometres along strike. Technological advances are driving the current and future state-of-the-art in the study of layered intrusions and, looking forward, it is clear that these bodies will continue to inspire and challenge our understanding of magmatic systems and magma solidification for many years to come.
The Skaergaard Intrusion of East Greenland is the quintessential example of low-pressure closed-system fractionation of basaltic magma. Of particular significance is the lack of consensus about the microstructural record and the mechanisms by which interstitial liquid is expelled from solidifying crystal mushy zones.
Skaergaard remains a cradle for new insights into igneous processes, with recent work highlighting the importance of separation of immiscible liquids on magma evolution. Millimeter—centimeter thick layers of chromite-rich rock chromitites are rare, but ubiquitous, features of the Bushveld South Africa and Rum Scotland layered intrusions. Despite their meager dimensions, the chromitites provide insight into processes that modify igneous layering and, in the Bushveld, the formation of the platinum-group element—rich Merensky Reef.
The Merensky Reef chromitites represent reaction zones formed in a compositional gradient between hydrous silicate melt and a crystalline cumulate assemblage, analogous to reaction zones in metamorphic systems. At Rum, the chromitites formed at the melting front between newly injected magma and the magma chamber floor, an analogous process but one driven by thermal, rather than chemical, energy. Major-element zoning in plagioclase is best explained by trapped liquid in the pore spaces between cumulus crystals, which is a result of the complex interplay between the rate of crystal growth and the cooling rate.
Isotopic zoning in feldspars likely reflects crystal growth in a magma that is becoming, or has become, isotopically contaminated through wall rock partial melting and assimilation processes.
Mineral-scale isotopic zoning, such as detected in plagioclase, can be used to infer the cooling rates of layered intrusions. Rock textures provide a key to deciphering the physical processes by which gabbro forms in mafic intrusions. Developments in both direct optical and crystallographic methods, as well as indirect magnetic fabric measurements, promise significant advances in understanding gabbroic textures.
Here, we illustrate how bulk magnetic fabric data, particularly from intrusions with sparse silicate-hosted magnetite, may be used to extend direct crystallographic observations from thin sections. We also present a scheme for characterizing crystallographic foliation and lineation and use this to suggest that the strength of gabbro plagioclase foliations and lineations varies significantly with geodynamic environment.
Several of the bodies contain world-class ore deposits, notably the Kemi chromium deposit and the Pechenga nickel deposits.
Other deposits include nickel and copper at Kevitsa, Kotalahti and Sakatti; vanadium at Koillismaa; and platinum-group elements at Portimo and Penikat.
0コメント