V54A-01
Sills as Piecemeal Emplacements of a Layered Intrusion: The Ferrar Magmatic System, Dry Valleys, Antarctica
It is especially difficult to discern for mafic layered systems the initial magmatic conditions giving rise to the end product. What was the style and duration of filling? What was the compositional variation and phenocryst content of the individual magmatic pulses? And, if phenocrysts are present, were they sorted prior to injection during ascension? The Ferrar dolerite magmatic system (~180 ma; > 5,000 km3) contains information fundamental to these questions. The central stack of (at least) four massive sills (~300-750 m thick) represents an episodic or serial loading of a magmatic system with compositionally and temporally contiguous aliquots of phenocryst-free tholeiitic and phenocryst-laden ultramafic magma. The same magmatic-volcanic episode gave rise to the overlying extensive Kirkpatrick basalt (e.g., Fleming et al., 1995; CMP,121;217). Whole rock compositions (e.g., CaO vs MgO) for the suite of sills shows a sweeping pattern very similar to that of Hawaii except that the context of each sample in its natural setting, as opposed to a stack of lavas, is well known. Rock composition can be related to the physical setting of the sample and the spatial setting of the contiguous rock mass. The internal structure of the ultramafic Basement Sill, the youngest and most mafic sill, reflects a long succession of filling events. It is stuffed by a vast tongue of large (1-20 mm) orthopyroxene and much smaller (< 1 mm) plagioclase; both were present in the inflowing magma, with plagioclase growing during transport. Bulk composition (7-20 MgO) and opx size and composition (En40-En90) symmetrically vary vertically through the sill, reflecting sorting during ascent. (The marginal quench zones argue against significant post emplacement re-equilibration.) The size and high concentration of opx has generally allowed little relative post- emplacement settling, but this opx matrix has acted as a sieve to the plagioclase, forming anorthosite stringers and layers pervasively and exclusively in the tongue over a vast region (~104 km3). The Opx Tongue thickness decreases radially throughout the region from Bull Pass. Everywhere within the Tongue, and only within the Tongue, are pervasive signs of crystal sorting and layering. Local ponding in a deeper recess has formed a small (~600 m thick), well-defined, layered body with many of the sorting features of large layered bodies. This is the Dais Intrusion. Rapid cooling over ~2,000 yrs. has preserved diagnostic textural relations that are annealed out in much larger layered bodies. The next highest sill, Peneplain Sill, also shows layering and the incipient development of an Opx Tongue near Bull Pass, but the other upper two sills each show no Tongue and are progressively poorer in MgO. The overall package shows strong differentiation radially and vertically. The lesson: were these same injections to form a single chambered magma, following the enumerated emplacement sequence, beginning with phenocryst-free tholeiites and concluding with heavy inputs of opx-rich ultramafics, the final result would most likely be a major layered body. (Continual stretching with intermittent injection would mimic an ocean ridge system.) Slow cooling would promote an extensive annealing to sharpen and accentuate the initial modal layering and enhance the initial cryptic layering. Initial conditions are exceedingly difficult to uncover, but their importance is fundamental to understanding the final product.
V54A-02
Compositional Convection Trumps Silicate Liquid Immiscibility (and Other Matters)
Published data for the Skaergaard (Skd) intrusion through MZ-UZc suggest that all the average cumulates are hyper-mafic and poor in felsic components relative to a cotectic condition. The "lost" felsic components presumably reside in the SHR and UBS. Reports of conjugate mafic-felsic liquids in Skd have always run afoul of the difficult extrapolation of scale from the microscopic emulsion to the megascopic separation of two liquids in the intrusion. This is a false question. When Mt+ occurs below the base of MZ, the cumulates are exceedingly dense and are nearly perfect adcumulates (shown by least plagioclase zoning: Toplis et al. ms CMP 07). The resulting rejected solute (RS) is, therefore, clearly felsic, buoyant, hot, and evolved relative to the resident bulk magma. This RS will rise freely in plumes and collect near the roof. The challenge to the 2-liquid hypothesis is, then, how to distinguish it from classical cumulate theory. From current evidence, this challenge cannot be met. Compositional convection is a well-understood and quantifiable property of freezing liquids, from the formation of sea ice to the crystallization of the inner core of the Earth and the corrosion of troctolite above thin dunite sills at Rum. Its power has been quantified to a first approximation (Morse 1986 JPet 27:1183) and shown to be appreciable. In contrast, the ability of emulsions with high variance of NBO/T to separate megascopically has never been shown. The probability of solute rejection and pluming of felsic residua is 1.0, whereas the contrary hypothesis of liquid separation has no known probability. Even if it occurred, it would be redundant in the face of assured compositional convection. The issue is moot at best, and unworthy of further exploration. Both ideas produce an evolving felsic melt near the intrusion roof, and that leads to increasing polymerization and inability to achieve adcumulus growth, so the UZ of the intrusion has greater plagioclase zoning as evidence of more orthocumulus growth. Multiphase Rayleigh fractionation and the internal reservoir of large magma chambers (Morse 2007 abstr Frontiers Min Sci, minersoc.org p. 228) may explain the lower Mg No. of sub-cotectic (i.e., scarcer) pyroxene in a Bushveld drill core: the local depletion factor overwhelms any recharge from within the magma reservoir (if present).... Other matters concern sawtooth compositional profiles in some intrusions: these represent occasional or regular involvement of the internal reservoir, alternating with isolation from it; they therefore represent magma-dynamic oscillations rather than clever (Mg-wise) external recharge events....The "M&M" packing fraction of PL+Ol troctolites may be as high as 0.8 or more, rather than the 0.6 obtained from Al tablets in water by Wager et al. (1960)....The chief and now archaic problem of layered intrusions is to refine the cooling history with time using realistic boundary conditions. A recent primary advance has been that of Holness and her group at Cambridge (JPet submitted), who demonstrate that cumulate maturation at Skd reflects aspects of the liquidus history of the magma body. In short, the enthalpy budget goes discontinuously from dominance by specific heat to latent heat at each addition of a cumulus (liquidus) crystal phase, to an isothermal end point where the enthalpy transfer to the surroundings is all latent heat and no sensible heat. These findings represent true advances in the understanding of the evolution of mafic magma bodies, of which layered intrusions are our most precious relic.
V54A-03
Cyclicity as Revealed by Stratigraphic Mineralogy and Geophysics in the Bushveld Complex, South Africa
The 65,000 km2 Bushveld Complex of South Africa has been well characterized, but most work has focused on relatively thin PGE-, Cr- and V-bearing horizons such as the Merensky Reef, UG-2 and Upper Zone magnetite layers. We have been acquiring a near-complete set of mineralogical, geochemical and geophysical data in a stratigraphic context for almost the entire 5,500 thickness of cumulate rocks in the Northern (Potgietersrus) Lobe, from 2 drill cores at Bellevue (roof, Upper Zone, Main Zone, ~3000 m) and at Moordkopjie (Main Zone, Platreef, basement, ~1500 m). Our dataset includes mineral compositions and WR major & trace elements (avg. every 10 m), magnetic susceptibility (avg. every 2 cm) and density (avg. every 2 m). The results document, for example, the broad-scale magmatic fractionation trends (An80 à An21, Mg# = 80 à 06), but there are prominent discontinuities and reversals at scales ranging from 100s to 10s of m, which probably represent injections of new magmas to actively forming cumulates. Some, but not all of these departures from continuous frationation trends are associated with similar patterns in physical properties, especially density. Of particular interest in the Main Zone, which is perceived by many as a largely homogeneous body of gabbronoritic rocks, is a prominent cyclicity on a scale of ~50-200 m, with progressive upward increases in density reflecting gradual upward increases in modal color index. Some of these cycles are associated with broad reversals in chemical fractionation trends (e.g. upward increases in Mg# of pyroxenes), arguing against simple fractionation processes. We suggest that such cycles may represent blending zones in which dense liquids and/or crystals drain downwards into the existing cumulate piles. If so, magmatic cumulates in large layered intrusions like Bushveld, and possibly smaller examples, were constructed by 10s to 100s of successive magmatic influxes. We intend to test this isotopically using mineral separates.
V54A-04
Structural and Compositional Constraints on the Emplacement of the Bushveld Complex, South Africa
Despite a plethora of petrological studies, the emplacement mechanics of the world's largest layered intrusion, the 2.06 Ga Bushveld Complex in South Africa, are still poorly understood. Early models considered the intrusion to comprise separate, lopolithic intrusions or even concentric cone sheets, but recently, overwhelming support for a sill-like intrusional form has emerged [1,2]. Examination of the contact aureole reveals three groups of emplacement related structures. Interfinger deformation zones and bridges, formed between intruding and dilating magma fingers, are preserved in both the western and eastern parts of the Complex. Two magma conduits are also known – both of these conduits are associated with strongly deformed wall-rocks and at least one of them was subjected to high magma fluxes and hosts economically significant Ni deposits. The final group of structures are diapiric domes that characterize the eastern contact aureole. These domes formed by diapiric amplification of initial interfinger deformation zones associated with the earliest mafic-ultramafic pulse of the Bushveld Complex [3,4]. The diapiric domes, interfinger deformation zones and magma conduits exhibit a strong NW-SE preferred orientation, while longitudinal conduit terminations and divergence of the conduits away from their source horizons indicates magma emplacement towards the SE. This emplacement direction is supported by numerous and varied petrological data, from bulk lithological facies variations in the layered sequence, to mineral chemical variations within specific horizons. The thickest and most chemically primitive accumulations of the lower zone of the Complex are found adjacent to the Thabazimbi-Murchison lineament (TML), a crustal scale lineament which has undergone polyphase reactivation from at least 2.7 Ga [5], and these accumulations thin towards the NW and SE away from the lineament. The Bushveld Complex was most likely fed by a feeder dyke that utilized the TML and spread laterally from the dyke-axis to form its current sill-like geometry. This implies the stress field at 2.06 Ga was suitably oriented to allow for dilation of the ENE-trending TML, and suggests the Kaapvaal craton was subject to a component of NW-SE extension. The Limpopo belt, however, records NW-SE directed transpressional collision of the Kaapvaal and Zimbabwe cratons at this time [6], suggesting that Bushveld emplacement occurred under conditions of far field extensional stress within this transpressional setting, and may add support to a subduction origin for the Bushveld Complex. References: [1] Cawthorn RG et al., S. Afr. J. Geol., 101, 291-298, 1998; [2] Kruger FJ, Econ. Geol. Res. Inst. Univ. Witwatersrand, 377, 26 p, 2004; [3] Uken R and MK Watkeys, Geology, 25, 723-726, 1997; [4] Gerya TV et al., Geology, 31, 753-756, 2003; [5] Good N and MJ de Wit, J. Geol. Soc. London, 154, 93-97, 1997; [6] Holzer L et al., J. Afr. Earth Sci., 28, 383-402, 1999.
V54A-05
Compositional Convection-Driven Differentiation in the Skaergaard Intrusion: A Reaction- Transport Model
Considerable debate has focused on the role of thermal versus compositional convection and late-stage melt and volatile migration in the differentiation of layered intrusions, including the Skaergaard Intrusion. The result of these coupled processes is a hierarchy of structures from textural re-equilibration, to mm-scale rhythmic layering, to large-scale mobilization and recrystallization involving melt and volatiles. In the Skaergaard Intrusion, there is evidence that the base of the intrusion crystallized from melts strongly enriched in iron, presumably derived from the walls and/or roof. To investigate the scenario that iron-rich melts migrated from or through the crystallizing walls and ponded on the floor, we developed a two-dimensional reaction-transport model having the projected cross-section of the intrusion. Simulations of coupled flow and reaction of melt, heat, and minerals were performed using the RCTMAG code developed by the authors. Processes include conservation of fluid mass, energy, advective and diffusive multicomponent transport, and crystallization/melting. Crystal-melt equilibria and compositions are treated using distribution coefficients based on literature values or derived from lab and/or field data. Permeability and porosity changes are coupled to crystallization and melting, with the resulting volume changes affecting flow. Simulations show that iron-rich melt develops within the sidewall mush and tends to migrate through the mush toward the base. Compositional convection dominates over thermal convection because heat loss through the walls and roof lead to crystallization and melt compositional changes, affecting density more than temperature. Chemical and thermal diffusion within the mush has subtle effects on mineral compositions and modes, primarily because water and alkalis diffuse faster than other components. The propensity for melt to migrate through the mush is clearly aided by the increase in iron and volatiles, counteracting the typical large increase in viscosity that accompanies silica-enrichment.
V54A-06
Early Silicate Liquid Immiscibility in the Skaergaard Intrusion: Evidence from high Temperature Centrifugation Experiments
Immiscible droplet textures are common in groundmass glasses and plagioclase-hosted melt inclusions of tholeiitic basalts (Philpotts, 1982). Our experiments on synthetic analogues of natural immiscible basaltic-rhyolitic glasses showed that conventional quenching experiments in 1-atm gas mixing furnaces were in most cases unable to reproduce unmixing yielding instead either turbid, opalescent glasses, or crystallization of tridymite and pyroxenes. In contrast, experiments involving in situ high-temperature centrifugation at 1000g on some of the liquids did yield macroscopic unmixing and phase separation. It appears that experimental reproduction of immiscibility in complex ferrobabsaltic aluminosilicate melts is hampered by nucleation barrier, metastable crystallization, and sluggish phase separation kinetics. Three-four hours of centrifugation were insufficient to complete phase segregation, and resulted in sub-micron immiscible emulsions in quenched glasses. For a model liquid composition of the Middle Zone of the Skaergaard intrusion obtained from experiments by Toplis and Carroll (1995) centrifugation at super-liquidus temperatures of 1110-1120 degrees C, produced a thin, silicic layer (64.5 wt.% SiO2 and 7.4 wt.% FeO) at the top of the main Fe-rich glass (46 wt.% SiO2 and 21 wt.% FeO). Transmission electron microscopy of the quenched products revealed silica-rich immiscible globules of about 20—30 nm in diameter suspended in the Fe-rich glass. The globules are however not a quench feature because they moved during centrifugation over a few millimeters of the sample length and eventually accumulated in the thin (0.2 mm) silicic liquid layer at the top. The divergent compositions of the top and at the bottom were shown in a series of static runs to crystallize very similar crystal assemblages of plagioclase, pyroxene, olivine, and Fe-Ti oxides. In light of our centrifuge experiments, immiscibility in the Skaergaard intrusion may have started already at the transition from the Lower to the Middle Zone. Thus, magma unmixing may be an important factor of the Fe enrichment documented in cumulates of the Skaergaard Layered Series. Philpotts A.R. (1982) Contrib. Mineral. Petrol. 80, 201-218. Toplis M.J. and Carroll M.R. (1995) J. Petrol. 36, 1137–1171.
V54A-07
Assimilation of Consanguineous Mafic Intrutions: Layered Crustal Sill Complexes as Reactive Filters for Continental Basalts
Continental basalts commonly display variations in their chemical compositions that are inferred to reflect fractionational crystallization (FC), recharge-FC (RFC), assimilation-FC (AFC), or recharge-AFC (RAFC). The dominance of AFC-related processes reflects the intrinsic linkage between crystallization (which releases latent heat) and assimilation (which consumes latent heat). One of the central questions in any assimilation process, however, is what exactly is being assimilated. It is commonly assumed in most AFC models for the intrusion of basalt into continental crust that the contaminant is pre-existing continental crust – that is, felsic gneiss of roughly granodioritic to tonalitic composition, which is enriched in K2O and other large ion lithophiles relative to mantle-derived basalts. These continental gneisses are commonly Precambrian in age and are enriched in the lithophilic isotope ratios 87Sr/86Sr, 207Pb/204Pb, and 208Pb/204Pb, and depleted in 143Nd/144Nd. As a result, AFC-related processes involving this ancient continental crust component typically result in basaltic lavas that are enriched in LILE (e.g., K) relative to high-field strength elements (e.g., Ti, P) and enriched in the heavy isotopes of Sr, Pb, and Nd compared to the primary or parental magma. Contrary to these expectations, basalts of the Snake River volcanic province that display chemical variations diagnostic of AFC (e.g., increasing La/Lu with decreasing mg#) are commonly characterized by essentially constant isotopic ratios of Sr, Pb and Nd, and by LILE/HFSE ratios (e.g., K/P) that decrease with decreasing mg#. We propose that these basalts assimilated a ferrogabbro derived from a parent magma that was the same or similar to the magmas being intruded to recharge the system. Melts derived from this ferrogabbro would be low in K and enriched in Fe, Ti, P, and La/Lu relative to the primitive recharge magma; the isotopic composition would be the same as the primitive recharge magma. We infer that this exchange took place within a 10 km thick mafic sill complex that has been imaged seismically at depths of 12-22 km the middle crust. We propose that this process may apply to a wide range of continental basalts.
V54A-08
Empirical Calibration of Vanadium Partitioning Between Magnetite and Ilmenite as an Oxybarometer
Vanadium is measured in magnetite and ilmenite collected in two Fe-Ti ore deposits characterized by different fO2. In the Grader ilmenite deposit (Havre-Saint-Pierre anorthosite complex, Quebec), the fO2 has been estimated at ΔNNO ca. +0.6 log units, whereas in the Fedorivka layered intrusion (Korosten anorthosite, Ukraine) a lower fO2 ranges from ΔFMQ >+0.7 log units down to -1.4 log units. After correction for subsolidus compositional re-equilibration of the magnetite, the partition coefficient of vanadium between magnetite and ilmenite D is compared to the (Eu2+/Eu3+)plag ratio calculated from the Eu distribution between pairs of plagioclase and apatite. Both parameters show good correlation, (Eu2+/Eu3+)plag varying from ca. 30 to ca. 200, and D from ca. 1 to 27. Calibrations of the two oxybarometers are proposed. Though further experiments are needed better to ascertain the relationships, comparison with independent data shows that the empirical estimates of fO2 with D are credible.