Rosamond J. Kinzler

Primary magmas of mid-ocean ridge basalts 2. Applications

Variable initial mantle composition and extent of depletion during dynamic melting processes strongly influence compositions of primary basaltic magmas. The descriptions of the equilibria that pertain to melting in the upper oceanic mantle presented in the companion paper (Kinzler and Grove, this issue) are used to estimate the major element compositions and temperatures of aggregate primary magmas of mid-ocean ridge basalt (MORB) generated in the adiabatically upwelling mantle beneath oceanic spreading centers. Primary MORB magmas with high Na2O abundances that are produced from more fertile mantle compositions or represent initial melts of a depleted spinel-lherzolite have higher SiO2 and Al2O3 and lower MgO, FeO, and CaO abundances, relative to low-Na2O primary magmas. Na2O abundance variation in the mantle source during polybaric, near-fractional melting processes causes melt compositions to vary significantly. The total extents of depletion achieved by the decompression melting process to yield the observed variation in major elements of MORB range from ∼ 6 to 18%; the range of mean pressures of melting is relatively narrow, 8–15 kbar; the total range modeled for the adiabatic, near-fractional melting process is 4 to 25 kbar. Aggregate primary magmas of MORB are not picritic, nor do they resemble sampled primitive MORB (MORB with MgO > 9.0 wt%). Much of the variation in major element composition observed in sampled MORB can be explained by melting a similar depleted MORB-mantle source. The ambient temperature range of the upper mantle beneath the global ridge system required to explain the observed chemical variations is 1475°–1315°C.

Primary magmas of mid‐ocean ridge basalts 1. Experiments and methods

This paper reports experiments carried out between 9 and 16 kbar (0.9–1.6 GPa) using natural, primitive mid-ocean ridge basalt compositions and synthetic analogs of mid-ocean ridge basalts to investigate the effects of pressure, temperature, and variable bulk composition on the composition of melts multiply saturated with the minerals present in the upper oceanic mantle: olivine, orthopyroxene, augite, and plagioclase or spinel. For this low-variance, five-phase assemblage, equations involving pressure, melt NaK # ([Na2O+K2O]/[Na2O+K2O+CaO]; weight ratio), melt Mg # (Mg/[Mg+Fe2+]; total iron as Fe2+), and weight percent TiO2 in the melt predict temperature and major element compositions of magmas produced by melting spinel and plagioclase lherzolites at upper mantle pressures. The equations are estimated using a selected set of data from this experimental study and published experimental studies that report compositions of glasses coexisting with olivine, low-Ca pyroxene, augite, and plagioclase and/or spinel. An experimental test of a liquid compositionally similar to melts produced in a subset of peridotite-basalt sandwich experiments is presented. The composition tested was reported as multiply saturated (with orthopyroxene + augite + spinel + olivine) in the sandwich experiment, but it does not crystallize these phases at the conditions of the experiment. We exclude this liquid and the subset it represents (data from Fujii and Scarfe [1985] and Falloon and Green [1987]) from the data set used to constrain the melting equilibria. With estimates of the melt NaK #, melt Mg #, and weight percent TiO2 of the melt the quantitative descriptions of the melting equilibria can be used to predict the temperatures and major element compositions of melts from lherzolite. Methods are described for estimating these compositional parameters with the nonmodal batch melting equation (for Na2O, CaO, K2O, and TiO2) and a mass balance calculation (for MgO and FeO) from the initial composition and phase proportions of the mantle source, the amount of melt produced and the nature of the melting process, and the stoichiometric coefficients of the mantle melting reaction.

Coupled CaAl-NaSi diffusion in plagioclase feldspar: Experiments and applications to cooling rate speedometry

Abstract The rate of CaAl-NaSi interdiffusion in plagioclase feldspar was determined under 1 atm anhydrous conditions over the temperature range 1400° to 1000°C in calcic plagioclase (An80−81) by homogenizing coherent exsolution lamellae. The dependence of the average interdiffusion coefficient on temperature is given by the expression: D = 10.99 ( cm 2 /sec) exp (−123.4( kcal/mol )/RT), (T in °K). This value is for diffusion perpendicular to the (03 1) interface of the lamellae. CaAl-NaSi interdiffusion is 4 to 5 orders of magnitude slower than oxygen diffusion in the temperature range 1400° to 1200°C and possibly 10 orders of magnitude slower at subsolidus temperatures. The large differences in diffusion rates explain the apparent contradiction posed by the plagioclases of large layered intrusions (e.g., the Skaergaard), which retain delicate Ca, Na compositional zoning profiles on the micron scale, but have undergone complete oxygen isotopic exchange with heated meteoric groundwater from the surrounding wall rocks. CaAl-NaSi diffusion is slow, the closure temperature is high (within the solidus-liquidus interval), and Ca-Na zoning is preserved. Oxygen diffusion is faster, the closure temperature is lower (350°-400°C) and the feldspars exchange oxygen with the low-temperature hydrothermal fluids. The complex micron-scale oscillatory zones in plagioclase can also be used as cooling rate speedometers for volcanic and plutonic plagioclase. Cooling histories typical of large mafic intrusions (e.g. the Stillwater) are slow, begin at high initial temperatures (1200°C) and result in homogenization of oscillatory zones on the scale of 10 microns. The oscillatory zones found in the plagioclase of granodioritic plutons are preserved because cooling is initiated at a lower temperature (1000°C) limiting diffusion to submicron length scales despite the slow cooling rate of the intrusion.

High pressure phase relations of primitive high-alumina basalts from Medicine Lake volcano, northern California

Anhydrous phase relations were determined at 1 atm and 10 to 15 kbar for primitive high-alumina basalts (79–35g and 82–72f) from Giant Crater at Medicine Lake volcano. These compositions are multiply saturated with olivine+augite+plagioclase+spinel+/-orthopyroxene near the liquidus at about 11 kbar. Experiments on mixtures of sample 79–35g with orthopyroxene and olivine determined the location of the multiple saturation boundaries where liquid coexists with the assemblage olivine+augite+orthopyroxene+plagioclase at 10 kbar and olivine+augite+orthopyroxene+spinel at 15 kbar. The mix experiments showed that primitive Medicine Lake high alumina basalts (HABs) are close in composition to liquids in equilibrium with a mantle lherzolite source containing olivine+augite+ orthopyroxene+spinel+plagioclase at 11 kbar. Orthopyroxene observed as a near liquidus phase in an 11 kbar experiment on sample 82–72f supports this conclusion. The most primitive HABs from Medicine Lake are low in K2O (0.07 wt.%), high in MgO (>10 wt.%) and Ni (231 ppm), and have light-rare earth element depletions and large ion lithophile element enrichments. A model for the origin of these near-primary high-alumina basalts is that they are partial melts of a MORB-like mantle lherzolite source that has been enriched by a fluid component derived from the subducted slab. The HAB magma segregated from its mantle residue just below the base of the crust near the crust-mantle boundary.

An experimental study on the effect of temperature and melt composition on the partitioning of nickel between olivine and silicate melt

Abstract Experiments in the simple system CaO-MgO-Al2O3-SiO2-Na2O-FeO were carried out to investigate the control of temperature and melt composition on the partitioning of nickel between olivine and silicate melt ( D oliv/liq Ni ). Eleven experiments determine the influence of changing forsterite (Fo) content on D oliv/liq Ni in this simple system. The equation of Hart and Davis (1978) that accounts for the variation of D oliv/liq Ni in terms of MgO content of the silicate liquid is tested using experimental data from ironbearing simple and natural systems and found to be inadequate to explain the observed variation of D oliv/liq Ni ( ±39% relative average error of prediction). Two different equations are formulated to describe the partitioning behavior of Ni between olivine and silicate melt. The first is similar to that of Hart and Davis (1978) and uses an expression for Ni-Mg exchange between olivine and silicate melt. The second uses an expression for the Ni-olivine formation reaction. The Ni-Mg exchange equation for D oliv/liq Ni depends on the forsterite content of the olivine and the mole fraction MgOliq, and predicts the experimentally determined values within ±13% relative average error. The Ni-olivine formation reaction equation for D oliv/liq Ni depends on temperature, mole fraction SiOliq2, and melt compositional terms that arise from a symmetric, binary, Margules formulation of the activity coefficients for NiOliq and SiOliq2. This equation predicts the experimentally determined values within ±9% relative average error.

Assimilation of granite by basaltic magma at Burnt Lava flow, Medicine Lake volcano, northern California: Decoupling of heat and mass transfer

At Medicine Lake volcano, California, andesite of the Holocene Burnt Lava flow has been produced by fractional crystallization of parental high alumina basalt (HAB) accompanied by assimilation of granitic crustal material. Burnt Lava contains inclusions of quenched HAB liquid, a potential parent magma of the andesite, highly melted granitic crustal xenoliths, and xenocryst assemblages which provide a record of the fractional crystallization and crustal assimilation process. Samples of granitic crustal material occur as xenoliths in other Holocene and Pleistocene lavas, and these xenoliths are used to constrain geochemical models of the assimilation process.A large amount of assimilation accompanied fractional crystallization to produce the contaminated Burnt lava andesites. Models which assume that assimilation and fractionation occurred simultaneously estimate the ratio of assimilation to fractional crystallization (R) to be >1 and best fits to all geochemical data are at an R value of 1.35 at F=0.68. Petrologic evidence, however, indicates that the assimilation process did not involve continuous addition of granitic crust as fractionation occurred. Instead, heat and mass transfer were separated in space and time. During the assimilation process, HAB magma underwent large amounts of fractional crystallization which was not accompanied by significant amounts of assimilation. This fractionation process supplied heat to melt granitic crust. The models proposed to explain the contamination process involve fractionation, replenishment by parental HAB, and mixing of evolved and parental magmas with melted granitic crust.

Corrections and further discussion of the primary magmas of mid-ocean ridge basalts, 1 and 2

An error has been detected in the methodology for estimating batch melt compositions from plagioclase and spinel Iherzolites at mantle conditions presented by Kinzler and Grove (1992a,b) (hereafter referred to as K&G). In a mass balance calculation employed by K&G the melt fraction was not properly transformed from weight units to molar units, resulting in an overestimate of the effective melt fraction employed in the calculation of the Mg # (Mg/[Mg+Fe2+]) of the melt, at melt fractions greater than 0.01. In the batch melts estimated by K&G, the abundance of FeO increases then decreases in the melt with increasing melt fraction from 0.01 to 0.20. Using the corrected methodology described here, the abundance of FeO in melts produced via batch melting increases as melt fraction increases from 0.01 to 0.20 (up to the point of exhaustion of clinopyroxene). The correction has essentially no effect on the compositions of melts produced via the near-fractional melting processes described in K&G because melt fractions employed in the near-fractional melting calculations never exceeded 0.01. In addition we apply the methodology of K&G to estimate the mantle temperature, average pressure of melting and average extent of melting beneath a mid-ocean ridge segment from the FeO and Na2O content of a mid-ocean ridge basalt from that segment.

Origin of compositional zonation (high‐alumina basalt to basaltic andesite) in the Giant Crater Lava Field, Medicine Lake Volcano, northern California

The Giant Crater lava field erupted high-alumina basalt (HAB), basalt, and basaltic andesite (BA) lavas about 10,500 years ago on the south flank of Medicine Lake volcano. Lava flows from several adjacent vents constitute a compositionally zoned eruption of ≈4.4 km3. The earliest lavas are the most evolved (53.5 wt % SiO2 and 6.5 wt % MgO), and latest lavas are the most primitive (48 wt % SiO2 and 10.5 wt % MgO). The evidence from major and trace element chemistry, isotopes and mineral chemistry in lavas and inclusions indicates that fractional crystallization and assimilation of a granitic crustal component played important roles in producing the observed compositional variations. Models that involve fractionation of primitive HAB, assimilation of granitic crust, replenishment of the magma reservoir by primitive HAB and mixing of these components (FARM) are presented for the development of the compositionally zoned eruption. In these models, extensive fractionation of HAB and melting of silicic crust occurs with little or no chemical interaction between fractionated basalt and melted crust. The melted granitic crust is produced by heat supplied through fractionation of large volumes of basaltic magma. In an initial event, melted crust and highly fractionated basalt (ferrobasalt) mix with primitive HAB magma. Subsequent injections of primitive HAB mix with the ferrobasalt and the magma produced by the previous mixing event, and this sequence of mixing, reinjection, and mixing leads to the compositional zoning of the Giant Crater lava field.