Phosphorus Zoning in Olivine: A new Source of Information on the Early Magmatic Histories of Igneous Rocks
We observed P zoning in olivine (ol) from a wide range of igneous rocks: terrestrial basalts, andesites, dacites, and komatiites; martian meteorites; lunar basalts; and ordinary chondrites (OCs). P contents vary from below the detection limit (∼0.003 wt%) to 0.1-0.2 wt% over a few microns, with no correlated variations in Fo content. Zoning patterns include P-rich cores with skeletal, hopper, or euhedral shapes; oscillatory zoning; structures suggesting replacement of P-rich ol by P-poor ol; and sector zoning. Crystallization experiments on basaltic compositions at cooling rates of 1-30 deg-C/h reproduce many of these features. We infer that P-rich zones in experimental and natural ol reflect incorporation of P in excess of equilibrium partitioning during rapid growth and that zoning patterns primarily record crystal-growth-rate variations. High-P cores may reflect pulses of rapid crystal growth following delayed nucleation due to undercooling. Oscillatory zoning in P generally reflects internal factors at the crystal-melt interface (i.e., oscillating growth rates that occur without external forcings), but some P zoning in natural ol may reflect external forcings (e.g., magma mixing, eruption) resulting in varying crystal growth rates and/or P contents in the magma. In experimental and some natural ol, Al, Cr, and P contents are generally positively correlated, but in natural phenocrysts, Cr zoning is usually less intense than P zoning and Al zoning is weak to absent. We propose that magmatic ol grows with correlated zoning in P, Cr, and Al superimposed on normal zoning in Fe/Mg; rapidly diffusing 2+ cations homogenize during residence in hot magma; slower diffusing Al and Cr3+ only partially homogenize; and delicate P zoning is preserved because P diffuses at a much lower rate. This interpretation is consistent with the preservation of zoning in P, Cr, Al, and Fe/Mg in ol with short residence times at high T (e.g., experimentally grown ol, komatiitic ol, groundmass ol, and rims of ol phenocrysts grown during eruption). Preservation of P zoning in ol from type 6 OCs and a Bushveld apatite-ol-bearing gabbro is consistent with a very low diffusivity of P in ol. Melt inclusions in ol are usually spatially associated with regions of high-P ol, but the ol in direct contact with melt inclusions is usually low in P. Moreover, low-P ol zones directly surrounding melt inclusions often crosscut (and appear to replace) high-P features (e.g., in one case a chevron-shaped wedge of low-P ol surrounding a melt inclusion crosscuts high-P, oscillatory zoned ol). Low-P regions directly around melt inclusions vary from <10 to many 10s of microns across. In some cases, these low-P regions probably formed by slow crystallization on the inclusion wall. In many cases, however, we infer that melt inclusions were trapped by rapidly crystallized high-P ol, but that the enclosing high-P ol was then dissolved by the melt inclusion, followed by slow reprecipitation of the inclusion walls as low-P ol. This is consistent with melt inclusions from the Galapagos that vary by a factor of ∼5 in P but are essentially identical in trace element patterns. P zoning is widespread in magmatic ol, revealing complex details of growth and intra-crystal stratigraphy in what otherwise appear to be relatively featureless crystals. Since it is preserved in early-formed ol with prolonged residence times in magmas at high temperatures, P zoning has promise as an archive of information about an otherwise largely inaccessible stage of a magma's history.
Bubble Growth in Lunar Basalts
Although Moon is usually said to be volatile-"free", lunar basalts are often vesicular with mm-size bubbles. The vesicular nature of the lunar basalts suggests that they contained some initial gas concentration. A recent publication estimated volatile concentrations in lunar basalts (Saal et al. 2008). This report investigates bubble growth on Moon and compares with that on Earth. Under conditions relevant to lunar basalts, bubble growth in a finite melt shell (i.e., growth of multiple regularly-spaced bubbles) is calculated following Proussevitch and Sahagian (1998) and Liu and Zhang (2000). Initial H2O content of 700 ppm (Saal et al. 2008) or lower is used and the effect of other volatiles (such as carbon dioxide, halogens, and sulfur) is ignored. H2O solubility at low pressures (Liu et al. 2005), concentration-dependent diffusivity in basalt (Zhang and Stolper 1991), and lunar basalt viscosity (Murase and McBirney 1970) are used. Because lunar atmospheric pressure is essentially zero, the confining pressure on bubbles is completely supplied by the overlying magma. Due to low H2O content in lunar basaltic melt (700 ppm H2O corresponds to a saturation pressure of 75 kPa), H2O bubbles only grow in the upper 16 m of a basalt flow or lake. A depth of 20 mm corresponds to a confining pressure of 100 Pa. Hence, vesicular lunar rocks come from very shallow depth. Some findings from the modeling are as follows. (a) Due to low confining pressure as well as low viscosity, even though volatile concentration is very low, bubble growth rate is extremely high, much higher than typical bubble growth rates in terrestrial melts. Hence, mm-size bubbles in lunar basalts are not strange. (b) Because the pertinent pressures are so low, bubble pressure due to surface tension plays a main role in lunar bubble growth, contrary to terrestrial cases. (c) Time scale to reach equilibrium bubble size increases as the confining pressure increases. References: (1) Liu Y, Zhang YX (2000) Earth Planet. Sci. Lett. 181, 251. (2) Liu Y, Zhang YX, Behrens H (2005) J. Volcanol. Geotherm. Res. 143, 219. (3) Murase T, McBirney A (1970) Science 167, 1491. (4) Proussevitch AA, Sahagian DL (1998) J. Geophys. Res. 103, 18223. (5) Saal AE, Hauri EH, Cascio ML, et al. (2008) Nature 454, 192. (6) Zhang YX, Stolper EM (1991) Nature 351, 306.
Partial Melts in the Seismic Low Velocity Zone
The possibility that small amounts of CO2± H2O rich magma could plausibly be stable in the seismic low veloxity zone (LVZ) continues to be debated 40 years since it was first propos,with some arguing that the LVZ arises from the properties of solid peridotite close to its solidus and others advocating hydrous silicate or carbonatitic partial melts as the responsible agents . Experiments defining the partitioning of H2O between peridotitic minerals and silicate melts demonstrate that H2O is modestly incompatible during partial melting of the mantle. Consequently, near- solidus partial melts of mantle with normal sub-oceanic H2O concentrations (100 ppm) are not particularly H2O-rich, ranging from ∼1.25 wt. % at 100 km to ∼2.5 wt.% at 200 km. These small concentrations are insufficient to incite partial melting in the LVZ beneath mature oceanic lithosphere without a significant influence of CO2. On the other hand, carbonate is effectively incompatible in peridotite once the stability of magnesite is exceeded at ∼300 km, and therefore carbon-rich melts may be stable throughout the LVZ. However, these melts may be carbonatitic only in the deepest (>150 km) and oldest (>60 Ma for a cooling half-space model, older for a plate model) portions of the LVZ. In the younger shallower portions of the LVZ, carbonatite will react with peridotite to form hydrous carbonated silicate melts with compositions ranging from melilitite to alkali basalt as the volatiles become diluted in the shallowest and youngest parts of the LVZ. Thus, partial melts are thermodynamically stable throughout the LVZ and their composition varies spatially. However, this does not resolve the debate as to whether the LVZ is caused by small amounts of retained partial melt. First, the fractions of partial melt generated from normal sub-oceanic mantle are very small (<0.02% carbonatite and <0.1% carbonated silicate, except at depths >80 km and lithospheric ages <20 Ma) and may not be sufficient to account for the seismic properties of the LVZ. Second, thermodynamic stability is not the same as dynamic stability, as the buoyant melts may be extracted rapidly from the LVZ by compaction. .
Lunar Magmatic Volatiles
Samples returned from the Apollo Missions prompted a variety of experimental investigations (e.g., [1-4]) which form the basis of our current understanding of lunar compositional evolution. The observed low abundances of solidus temperature-suppressing volatiles justified volatile-free experiments. However, the low-pressure nature of the samples makes it unlikely that volatiles were retained during magma ascent and eruption. In an effort to re-assess the lunar mantle volatile budget, we are focusing on the mineral apatite because of its incorporation of F, Cl, and OH as essential structural constituents and its greater ability to retain such volatiles relative to melt. Apatite grains analyzed from magnesian- and alkali-suite rocks (14161,7111, 14161,7269 and 14161,7264), KREEPy impact melt rocks associated with magnesian- and alkali-suite rocks (14161,7233; 14161,7110; 14161,7062; 12033,634-25; SaU 169-4), and mare basalts (79195; 12037,224; 74246; 12023,147,1; 10084; LAP 02205; LAP 03632; NWA 2977) by electron microprobe using the technique of [5,6] show two distinct compositional groups. Apatite from the mare basalts analyzed are primarily mixtures of fluor- "missing component" (OH?) apatite with low Cl abundance, while that from the magnesian- and alkali-suite rocks are fluor-chlor mixtures. Apatite/basaltic melt partition coefficients for F, Cl, and H2O from the data of  provide first estimates of magmatic volatile abundances in lunar magmas. They suggest that magmatic water may have been more abundant than F and Cl at the stage of apatite crystallization in mare basalts. In contrast, at this stage, the magmas that produced the Mg-and alkali suite minerals were F- and Cl-dominated. These results have wide-reaching implications regarding the chemical and physical evolution of the Moon and therefore, the next generation of experimental investigations.  Walker et al. 1973 EPSL 20, 325-336.  Walker et al. 1975 GCA 39, 1219-1235.  Longhi 1992 GCA 69, 1275-1286.  Longhi 2003 JGR 108, E8, doi:10.1029/2002JE001941.  Stormer et al. 1993 Am Min 78, 641.  McCubbin et al. 20081st NLSI Conference.  Mathez and Webster 2005 GCA 69, 1275-1286.
The volatile contents and D/H ratios of the Apollo 15 lunar volcanic glasses
The notion that highly volatile elements, especially hydrogen, were completely evaporated away during the catastrophic heating events that formed the Moon has changed. New evidences of indigenous water in the primitive lunar volcanic glasses indicate the presence of a deep source within the Moon relatively rich in volatiles1. Here we report new volatile data (C, H2O, F, S, Cl) for over 200 individual Apollo 15 lunar glasses with composition ranging from very low to high Ti contents (sample 15427,41; 15426,138; 15426,32). Our new SIMS detection limits (∼0.15 ppm C; ∼0.4 ppm H2O, ∼0.05 ppm F, ∼0.21 ppm S, ∼0.04 ppm Cl by weight determined by the repeated analysis of synthetic forsterite located on each sample mount), represent at least 2 orders of magnitude improvement over previous analytical techniques. After background correction the volatile contents are 0-0.14 ± 0.13 ppm for C suggesting values just above background; 0-70 ppm for H2O; 1.6-60 ppm for F; 58-885 ppm for S; and 0-3 ppm for Cl . Our new values represent an increase in the volatile concentrations by a factor of 2 from previously reported data1. Two outstanding features of the data are the significant correlation among H2O, Cl, F and S contents, and the clear relationship between the volatile and the major element contents of the glasses. D/H ratios measured in the lunar glasses range from +700‰ to +5400‰ and are inversely correlated with water contents. Part of the D enrichment may result from in-situ spallation from interactions with solar and galactic cosmic rays. At the lowest H2O contents, spallogenic D can potentially account for half of the abundance of D in the interiors of the glass beads based on estimated D production rates2, but the production rates would have to be underestimated by a factor of 100 to account for the D/H ratios of glasses with high H2O. The D/H ratios of magmatic water in H2O-rich glasses are thus undisputably fractionated from terrestrial values. The data are consistent with kinetic fractionation of D from H during post- eruptive degassing, from a pre-eruptive H2O-D/H composition similar to terrestrial basalts, provided that H and D diffuse as protons/deuterons. Alternatively, the high D/H ratio of the water in the glasses could be inherited from gas condensed within the Moon from a residual atmosphere surrounding the proto-lunar disk after the giant impact. 1. Saal et al. (2008) Nature 454, 192-195. 2. Merlivat et al. (1976) LPSC v. 7, 649-658
Origin of the Lunar Magnesian Suite Cumulates
Calculation of chemical evolution during lunar magma ocean (LMO) crystallization strongly suggests that the highly magnesian olivines with relatively low Ni concentrations that are characteristic of the magnesian suite cumulates ultimately derive from the earliest dunite cumulates of the magma ocean. However, the evolution from pure olivine deposited at the base of the magma ocean to olivine-plagioclase cumulates (magnesian suite troctolites) crystallized within the crust is complex. New melting calculations on various combinations of rock types formed in the lunar magma ocean suggest 2 alternative scenarios. In the first, overturn of the LMO cumulate pile brings the hottest cumulates (dunite) to the base of an anorthositic crust where mechanical mixing plus radioactive heating provided by the late stage LMO residual liquid (KREEP) induce extensive melting. The melts are suitable parents of the magnesian suite troctolites and norites. In the second scenario no anorthositic crust forms, rather norite and later gabbronorite accumulate at the base of the ocean once plagioclase becomes stable. Overturn of the cumulate pile brings hot dunite to the base of the zone of mafic cumulates where melting and mixing ensue. Low-degree, hybrid melts of a mixture of primary dunite, norite, and late-stage MO liquid have the necessary Ni, Co, Mg', Al2O3, and REE to crystallize the distinctive lithologies of the magnesian suite. The KREEP signature in magnesian suite cumulates, which includes depletion of Ti relative to REE, implies crystallization of the LMO beyond 99%. Thermal models show that a thick KREEP layer beneath an insulating crust extends the duration of the LMO beyond the crystallization age of the oldest magnesian suite cumulates. This result plus the sparsity of anorthosite in the region of crustal KREEP enrichment (the PKT or Procellarum KREEP Terrane) argues for lack of formation of anorthosite and crystallization of the LMO to shallow levels (5 km or less) in the PKT, thus favoring the second scenario. A major concern is whether the sparsity of anorthosite in the PKT was caused by a local phenomenon such as a convection head or whether it was a Moon-wide feature, implying post-magma ocean formation of ferroan anorthosites.
Source Variations Along the EPR Identify Melt flow and Influence Segmentation
Understanding the melting processes and the relation between the source variations and the melting process is crucial in understanding the sub-ridge processes. We have analyzed at high "density" samples from the EPR between 8-18N for trace elements and isotopes. At the EPR we observe a systematic variation in the chemical composition of the basalts related to ridge discontinuities, both at fracture zones and at overlapping spreading centers. The variations in the chemistry are of two types: 1. There is a discontinuity in composition across a fracture zone or overlapper. This discrete jump in composition can be identified in both the trace element ratios as well as the isotopic compositions. 2. The chemical variations in each individual ridge segment indicates two component mixing. However, the two components differ from segment to segment. The first type of variation can be explained by low degrees melts traveling across the ridge continuity. At migrating ridges such as the EPR leading (LE) and trailing edges (TE) of ridge segments have been identified. LEs have thicker crust suggesting a larger accumulation of melt. The low degree melts generated of-axis on the TE of the ridge segment can find a shorter route to the ridge by crossing the transform fault plane. The LE therefore has additional low-degree melts which are missing at the TE. The area on the EPR we covered contains four fracture zones (Siqueros, Clipperton, Orozco and 18N) as well as three overlapping spreading centers. We observe discontinuities in the chemical composition of the basalts at all eight ridge discontinuity. The changes in the trace element ratios like Ce/Yb, Ba/La, Sm/Nd at six of the seven discontinuities are consistent with the LE receiving a larger amount of low degree melt, as predicted by the geophysical model. The Clipperton Fracture zone is the only discontinuity that has chemical variations that are the reverse of what is expected based on the model. Secondly, and perhaps most importantly, the chemical variations within the individual ridge segments are distinct from each other. Each individual segment represents two component mixing, but the components change from segment to segment. The coincidence of chemical discontinuities with ridge discontinuities is consistent for all eight ridge segments and indicates that the segmentation is influenced by the mantle composition. It is hypothesized that changes in mantle source composition, like variations in solidus or mineralogy, can potentially result in changes in the melting regime like depth of melting and melt rate that causes enough stresses to locate the ridge discontinuities at places where the discontinuities n the mantkle composition are the largest.