Isotopic Petrology: The Curious Case of the Shergottite Meteorites
The shergottites comprise a diverse suite of martian basalts and basaltic cumulates. As of 1985, there were three proposed igneous ages for this group of basaltic rocks: (i) 4-4.5 b.y. [1; Caltech]; (ii) 1.3 b.y. [2; JSC]; and (iii) 360 m.y. [3; Mainz]. At that time I proffered that petrographic observations demanded that the shergottites were only 180 m.y. old . By 1985, all the above geochronology groups had presented evidence of a young 200 m.y. age, but interpreted that as a metamorphic resetting. My observation was considered extremely controversial. However, John Longhi was instrumental, perhaps pivotal, in causing this new, controversial interpretation to be accepted, at least among petrologists. John then used this newfound knowledge to infer the Nd isotopic composition of the martian crust . This new interpretation of shergottite chronology has led to petrologic insights that would not otherwise have been possible: (I) There were melt extraction events in the shergottite mantle immediately(?) preceding shergottite formation; and (II) the variation in enriched vs. depleted characteristics of the shergottites is best explained by assimilation of ancient, enriched crust by young magmas from a depleted source region. I. Internal, mineral isochrons of the shergottites (15 years later) vary from 165 m.y. to 575 m.y. . Without exception, the Sm/Nd ratios of the shergottites themselves are larger than the time-integrated Sm/Nd ratio of their source regions . This means that there has been a LREE-enriched phase that has fractionated from the shergottites. There are no solid phases in the martian mantle that are capable of this. This implies that LREE-enriched magmas escaped the shergottite source regions just prior to shergottite petrogenesis. II. Therefore, the shergottites can be characterized in terms of three Sm-Nd components: (i) a primitive shergottite magma from a depleted source region; (ii) an enriched crust; and (iii) a missing LREE-enriched melt. Interestingly, the 180 m.y. shergottites require only two Sm-Nd components, because they fall along a two-component mixing line. This implies that these shergottites were derived from a single magma, that was generated from a single, LREE-melt-extracted source region, which subsequently assimilated various amounts of enriched crust. Otherwise, the amount of assimilated crust for each shergottite would be required to be linked to the amount of missing, LREE-enriched magma, in such a way that a two-component linear array is generated. Since LREE melt extraction had to occur before depleted-shergottite petrogenesis and since crustal assimilation must have occurred after, these two physical processes have little chance of being coupled. In addition, this linear shergottite mixing array must be of sufficient quality that it could initially be interpreted as an isochron . This interpretation of 180 m.y. shergottite petrogenesis reinforces Longhi's inference of the Nd isotopic composition of the martian crust , at least in one particular martian terrain. Bottom line: In order to understand martian chronology, you have to understand martian petrology. And in order to understand martian petrology, you have to understand martian chronology. This is an aphorism that (I think) John would endorse.  Chen and Wassserburg (1986) Geochim. Cosmochim. Acta 50, 955-968.  Shih et al. (1982 Geochim. Cosmochim. Acta 46, 2323-2344.  Jagoutz and Wanke (1986) Geochim. Cosmochim. Acta 50, 939-953.  Jones (1986) Geochim. Cosmochim. Acta 50, 969-977.  Longhi (1991) Proc. Lunar Planet. Sci. Conf. 21st, 695-709.  Nyquist et al. (2001) In Chronology and Evolution of Mars 96, pp. 105-164.  Borg (2003) Geochim. Cosmochim. Acta 67, 3519-3536.
Two Parent Magmas for the Same Anorthosite Pluton? The Egersund-Ogna Case
The Egersund-Ogna massif (Rogaland anorthositic province, Norway) is a typical diapir with andesine anorthosite (AA) in the central part and a deformed leuconoritic margin (LM) with labradoritic plagioclase. Orthopyroxene is the only mafic mineral in the centre and the margin, except for local occurrence of olivine Fo70 in the margin. Both the AA and LM contain typical high-alumina orthopyroxene megacrysts (HAOM) coexisting with plagioclase megacrysts. A new series of HAOM analyses shed new light on the petrogenesis of the intrusion as formely developed (Duchesne et al. 1985 NATO ASI C158, 449-476; Longhi et al. 1999, J. Petrol. 40, 339-362): (1) HAOM in AA display wide ranges in Al contents and mg# (Al2O3 from 2.5% to 8.5%; mg# from 58 to 79) and thus crystallized continuously from 11-13 kbar to 3-5 kbar. (2) The range of mg# in HAOM from the LM is similar to that in AA. Al2O3 clusters between 3-6% but some values extend up to 7.8%. The LM magma has thus started crystallizing at the same pressure than AA. (3) HAOM from AA and LM differ by their MnO contents which are higher by 0.05% in LM at equivalent mg#. Cr in HAOM can reach values as high as 1400 ppm in AA (thus >900 ppm as previously stated) and as high as 2200 ppm in LM (thus higher than previous values of 1500 ppm). (4) At equivalent mg# and Al content of the associated HAOM, plagioclase megacrysts in LM are ca. 10% richer in An than in AA. The Sr contents is also quite distinct (AA: 900 ± 50 ppm; LM: 450 ± 25 ppm). This strongly suggests that AA and LM have crystallized from two different magmas. (5) The HAOM trace element variation is controlled by fractional crystallization, and modelling of Zn, Co, Ni, V, Mn and Ti leads to a well-constrained value for the plagioclase proportion in the plag-opx cotectic relation (Xplag = 0.685) for liquid fraction ranging from 1.0 to 0.7. The partition coefficients of these elements between opx and melt do not vary with pressure. (6) Sr and An variations in plagioclase mainly reflect the pressure effect on the plag/melt DCaO and DSr rather than fractionation processes. (7) The LM most primitive assemblage can have crystallized at 11-13 kbar in a high-alumina basalt similar to the HLCA melt (Fram & Longhi, 1992 Am. Miner. 77, 605-616). (8) It has been suggested (Longhi et al., 1999 op. cit.) that the AA parent magma was generally similar to a primitive jotunite but slightly less evolved. However, the extended range of Cr contents in HAOM (up to 2200 ppm) and the high mg# provided by the new data are more compatible with a high-alumina basalt similar to HLCA. (9) Thus AA and LM parent magmas are high-alumina basalts similar in mg#, different in Cr and Mn, and quite distinct in Sr and Ca contents. (10) We concur with the former conclusion that both magmas have wollastonite content lower than other basaltic rocks and lie astride the thermal maximum on the opx+cpx+plag liquidus boundary. This requires melting of plag-bearing mafic sources yielding melts different in Sr, Ca, Cr and Mn contents. (11) The intrusion emplaced in two pulses. The first magma (LM) started crystallizing at depth (Al-rich HAOM) and intruded a magma chamber at intermediate depth (HAOM with intermediate Al contents) where it produced a plagioclase mush, which eventually intruded the final magma chamber. A second magma (AA) also started crystallizing HAOM and plagioclase megacrysts at depth and continued to crystallize HAOM and plagioclase en route to the final chamber (range of Al contents in HAOM). The AA diapir followed the same conduit as the first one and intruded the LM magma chamber to deform it by ballooning. The Egersund-Ogna massif is thus a nested-diapir.
Sveconorwegian Massif Type Anorthosites (Rogaland Complex) and Penecontemporaneous Granitoids Suites Result From Postcollisional Remelting of Subduction-Related Underplated Mafic Cumulates
Two distinct magmatic suites have been identified in the postcollisional setting of the Sveconorwegian orogen, the Anorthosite-Mangerite-Charnockite (AMC) suite of the Rogaland complex (0.93 Ga) located in the westernmost part of the orogen and the hornblende- and biotite-bearing granitoids (HBG) suite (0.97-0.94 Ga) outcropping all over southern Norway east of a Sveconorwegian opx-in isograd. New geochronological data (U- Pb on zircon) confirm previous age estimates for the two suites. The least differentiated mafic facies of the two suites, supposed to represent the parent magmas, have similar geochemistries but with some significant differences : higher CaO and Mg # (MgO/(MgO+FeOt)) and lower K2O in the HBG suite. Moreover, the mafic facies of the two suites have very similar initial Sr, Nd and Pb isotopic compositions but their differentiation is affected by two distinct crustal contaminants. The crustal contaminant involved in the AMC suite has high initial 87Sr/86Sr, slightly negative εNdt and highly radiogenic lead isotopic compositions. The crustal contaminant involved in the HBG suite has low initial 87Sr/86Sr, strongly negative εNdt and less radiogenic lead isotopic composition. This suggests that the AMC and HBG suites were emplaced in two different crustal segments which were already part of a single piece of continent at the onset of the Sveconorwegian orogeny. The initial Sr, Nd and Pb isotopic compositions of the mafic facies of the AMC and HBG suites are very close to the initial Sr, Nd and Pb isotopic composition of the mafic facies of subduction-related augen gneisses, the Feda suite (1.05 Ga). Additionally, experimental data indicate that plausible parent magmas of massif type anorthosites (AMC suite) lie on thermal high in relevant phase diagrams and thus cannot be produced by fractionation from mantle melts but by melting of gabbronoritic sources in the lower crust (Longhi et al., 1999 ; Longhi, 2005). This conclusion can be extented to the mafic facies of the HBG suite on the basis of similar geochemistry. Combining these isotopic and experimental constraints, we suggest that the parent magmas of the HBG and AMC suites were produced by remelting of mafic cumulates produced by crystallization of mafic magmas which were underplated at the base of the lower crust when the the Feda suite was emplaced. The different geochemistries of the mafic facies of the HBG and AMC suites can be partly explained by modifying the lower crustal mafic cumulates during the high grade Sveconorwegian metamorphism (1.035- 0.97 Ga) which took place between the emplacement of the Feda suite (1.05 Ga) and of the postcollisional suites (0.97-0.93 Ga). The granulitic conditions occurring west of the opx-in isograd have induced dehydration of the mafic cumulates and decreased their Mg # (MgO/(MgO+FeOt)).
High Pressure Crystallization of Mafic Magma: Field Observations, Compositional Measurements and Computer Modeling
The Cortlandt Complex is a small early Silurian composite, mafic to ultramafic, anorogenic deep crustal pluton about 60 km N of New York City in which most rocks in the six mapped plutonic phases have resulted from either fractionation or contamination or both. Bender et al (AJS-1984) estimated Cortlandt parental composition as an alkalic gabbro based on the nature of early plutons. The youngest and easternmost pluton consists largely of concentrically layered pyroxenites and olivine pyroxenites (with subequal modal proportions of opx and cpx, and only minor ol) and it appears to have had a different parental magma. Samples collected through a series of layers reflecting a few hundred meters of stratigraphy in layered pyroxenites indicate wide variation in F/FM (0.18 to 0.3) and both Al and Ti contents of pyroxenes are unusually high (cpx - Al2O3 from 6-7 wt percent and TiO2 from 1.0 - 1.5 wt percent; opx - Al2O3 from 3.8 - 5.7 wt percent and TiO2 from 0.2 to 0.7 wt percent). All pyroxenes show significant exsolution of ilm lamellae. Crystallization pressure has been well constrained by thermobarometry of metapelites in the thermal aureole at roughly 0.8 GPa, making this a very unusual example of very high P cumulate formation. Several magma compositions were tested as suitable parents by running computer simulations using MELTS (Ghiorso and Sack, 1995, CMP; Asimow and Ghiorso, 1998, Am. Min.). These MELTS runs quickly eliminated the proposed alkalic gabbro parent - it did not crystallize opx at any P. In this preliminary modeling, the most suitable parental magma for the cumulates was a picritic Karoo basalt (SiO2 - 46.9, TiO2 - 1.6, Al2O3 - 9.3, FeO - 12.2, MgO - 15.9, CaO - 9.1; Na2O - 1.3, K2O - 0.6, P2O5 - 0.2). MELTS runs at 8 kbar and FMQ showed a close approximation to both mineral proportions and mineral chemistry of the Cortlandt samples. The initial liquidus phase was opx at 1421C, ol at 1407C (L 4.4percent crystallized), spl at 1301C (L 19.7 percent crystallized), cpx at 1275C (L 21.2 percent crystallized). This MELTS run terminated at 1035C (16 percent L remaining) with no feldspar. Relative proportions predicted at 1230C by MELTS were L - 53.8 percent; opx - 22.7 percent; cpx - 19.1 percent; ol - 4 percent. At 1200C: L - 43.5 percent; opx - 25.2 percent; cpx - 26.8 percent; ol - 3.9 percent. At 1170C: L - 34.1 percent; opx - 29.3 percent; cpx - 32.3 percent; ol - 2.8 percent. Al2O3 and TiO2 contents of both opx and cpx predicted by MELTS in the T range of 1230C - 1170C agreed very well with measured compositions of Cortlandt pyroxenes. F/FM ratios predicted by MELTS were slightly higher than those measured (with ferrous/ferric ratios calculated based on stoichiometry) but this minor discrepancy can likely be corrected by minor adjustment of the modeled composition. Interestingly, mineral thermometry in UHT metapelite xenoliths within the layered pyroxenite sequence yield equilibration T of 1175 - 1200C. Abrupt swings in the F/FM, Al2O3 and TiO2 of pyroxenes may represent injections of fresh magma into already formed cumulate layers. I had earlier speculated that the residual liquid from formation of these fractionally crystallized cumulates might be of anorthositic or gabbroic anorthosite character. The composition of liquid remaining at 1200C (ca. 43percent of original L) as predicted by MELTS was normatively feldspathic, but of more troctolitic character (pl - 46percent, or - 8percent, di - 14percent, ol - 21percent, ne - 3percent) although this normative mineralogy could potentially reflect modal mineralogy of a hornblende gabbro or diorite, both of which rock types crop out in the Cortlandt Complex Pluton 6.
Source Mineralogy for Hawaiian Tholeiites
Hawaiian tholeiites commonly have been thought to be the melting product of garnet lherzolite. However, Sobolev, et al. (2005, Nature, 434, 590) proposed a pyroxenite source. This idea is based on their claim that melting of garnet lherzolite at P > 3 GPa would yield magmas with low SiO2 (< 47%), a feature that is not characteristic of Hawaiian tholeiites. Phase relations for the CaO-MgO-Al2O3- SiO2 (CMAS) system show that melt at the lherzolite solidus with the lowest SiO2 at any pressure occurs just at the solidus transition from spinel lherzolite to garnet lherzolite (3 GPa), and that solidus melts become progressively richer in SiO2 as pressure either decreases (spinel lherzolite field) or increases (garnet lherzolite field) from 3 GPa. Absolute values of melt compositions in CMAS do not precisely reproduce natural compositions, but the trends hold. As pressure increases from 3 to 6 GPa (the high-pressure limit of the experimental data), model basalt liquid compositions at the garnet lherzolite solidus increase significantly in both MgO (22.1 to 30.8 wt. %) and SiO2 (47.3 to 51.6 wt. %). This is consistent with the characteristically high SiO2 of Hawaiian tholeiites and strongly supports a source with a garnet lherzolite mineralogy, olivine + orthopyroxene + clinopyroxene + garnet. This result holds for a wide range of lherzolitic to pyroxenitic bulk source compositions, as long as the bulk compositional variations are not so extreme that one or more of these minerals is lost. Phase relations in the CMAS tetrahedron at 3 GPa (Milholland and Presnall, 1998, J. Petrol., 39, 3-27) show that an eclogitic mineralogy produces melts that are enriched in SiO2 and low in MgO - very different from the least-fractionated Kilauea glasses. Glass compositions at the least- fractionated end of the olivine-controlled trend for the Puna Ridge at Kilauea (Clague, D. A., et al., 1995, J. Petrol., 36, 299-346) plot very close to the experimental determination of the initial melt composition for a CMAS model garnet lherzolite mineralogy at ∼ 4-5 GPa. This shows that some Hawaiian melts are extracted from a depth of ∼150 km with almost no crystallization during passage to the surface.
Melt Generation in Heterogeneous Mantle Sources: A Three-Legged Stool Approach
The compositions and volumes of basalts generated by adiabatic decompression melting are primarily a function of three factors: mantle potential temperature, the style of mantle upwelling, and source composition. Attempts to use basalts to infer the relative importance of these three factors in specific localities are made difficult because even for homogeneous mantle sources, basalts are aggregates of melts generated over a range of pressures and temperatures within the melting regime. When source heterogeneity and differences in the melting behavior of source lithologies are accounted for, the complexity of relating basalts to the conditions of melt generation increases substantially. Advances in our understanding of mid - ocean ridge basalt petrogenesis have demonstrated the utility of creating geochemical models for melt generation that are constrained by experimental petrology [e.g. 1]. To better relate basalt compositions to the melting processes within a heterogeneous mantle source, we have developed a forward polybaric melting model that simulates the melting of a source comprised of pyroxenite and peridotite. The model uses thermodynamically - derived polybaric melting functions based on parameterizations of pyroxenite and peridotite melting [2, 3]. The model takes into account mantle potential temperature, style of mantle upwelling and variable amounts of pyroxenite, and outputs the isotopic and trace element compositions and volumes of pooled melts using the residual mantle column method . We propagate uncertainties in model input parameters to assess robustness and compare our results with previous models [5-7]. We apply our model to ocean island and large igneous province environments to constrain potential temperature, upwelling rate and abundance of pyroxenite in the mantle source from observed basalt compositions and volumes.  Longhi 2002, G-cubed, doi:10.1029/2001/GC000204;  Katz et al. 2003, G-cubed, doi:10.1029/2002GC000433;  Pertermann and Hirschmann 2003, JGR, v 108, doi:10.1029/2000JB000118 ;  Plank and Langmuir 1992, JGR, v 97, p 19749-119770;  Ito and Mahoney 2005, EPSL, v 230, p 29-46;  Sobolev et al 2005, Nature, v. 434, p 590-597;  Stracke and Bourdon 2008, GCA, v. 73, p 218-238.
Longhi Games, Internal Reservoirs, and Cumulate Porosity
Fe in plagioclase at an early age, T-rollers (or not) on the Di-Trid boundary in Fo-Di-Sil, the mantle solidus, origins of anorthosites, esoteric uses of Schreinemakers rules and many more topics are all fresh and pleasant memories of John Longhi's prolific and creative work. The Fram-Longhi experimental effect of pressure on plagioclase partitioning with liquid in mafic rocks became essential to an understanding of multiphase Rayleigh fractionation of plagioclase in big layered intrusions. Only by using the pressure effect could I find a good equation through the data for the Kiglapait intrusion, and that result among others required the existence with probability 1.0 of an internal reservoir (Morse, JPet 2008). Knowledge of cumulate porosity is a crucial key to the understanding of layered igneous rocks. We seek both the initial (inverse packing fraction) and residual porosity to find the time and process path from sedimentation to solidification. In the Kiglapait Lower Zone we have a robust estimate of mean residual porosity from the modes of the excluded phases augite, oxides, sulfide, and apatite. To this we apply the maximum variance of plagioclase composition (the An range) to find an algorithm that extends through the Upper Zone and to other intrusions. Of great importance is that all these measurements were made in grain mounts concentrated from typically about 200 g of core or hand specimen, hence the represented sample volume is thousands of times greater than for a thin section. The resulting distribution and scatter of the An range is novel and remarkable. It is V-shaped in the logarithmic representation of stratigraphic height, running from about 20 mole % at both ends (base to top of the Layered Series) to near-zero at 99 PCS. The intercept of the porosity-An range relation gives An range = 3.5 % at zero residual porosity. Petrographic analysis reveals that for PCS less than 95 and greater than 99.9, the An range is intrinsic, i.e. pre-cumulus, for values less than 6 %. Hence all the many rocks below that value are perfect adcumulates with zero residual porosity. Two great surprises emerge from the data. First, there is an abrupt spike from a residual porosity of about 2 to 8 % at 90 PCS, attributed to the great over-production and recovery of Augite and Fe-Ti oxides arising from varied interaction with the internal reservoir of the large magma chamber (Morse, 1979 and 2008 JPet). Second, the Fo range of olivine dramatically shows the same pattern, with Fo range up to 6%; the disparately located grains of olivine have not equilibrated with each other at above 1,000 degrees C and for many thousands of years. This demonstrates the lack of interconnected trapped liquid, hence no mushy zone. Application to meager An-range data in the Skaergaard intrusion shows a similar V-shaped decline to zero porosity (at about 70 PCS) followed by a rise to high values at the Sandwich Horizon. In both intrusions, the late rise signifies increasing cooling by conduction and increasing dominance of feldspar networks that retain increasingly large volumes of trapped liquid. The hard ground of maximum adcumulus growth occurs near the top of the Skaergaard Middle Zone, probably explaining why the sulfide-related Au-PGE ores accumulated there. Quantification of INITIAL porosity in natural rocks is seldom if ever achieved, but can plausibly be mapped from the upper limit of the An range for the Kiglapait intrusion. This parameter must apparently also reach zero at 99 PCS, implying a packing fraction of 1.0 at a minimum rate of accumulation. Hence we may now learn the whole route from initial to final porosity for a large array of cumulate rocks. The results will constrain our hypotheses about accumulation, solidification, and mushy zones for a long time to come.