Reaction between slab-derived melts and peridotite in the mantle wedge: experimental constraints at 3.8 GPa

Laboratory experiments on natural, hydrous basalts at 1–4 GPa constrain the composition of “unadulterated” partial melts of eclogitized oceanic crust within downgoing lithospheric slabs in subduction zones. We complement the “slab melting” experiments with another set of experiments in which these same “adakite” melts are allowed to infiltrate and react with an overlying layer of peridotite, simulating melt:rock reaction at the slab–mantle wedge interface. In subduction zones, the effects of reaction between slab-derived, adakite melts and peridotitic mantle conceivably range from hybridization of the melt, to modal or cryptic metasomatism of the sub-arc mantle, depending upon the “effective” melt:rock ratio. In experiments at 3.8 GPa, assimilation of either fertile or depleted peridotite by slab melts at a melt:rock ratio 2:1 produces Mg-rich, high-silica liquids in reactions which form pyrope-rich garnet and low-Mg# orthopyroxene, and fully consume olivine. Analysis of both the pristine and hybridized slab melts for a range of trace elements indicates that, although abundances of most trace elements in the melt increase during assimilation (because melt is consumed), trace element ratios remain relatively constant. In their compositional range, the experimental liquids closely resemble adakite lavas in island-arc and continental margin settings, and adakite veins and melt inclusions in metasomatized peridotite xenoliths from the sub-arc mantle. At slightly lower melt:rock ratios (1:1), slab melts are fully consumed, along with peridotitic olivine, in modal metasomatic reactions that form sodic amphibole and high-Mg# orthopyroxene.     

Experimental Phase Relations in the System Tonalite-Peridotite-H2O at 15 kb; Implications for Assimilation and Differentiation Processes near the Crust-Mantle Boundary

Experiments at 15 kb in the tonalite-peridotite-H₂O system provideinformation on some of the phase equilibrium factors that mayinfluence reaction and assimilation processes between quartznormativemagmas and ultramafic rocks in the deep crust and upper mantle.Experiments were done with 5 or 10 wt.% H₂O added to powderednatural samples of tonalite, and mixtures of tonalite with 5or 10 wt.% peridotite added (TP5 and TP10, respectively). Theliquidus phase relations of these starting compositions wereinvestigated between 850 and 1100?C at 15 kb, using gold capsulesso that iron loss to the sample containers was not a problemand meaningful glass and mineral analyses could be obtained.Experiments on the tonalite alone show either liquidus garnet,for samples with 5% H₂O added, or liquidus hornblende, for sampleswith 10% H₂O. In contrast, orthopyroxene is the sole liquidusphase, irrespective of water content, in experiments using startingmixtures of 5 or 10 wt.% peridotite added to tonalite. Glassanalyses of partially crystallized tonalite define a crystallizationpath diverging significantly from the calc-alkaline trend towardshigher Ca/(Mg + Fe) in the CaO–(MgO + FeO)–?SiO₂triangle. In contrast, glasses from partially crystallized mixturesof tonalite with 5 or 10 wt.% peridotite added define a liquidtrend close to natural calc-alkaline compositions in terms ofCa/(Mg + Fe). Of more general significance, the proximity ofa field ofliquidus orthopyroxene on the high (Mg + Fe) sideof compositions along the calc-alkaline trend serves to limitthe Mgenrichment of such melts by interaction with ultramaficrocks. Unless heat is added to the system, reaction of tonaliticcomposition melts with ultramafic rocks will produce only slightlyMg-enriched melts: increasing degree of reaction simply resultsin further precipitation of orthopyroxene + garnet ? clinopyroxeneonce melt compositions reach the orthopyroxene field boundary.     

Magmatic processes and mantle heterogeneity beneath the slow-spreading northern Kolbeinsey Ridge segment,North Atlantic

We present new data on mineralogical, major and trace element compositions of lavas from the northernmost segment of the Kolbeinsey Ridge (North Kolbeinsey Ridge, NKR). The incompatible element enriched North Kolbeinsey basalts lie on a crystal fractionation trend which differs from that of the other Kolbeinsey segments, most likely due to higher water contents (~0.2%) in the NKR basalts. The most evolved NKR magmas erupt close to the Jan Mayen Fracture Zone, implying increased cooling and fractionation of the ascending magmas. Mainly incompatible element-enriched basalts, as well as some slightly depleted lavas, erupt on the NKR. They show evidence for mixing between different mantle sources and magma mixing. North Kolbeinsey Ridge magmas probably formed by similar degrees of melting to other Kolbeinsey basalts, implying that no lateral variation in mantle potential temperature occurs on the spreading axis north of the Iceland plume and that the Jan Mayen Fracture Zone does not have a cooling effect on the mantle. Residual garnet from deep melting in garnet peridotite or from enriched garnet pyroxenite veins does not play a role. The incompatible element-enriched source has high Ba/La and Nb/Zr, but must be depleted in iron. The iron-depleted mantle is less dense than surrounding mantle and leads to the formation of the North Kolbeinsey segment and its shallow bathymetry. The enriched NKR source formed from a relatively refractory mantle, enriched by a small degree melt rather than by recycling of enriched basaltic crust. The depleted mantle source resembles the mantle of the Middle Kolbeinsey segment with a depletion in incompatible elements, but a fertile major element composition.     

Sheared peridotite xenolith from the V. Grib kimberlite pipe,Arkhangelsk Diamond Province,Russia: Texture,composition, and origin

The petrography and mineral composition of a mantle-derived garnet peridotite xenolith from the V. Grib kimberlite pipe (Arkhangelsk Diamond Province, Russia) was studied. Based on petrographic characteristics, the peridotite xenolith reflects a sheared peridotite. The sheared peridotite experienced a complex evolution with formation of three main mineral assemblages: (1) a relict harzburgite assemblage consist of olivine and orthopyroxene porphyroclasts and cores of garnet grains (Gar1) with sinusoidal rare earth elements (REE) chondrite C1 normalized patterns; (2) a neoblastic olivine and orthopyroxene assemblage; (3) the last assemblage associated with the formation of clinopyroxene and garnet marginal zones (Gar2). Major and trace element compositions of olivine, orthopyroxene, clinopyroxene and garnet indicate that both the neoblast and clinopyroxene-Gar2 mineral assemblages were in equilibrium with a high Fe-Ti carbonate-silicate metasomatic agent. The nature of the metasomatic agent was estimated based on high field strength elements (HFSE) composition of olivine neoblasts, the garnet-clinopyroxene equilibrium condition and calculated by REE-composition of Gar2 and clinopyroxene. All these evidences indicate that the agent was a high temperature carbonate-silicate melt that is geochemically linked to the formation of the protokimberlite melt.     

雷琼地区晚新生代玄武岩地球化学：EM2成分来源及大陆岩石圈地幔的贡献

对雷琼地区21个晚新生代玄武岩样品的主量、微量元素和Sr、Nd、Pb同位素分别用湿化学法、ICP-MS和MC-ICPMS进行了测定.这些玄武岩主要为石英拉斑玄武岩,其次为橄榄拉斑玄武岩和碱性玄武岩.大多数样品的微量元素和同位素成分与洋岛玄武岩(OIBs)相似,而且随着SiO_2不饱和度增加,不相容元素含量也增加.除R4-1可能受到地壳混染外,其他样品相对均一的Nd同位素(ε_(Nd)=2.5-6.0)以及变化明显但范围有限的Sr同位素(0.703106～0.704481),可能继承了地幔源区的特征.~(87)Sr/~(86)Sr与~(206)Pb/~(204)Pb的正相关和~(143)Nd/~(144)Nd与~(206)Pb/~(204)Pb的负相关特征暗示DM(软流圈地幔)与EM2(岩石圈地幔)的混合.地幔捕虏体的同位素特征暗示EM2成分不可能存在于尖晶石橄榄岩地幔,而La/Yb和Sm/Yb系统表明岩浆由石榴石橄榄岩部分熔融产生,这意味着EM2成分可能存在于石榴石橄榄岩地幔.雷琼地区玄武岩的地球化学变化可以用软流圈地幔为主的熔体加入不同比例石榴石橄榄岩地幔不同程度熔融产生的熔体来解释:碱性玄武岩和橄榄拉斑玄武岩是软流圈熔体与石榴石橄榄岩地幔较低程度(7%～9%)熔融体混合,而石英拉斑玄武岩是软流圈熔体与石榴石橄榄岩地幔较高程度(10%～20%)熔融体的混合.     

Petrogenesis of the amphibole-rich veins from the Lherz orogenic lherzolite massif (Eastern Pyrenees, France): a case study for the origin of orthopyroxene-bearing amphibole pyroxenites in the lithospheric mantle

The Lherz orogenic lherzolite massif (Eastern French Pyrenees) displays one of the best exposures of subcontinental lithospheric mantle containing veins of amphibole pyroxenites and hornblendites. A reappraisal of the petrogenesis of these rocks has been attempted from a comprehensive study of their mutual structural relationships, their petrography and their mineral compositions. Amphibole pyroxenites comprise clinopyroxene, orthopyroxene and spinel as early cumulus phases, with garnet and late-magmatic K₂O-poor pargasite replacing clinopyroxene, and subsolidus exsolution products (olivine, spinel II, garnet II, plagioclase). The original magmatic mineralogy and rock compositions were partly obscured by late-intrusive hornblendites and over a few centimetres by vein–wallrock exchange reactions which continued down to subsolidus temperatures for Mg–Fe. Thermobarometric data and liquidus parageneses indicate that amphibole pyroxenites started to crystallize at P ≥ 13 kbar and recrystallized at P < 12 kbar. The high Al^VI/Al^IV ratio (>1) of clinopyroxenes, the early precipitation of orthopyroxene and the late-magmatic amphibole are arguments for parental melts richer in silica but poorer in water than alkali basalts. Their modelled major element compositions are similar to transitional alkali basalt with about 1–3 wt% H₂O. In contrast to amphibole pyroxenites, hornblendites only show kaersutite as liquidus phase, and phlogopite as intercumulus phase. They are interpreted as crystalline segregates from primary basanitic magmas (mg=0.6; 4–6 wt% H₂O). These latter cannot be related to the parental liquids of amphibole pyroxenites by a fractional crystallization process. Rather, basanitic liquids mostly reused pre-existing pyroxenite vein conduits at a higher structural level (P ≤ 10 kbar). A continuous process of redox melting and/or alkali melt/peridotite interaction in a veined lithospheric mantle is proposed to account for the origin of the Lherz hydrous veins. The transitional basalt composition is interpreted in terms of extensive dissolution of olivine and orthopyroxene from wallrock peridotite by alkaline melts produced at the mechanical boundary layer/thermal boundary layer transition (about 45–50 km deep). Continuous fluid ingress allowed remelting of the deeper veined mantle to produce the basanitic, strongly volatiles enriched, melts that precipitated hornblendites. A similar model could be valid for the few orthopyroxene-rich hydrous pyroxenites described in basalt-hosted mantle xenoliths. Received: 15 September 1999 / Accepted: 31 January 2000     

Petrogenesis of Hawaiian tholeiites: 1, phase equilibria constraints

The most magnesian olivine phenocrysts [Mg no.=100 Mg/(Mg+Fe)=90.5] in Hawaiian tholeiites provide evidence for the earliest stages of differentiation of Hawaiian magmas. Based on the correction of olivine fractionation effects, the primitive melt compositions which have crystallised these olivines are picritic with 16 wt% MgO. They are excellent primary-melt candidates. An experimental study on a new Hawaiian picritic primary-melt estimate demonstrates multiple saturation with peridotite (harzburgite) at 2.0 GPa and 1450° C. Garnet is not a liquidus phase at pressures below 3.5 GPa, and garnet peridotite is not a liquidus phase assemblage at any pressure or temperature. This result confirms previous experimental studies on Hawaiian primary-melt estimates and conflicts with trace-elementgeochemistry-based interpretations, which claim that melt generation occurs in the presence of residual garnet. If Hawaiian tholeiite primary magmas are picritic and have equilibrated with garnet-absent peridotite residues, the geochemical and isotopic characteristics of Hawaiian tholeiites (i.e. Sm/Nd chondrites and Nd>0) are consistent with their source recently having been enriched in incompatible elements. Previous modelling shows that such characteristics are consistent with source enrichment through the migration of small melt fractions generated at depth in the presence of garnet. This may be effected either at the time of Hawaiian magma genesis through dynamic melt segregation processes or, by melting of a previously enriched mantle source; possibly oceanic lithospheric mantle which has been infiltrated by melt fractions from the underlying asthenosphere prior to Hawaiian magmatism. Alternatively, if Hawaiian primary magmas are ultramafic in composition (20 wt% MgO) they may be generated in the presence of garnet peridotite at pressures 3.0 GPa.     

Melt/harzburgite reaction in the petrogenesis of tholeiitic magma from Kilauea volcano, Hawaii

We use the results of elevated pressure melting experiments to constrain the role of melt/mantle reaction in the formation of tholeiitic magma from Kilauea volcano, Hawaii. Trace element abundance data is commonly interpreted as evidence that Kilauea tholeiite is produced by partial melting of garnet lherzolite. We experimentally determine the liquidus relations of a tightly constrained estimate of primary tholeiite composition, and find that it is not in equilibrium on its liquidus with a garnet lherzolite assemblage at any pressure. The composition is, however, cosaturated on its liquidus with olivine and orthopyroxene at 1.4 GPa and 1425 °C, from which we infer that primary tholeiite is in equilibrium with harzburgite at lithospheric depths beneath Kilauea. These results are consistent with our observation that tholeiite primary magmas have higher normative silica contents than experimentally produced melts of garnet lherzolite. A model is presented whereby primary tholeiite forms via a two-stage process. In the first stage, magmas are generated by melting of garnet lherzolite in a mantle plume. In the second stage, the ascent and decompression of magmas causes them to react with harzburgite in the mantle by assimilating orthopyroxene and crystallizing olivine. This reaction can produce typical tholeiite primary magmas from significantly less siliceous garnet lherzolite melts, and is consistent with the shift in liquidus boundaries that accompanies decompression of an ascending magma. We determine the proportion of reactants by major element mass balance. The ratio of mass assimilated to mass crystallized (M_a/M_c) varies from 2.7 to 1.4, depending on the primary magma composition. We use an AFC calculation to model the effect of melt/harzburgite reaction on melt rare earth and high field strength element abundances, and find that reaction dilutes, but does not significantly fractionate, the abundances of these elements. Assuming olivine and orthopyroxene have similar heats of fusion, the M_a/M_c ratio indicates that reaction is endothermic. The additional thermal energy is supplied by the melt, which becomes superheated during adiabatic ascent and can provide more thermal energy than required. Melt/harzburgite reaction likely occurs over a range of depths, and we infer a mean depth of 42 km from our experimental results. This depth is well within the lithosphere beneath Kilauea. Since geochemical evidence indicates that melt/harzburgite reaction likely occurs in the top of the Hawaiian plume, the plume must be able to thin a significant portion of the lithosphere. Received: 4 February 1997 / Accepted: 27 August 1997     

Experimental data at 3 kbars pressure on parental magma to the Bushveld Complex

Experimental studies, mainly under 3 kbars pressure, have been undertaken on representative samples to determine if any of these compositions could be parental magma to the Bushveld Complex. One such composition, with 12.5% MgO, Mg/(Mg + Fe) of 0.72 and quartz-normative, crystallizes olivine, Fo₈₈, as liquidus mineral, at about 1,300° C, followed at only slightly lower temperature by orthopyroxene at 3 kbars pressure. There is a temperature drop of over 100° C before the appearance of plagioclase and finally clinopyroxene. This crystallization sequence is in excellent agreement with the observed sequence in the lower part of the Bushveld Complex.Results at higher pressures show that this composition cannot be a partial melt from mantle peridotite because olivine is replaced by orthopyroxene as the liquidus mineral at lower crustal pressures. A combination of olivine fractionation and contamination was probably involved in the early evolution of this magma.Experimental data on the other compositions show that they are not suitable as parental magma to the lowest portion of the complex. However, the data are used to construct phase diagrams within the basalt tetrahedron at 3 kbars pressure, which are of relevance to the crystallization of basic magmas in the upper crust.Research undertaken at the Grant Institute of Geology, University of Edinburgh, Scotland     

Anhydrous P-T phase relations of near-primary high-alumina basalt from the South Sandwich Islands

Anhydrous P-T phase relations, including phase compositions and modes, are reported from 10–31 kbar for a near-primary high-alumina basalt from the South Sandwich Islands in the Scotia Arc. The water content of natural subduction-related basalt is probably <0.5 wt.% and thus, these results are relevant to the generation of primary basaltic magmas in subduction zones. At high pressures (>27 kbar) garnet is the liquidus phase followed by clinopyroxene, then quartz/coesite at lower temperatures. At intermediate pressures (17–27 kbar), clinopyroxene is the liquidus phase followed by either garnet, quartz, plagioclase, then orthopyroxene or plagioclase, quartz, garnet, then orthopyroxene depending on the pressure within this interval. At all lower pressures, plagioclase is the liquidus phase followed at much lower temperatures (100° C at 5 kbar) by clinopyroxene. The absence of olivine from the liquidus suggests that the composition studied here could not have been derived from a more mafic parent by olivine fractionation at any pressure investigated, and supports the interpretation that it is primary. If so, these results also preclude an origin for this melt by partial melting of olivine-rich mantle periddotite and suggest instead that it was generated by partial melting of the descending slab (quartz eclogite) leaving clinopyroxene, garnet, or both in the residue. The generally flat REE patterns for low-K series subduction related basalts argue against any significant role for garnet, however, and it is thus concluded that the composition studied here was extracted at 20–27 kbar after sufficiently high degrees of partial melting (50%) to totally consume garnet in the eclogite source. Melting experiments on three MORB composition, although not conclusive, are in agreement with this mechanism. Results at 30 kbar support an origin for tonalite/trondhjemite series rocks by lower degrees of melting (15–30%), leaving both garnet and clinopyroxene in the residue.     

Partial Melt Distributions from Inversion of Rare Earth Element Concentrations

Inverse theory is used to calculate the melt distribution requiredto produce the rare earth element concentrations in a wide varietyof terrestrial and extra-terrestrial magmas. The concentrationsof the major and minor elements in the source regions are assumedto be the same as those for the bulk Earth, and the peridotitemineralogy calculated from the mineral compositions by leastsquares. Rare earth element partition coefficients are thenused for inversion, assuming the melt generation is by fractionalmelting. The mean composition of the magmas is taken to be anestimate of the average composition of the melt. For n-typcand e-type MORB the results agree well with the adiabatic decompressioncalculations if the potential temperatures are 1300 and 1500?Crespectively. The major and minor element compositions calculatedfrom the melt distribution obtained from the inversion alsoagree well with those observed. The observations are consistentwith a melt fraction that increases monotonically towards thesurface, starting at 80 km and producing 9 km of melt in thecase of n-type MORB, and at 120 km to produce 23 km in thecase of e-type MORB. The inversion calculations show that the melt fractions producedbeneath an intact plate by a plume like that beneath Hawaiiare smaller, and are also in agreement with the adiabatic calculationsif the potential temperature of the plume is 1500?C. Much ofthe melt is produced in the depth and temperature range of thetransition from garnet to spinel peridotite, in agreement withlaboratory experiments and with the full convective models ofthe Hawaiian plume. The inversion calculations show that thesource region for Hawaiian tholeiites changes with time fromprimitive to depleted mantle. This behaviour is likely to resultfrom percolation, and the processes involved can be understoodwith the help of a simple analytic model. The last, post-erosional,magmas produced on Oahu come from a source that has been uniformlyenriched in all rare earth elements by a factor of about two.Magmas associated with island arcs come from two sources. Oneresembles that of n-type MORB, and probably is produced by adiabaticupwelling. The other generates calc-alkaline basalt stronglyenriched in light rare earth elements, but with a smaller constantenrichment between Gd and Lu. This composition is consistentwith the extraction of a melt fraction of 1% from a source containing9% of amphibole. Such a source region can also account for thelow values of Ti and Nb, and perhaps also of Ta, observed inisland arc magmas. Basaltic andesites and andesites from islandarcs show the same amphibole signature, and can be producedfrom the calc-alkaline basalts by fractional crystallizationif amphibole separates with olivine and orthopyroxene. The percolationof a small melt fraction through a mantle wedge that containsconsiderable amounts of amphibole can only transport very incompatibleelements, such as He, U, Th, and Rb, towards the Earth's surface.Sr and Nd are likely to be too compatible to move against thematrix flow, but Pb may do so locally. These results have importantimplications for the isotopic systematics of the upper mantle. The melt distributions obtained from ophiolites are like thosefor island arc tholeiites, though a potential temperature of1400 ?C fits the results better than does one of 1300?C. Archaeantholeiites and basaltic komatiites give melt distributions similarto that of e-type MORB from Iceland, and can be produced byadiabatic decompression if the mantle potential temperatureis 1500cC, with tholeiites having lost more material by fractionalcrystallization. The melt distribution obtained from komatiitesrequires the melt fraction to reach 60% at the surface. Thoughthe calculated compositions agree with those observed, decompressionis unable to generate such large melt fractions. Inversion shows that plateau basalts can be produced from theupper mantle beneath the plates by adiabatic upwelling beneatha mechanical boundary layer 60 km thick. Many of the variedalkali-rich continental magmas are generated by melting an enrichedsource in the stability field of garnet peridotite. The averageenrichment required, by a factor of between two and five, canbe produced by the addition of a small melt fraction. Carbonatitesshow no evidence of amphibole involvement at any stage, a resultthat is consistent with their formation by liquid immiscibility.Inversion of the rare earth element concentrations in shalesgives a melt distribution similar to that from calc-alkalinebasalts from island arcs, with a strong amphibole signature.Generation of the continental crust by separation of calc-alkalinemagma from 40% of the mantle can account for the differencebetween primitive and depleted mantle. Low-K highland basalts from the Moon can be produced directlyfrom the average primitive lunar mantle if the melt fractioninvolved is ?0-5%, and if they were generated in the stabilityfield of plagioclase and spinel peridotite. Intermediate-K highlandbasalts come from a source that has been enriched by a factorof about two, and show no evidence of amphibole involvement.The rare earth concentrations in mare basalts require melt fractionsof up to 7% in the spinel peridotite stability field, and canbe generated by adiabatic upwelling of mantle whose potentialtemperature is 1300?C beneath a mechanical boundary layer thatis 150 km thick. Because lunar gravity is only one-sixth ofthat of the Earth, the thickness of the melting zone and thevolume of melt produced are six times greater for the Moon thanfor the Earth for the same value of Tp. Both low-Ti and high-Timare basalts may have lost as much as 70 and 85% respectivelyof their original material through crystal fractionation. Itis, however, difficult to understand how such an origin canaccount for the high magnesium concentrations. Basaltic achondritesinvolve melt fractions of 10-15%, generated in the spinel orplagioclase stability field.     

Petrogenesis of Tertiary Continental Intra-plate Lavas from the Westerwald Region, Germany

Tertiary volcanic rocks from the Westerwald region range frombasanites and alkali basalts to trachytes, whereas lavas fromthe margin of the Vogelsberg volcanic field consist of morealkaline basanites and alkali basalts. Heavy rare earth elementfractionation indicates that the primitive Westerwald magmasprobably represent melts of garnet peridotite. The Vogelsbergmelts formed in the spinel–garnet peridotite transitionregion with residual amphibole for some magmas suggesting meltingof relatively cold mantle. Assimilation of lower-crustal rocksand fractional crystallization altered the composition of lavasfrom the Westerwald and Vogelsberg region significantly. Thecontaminating lower crust beneath the Rhenish Massif has a differentisotopic composition from the lower continental crust beneaththe Hessian Depression and Vogelsberg, implying a compositionalboundary between the two crustal domains. The mantle sourceof the lavas from the Rhenish Massif has higher ²⁰⁶Pb/²⁰⁴Pband ⁸⁷Sr/⁸⁶Sr than the mantle source beneath the Vogelsbergand Hessian Depression. The 30–20 Ma volcanism of theWesterwald apparently had the same mantle source as the QuaternaryEifel lavas, suggesting that the magmas probably formed in apulsing mantle plume with a maximum excess temperature of 100°Cbeneath the Rhenish Massif. The relatively shallow melting ofamphibole-bearing peridotite beneath the Vogelsberg and HessianDepression may indicate an origin from a metasomatized portionof the thermal boundary layer. KEY WORDS: continental rift volcanism; basanites; trachytes; assimilation; fractional crystallization; partial melting     

Inter-mineral Fe isotope variations in mantle-derived rocks and implications for the Fe geochemical cycle

Iron isotope and major- and minor-element compositions of coexisting olivine, clinopyroxene, and orthopyroxene from eight spinel peridotite mantle xenoliths; olivine, magnetite, amphibole, and biotite from four andesitic volcanic rocks; and garnet and clinopyroxene from seven garnet peridotite and eclogites have been measured to evaluate if inter-mineral Fe isotope fractionation occurs in high-temperature igneous and metamorphic minerals and if isotopic fractionation is related to equilibrium Fe isotope partitioning or a result of open-system behavior. There is no measurable fractionation between silicate minerals and magnetite in andesitic volcanic rocks, nor between olivine and orthopyroxene in spinel peridotite mantle xenoliths. There are some inter-mineral differences (up to 0.2‰ in ⁵⁶Fe/⁵⁴Fe) in the Fe isotope composition of coexisting olivine and clinopyroxene in spinel peridotites. The Fe isotope fractionation observed between clinopyroxene and olivine appears to be a result of open-system behavior based on a positive correlation between the Δ⁵⁶Fe_{clinopyroxene-olivine} fractionation and the δ⁵⁶Fe value of clinopyroxene and olivine. There is also a significant difference in the isotopic compositions of garnet and clinopyroxene in garnet peridotites and eclogites, where the average Δ⁵⁶Fe_{clinopyroxene-garnet} fractionation is +0.32 ± 0.07‰ for six of the seven samples. The one sample that has a lower Δ⁵⁶Fe_{clinopyroxene-garnet} fractionation of 0.08‰ has a low Ca content in garnet, which may reflect some crystal chemical control on Fe isotope fractionation. The Fe isotope variability in mantle-derived minerals is interpreted to reflect subduction of isotopically variable oceanic crust, followed by transport through metasomatic fluids. Isotopic variability in the mantle might also occur during crystal fractionation of basaltic magmas within the mantle if garnet is a liquidus phase. The isotopic variations in the mantle are apparently homogenized during melting processes, producing homogenous Fe isotope compositions during crust formation.     

The Source Regions of Ocean Island Basalts

The geochemical modelling of many small-volume continental magmasshows that their source regions must have been depleted by basaltformation, and later enriched by the addition of a metasomaticmelt, formed by melting 03% of the MORB source. The presenceof such magmas throughout western Turkey and the Aegean, whereno plume is present, requires such magmas to be formed at temperaturesconsiderably below the dry solidus. Similar magmas elsewherebring up nodule suites, many of which have the same compositionas the source regions of the host magmas. Pressure and temperatureestimates from garnetbearing suites, and temperature estimatesfrom those without garnet, show that the nodules last equilibratedat pressures and temperatures close to those of the wet solidus.Magmas from the smaller oceanic islands and from some seamountsclosely resemble small-volume continental magmas, and also comefrom sources that have been metasomaticaUy enriched. However,no data sets from any of the oceanic islands that have yet beenmodelled require their source regions to have been depletedbefore being enriched The density of the sources of continentaland oceanic basalts can be obtained from their calculated modes.In the garnet peridotite stability field the sources of oceanisland basalts have densities that are slightly greater thanthat of the MORB source, whereas those of most small-volumecontinental magmas are lighter. Therefore ocean island sourcesalone are easily entrained into the thermal convection beneaththe plates. A numerical experiment shows that material in thehot and cold boundary layers of high Rayleigh number time-dependentconvection tends to remain in the boundary layers for severaloverturns, rather than moving into the interior of the circulation.A simple model that can account for the elemental and isotopiccomposition of ocean island basalts forms their sources by theaddition of metasomatic melt to the undcplcUd MORB source whileit forms the lower part of the mechanical boundary layer beneathcontinents. The isotopic differences between ocean island basaltand MORB are generated before the source becomes entrained inthe cold sinking plumes that fall to the base of the convectinglayer. At the base the material is heated and rises as partof a hot plume. Because the metasomatic melt contains waterand carbonates, the enriched regions start to melt and generatemore melt on decompression than does the MORB source. Such regionscan therefore generate islands and seamounts. Even when theenriched material moves into the interior of the circulationand acquires the mean potential temperature of the mantle, itwill still generate more melt on decompression than will theMORB source, and the isotopic and elemental composition willstill be distinctive. The model can therefore account for theobserved composition of magmas from seamounts that cannot beproduced from either the MORB or the primitive source. ^*Corresponding author     

Petrogenesis of Tertiary Mafic Alkaline Magmas in the Hocheifel, Germany

Primitive nephelinites and basanites from the Tertiary Hocheifelarea of Germany (part of the Central European Volcanic Province;CEVP) have high Mg-number (>0·64), high Cr and Nicontents and strong light rare earth element enrichment butsystematic depletion in Rb, K and Ba relative to trace elementsof similar compatibility in anhydrous mantle. Alkali basaltsand more differentiated magmatic rocks have lower Mg-numberand lower abundances of Ni and Cr, and have undergone fractionationof mainly olivine, clinopyroxene, Fe–Ti oxide, amphiboleand plagioclase. Some nephelinites and basanites approach theSr–Nd–Pb isotope compositions inferred for the EAR(European Asthenospheric Reservoir) component. The Nd–Sr–Pbisotope composition of the differentiated rocks indicates thatassimilation of lower crustal material has modified the compositionof the primary mantle-derived magmas. Rare earth element meltingmodels can explain the petrogenesis of the most primitive maficmagmatic rocks in terms of mixing of melt fractions from anamphibole-bearing garnet peridotite source with melt fractionsfrom an amphibole-bearing spinel peridotite source, both sourcescontaining residual amphibole. It is inferred that amphibolewas precipitated in the asthenospheric mantle beneath the Hocheifel,close to the garnet peridotite–spinel peridotite boundary,by metasomatic fluids or melts from a rising mantle diapir orplume. Melt generation with amphibole present suggests relativelylow mantle potential temperatures (<1200°C); thus themantle plume is not thermally anomalous. A comparison of recentlypublished Ar/Ar ages for Hocheifel basanites with the geochemicaland isotopic composition of samples from this study collectedat the same sample sites indicates that eruption of earlierlavas with an EM signature was followed by the eruption of laterlavas derived from a source with EAR or HIMU characteristics,suggesting a contribution from the advancing plume. Thus, theHocheifel area represents an analogue for magmatism during continentalrift initiation, during which interaction of a mantle plumewith the overlying lithosphere may have led to the generationof partial melts from both the lower lithosphere and the asthenosphere. KEY WORDS: alkali basalts; continental volcanism; crustal contamination; partial melting; Eifel, Germany     

Melting of the Shallow Upper Mantle: A New Perspective

Detailed examination of liquidus phase relationships in binaryand ternary joins of the CFMAS +Cr system has permitted a rigorousdetermination of the dry melting path of an initially fertilespinel peridotite composition resembling Bulk Silicate Earthor MORB-pyrolite. It is demonstrated that it is impossible tomodel mantle melting accurately using only one set of ratiosof phases entering the melt; this implies that the melting processis primarily controlled by solid solution rather than eutecticbehaviour. The proportions of phases entering a melt dependon whether a phase reacts and/or disappears from a system, andon the choice of the initial and final peridotite compositions.Four discrete domains in the melting regime of upper-mantleperidotites are distinguished, each characterized by differentphase melting coefficients, relating to the melting of: (1)lherzolites, (2) clinopyroxene-bearing harzburgites (i.e., free-clinopyroxene),(3) clinopyroxene-saturated harzburgites (i.e., clinopyroxenein solid solution in orthopyroxene), and (4) clinopyroxene-freeharzburgites (i.e., no clinopyroxene). The proposed non-linearfashion in which mantle lithologies melt explains the inadequacyof all previous models to reproduce the observed compositionsof upper-mantle peridotite melting residues. It is suggestedthat: (1) olivine and orthopyroxene will melt cotectically;(2) clinopyroxene and spinel will lose most of their aluminouscomponent after {small tilde}8% melting within the first 4 kb({smalltilde} 12 km) of ascent from the dry solidus; and that (3) clinopyroxenewill disappear completely from a MORB-pyrolite mantle after{small tilde}42% melting. Although such a number is significantlyhigher than that dictated by the position of the clinopyroxene-outcurves from peridotite isobaric equilibrium melting experiments({small tilde}22%), it is emphasized that the latter are a grossoversimplification of the natural melting process and are notequivalent to melting during adiabatic upwelling. It is concludedthat the commonly postulated disappearance of clinopyroxenefrom fertile peridotite compositions at {small tilde}22% meltingis greatly in error if melting in an adiabatically rising mantleis considered, thus providing an explanation for many unsuccessfulattempts by various authors to model the behaviour of transitionelements in sub-oceanic and supra-subduction-zone mantle andderivative magmas.     

Constraints from melt inclusions and their host olivines on the petrogenesis of Oligocene-Early Miocene Xindian basalts, Chifeng area, North China Craton

The geochemical characteristics of melt inclusions and their host olivines provide important information on the processes that create magmas and the nature of their mantle and crustal source regions. We report chemical compositions of melt inclusions, their host olivines and bulk rocks of Xindian basalts in Chifeng area, North China Craton. Compositions of both bulk rocks and melt inclusions are tholeiitic. Based on petrographic observations and compositional variation of melt inclusions, the crystallizing sequence of Xindian basalts is as follows: olivine (at MgO > ~5.5 wt%), plagioclase (beginning at MgO = ~5.5 wt%), clinopyroxene and ilmenite (at MgO < 5.0 wt%). High Ni contents and Fe/Mn ratios, and low Ca and Mn contents in olivine phenocrysts, combining with low CaO contents of relatively high MgO melt inclusions (MgO > 6 wt%), indicate that Xindian basalts are possibly derived from a pyroxenite source rather than a peridotite source. In the CS-MS-A diagram, all the high MgO melt inclusions (MgO > 6.0 wt%) project in the field between garnet + clinopyroxene + liquid and garnet + clinopyroxene + orthopyroxene + liquid near 3.0 GPa, further suggesting that residual minerals are mainly garnet and clinopyroxene, with possible presence of orthopyroxene, but without olivine. Modeling calculations using MELTS show that the water content of Xindian basalts is 0.3–0.7 wt% at MgO = 8.13 wt%. Using 20–25 % of partial melting estimated by moderately incompatible element ratios, the water content in the source of Xindian basalts is inferred to be ≥450 ppm, much higher than 6–85 ppm in dry lithospheric mantle. The melting depth is inferred to be ~3.0 GPa, much deeper than that of tholeiitic lavas (<2.0 GPa), assuming a peridotite source with a normal mantle potential temperature. Such melting depth is virtually equal to the thickness of lithosphere beneath Chifeng area (~100 km), suggesting that Xindian basalts are derived from the asthenospheric mantle, if the lithospheric lid effect model is assumed.     

Anhydrous Partial Melting of a Fertile and Depleted Peridotite from 2 to 30 kb and Application to Basalt Petrogenesis

Reversal experiments have been performed to check the methodof Jaques & Green (1980) to determine equilibrium partialmelts from peridotite compositions. Reversals of the Jaques& Green (1980) calculated equilibrium partial melt (CEPM)compositions have been carried out in two ways: (1) By runningCEPM compositions at original P and T conditions and testingfor multiple saturation in residual phases of the original experiments.(2) By sandwich/mixed experiments using the CEPM compositionplus peridotite (either Hawaiian pyrolite or Tinaquillo lherzolite). The glass (liquid) compositions from the first series of experimentsshow that the CEPM compositions of Jaques & Green (1980)are too olivine-rich. The glass (liquid) compositions from thesecond series of experiments define new olivine+orthopyroxene?clinopyroxenecotectics in a molecular normative tetrahedron. The new cotecticsplot towards the Qz apex of the tetrahedron, away from the cotecticsdefined by the CEPM compositions of Jaques & Green (1980).Partial melt compositions have also been determined at 20 and30 kb, using both the sandwich technique and the approach bymodal analysis and mass balance. The results of the experimental study are used to evaluate thepetrogenesis of mid-ocean ridge basalts, Hawaiian tholentesand primary magmas in intraoceanic convergent margin settings.     

Carbonation and melting reactions in the system CaO–MgO–SiO2–CO2 at mantle pressures with geophysical and petrological applications

Bowen's petrogenetic grid was based initially on a series of decarbonation reactions in the system CaO-MgO-SiO₂-CO₂ with starting assemblages including calcite, dolomite, magnesite and quartz, and products including enstatite, forsterite, diopside and wollastonite. We review the positions of 14 decarbonation reactions, experimentally determined or estimated, extending the grid to mantle pressures to evaluate the effect of CO₂ on model mantle peridotite composed of forsterite(Fo)+orthopyroxene(Opx)+clinopyroxene(Cpx). Each reaction terminates at an invariant point involving a liquid, CO₂, carbonates, and silicates. The fusion curves for the mantle mineral assemblages in the presence of excess CO₂ also terminate at these invariant points. The points are connected by a series of reactions involving liquidus relationships among the carbonates and mantle silicates, at temperatures lower (1,100–1,300° C) than the silicate-CO₂ melting reactions (1,400–1,600° C). Review of experimental data in the bounding ternary systems together with preliminary data for the system CaO-MgO-SiO₂-CO₂ permits construction of a partly schematic framework for decarbonation and melting reactions at upper mantle pressures. The key to several problems in the peridotite-CO₂ subsystem is the intersection of a subsolidus carbonation reaction with a melting reaction at an invariant point near 24 kb and 1,200°C. There is an intricate series of reactions between 25 kb and 35 kb involving changes in silicate and carbonate phase fields on the CO₂-saturated liquidus surfaces. Conclusions include the following: (1) Peridotite Fo+Opx+Cpx can be carbonated with increasing pressure, or decreasing temperature, to yield Fo+Opx+Cpx+Cd (Cd=calcic dolomite), Fo+Opx+Cd, Fo+Opx+Cm (Cm=calcic magnesite), and finally Qz+Cm. (2) Free CO₂ cannot exist in subsolidus mantle peridotite with normal temperature distributions; it is stored as carbonate, Cd. (3) The CO₂ bubbles in peridotite nodules do not represent free CO₂ in mantle peridotite along normal geotherms. (4) CO₂ is as effective as H₂O in causing incipient melting, our preferred explanation for the low-velocity zone. (5) Fusion of peridotite with CO₂ at depths shallower than 80 km produces basic magmas, becoming more SiO₂-undersaturated with depth. (6) The solubility of CO₂ in mantle magmas is less than about 5 wt% at depths to 80 km, increasing abruptly to about 40 wt% at 80 km and deeper. (7) Deeper than 80 km, the first liquids produced are carbonatitic, changing towards kimberlitic and eventually, at considerably higher temperatures, to basic magmas. (8) Kimberlite and carbonatite magmas rising from the asthenosphere must evolve CO₂ at depths 100-80 km, which contributes to their explosive emplacement. (9) Fractional crystallization of CO₂-bearing SiO₂-undersaturated basic magmas at most pressures can yield residual kimberlite and carbonatite magmas.     

Experimental tests of garnet peridotite oxygen barometry

We have performed experiments aimed at testing the calibration of oxygen barometers for the garnet peridotite [garnet (Gt)-olivine (Ol)-orthopyroxene (Opx)] phase assemblage. These involved equilibrating a thin layer of garnet sandwiched between layers of olivine and orthopyroxene at 1300°C and 23–35 kbar for 1–7 days. Oxygen fugacity was controlled (but not buffered) by using inner capsules of Fe?Pt alloy or graphitc or molybdenum sealed in welded Pt outer capsules. Post-experiment measurement of fO₂ was made by determining the compositions of Pt-Fe alloy sensors at the interface between garnet and olivine + orthopyroxene layers. The composition of alloy in equilibrium with olivine + orthopyroxene was approached from Fe-oversaturated and Fe-undersaturated conditions in the same experiment with, in general, excellent convergence. Product phase compositions were determined by electron microprobe and a piece of the garnet layer saved for ⁵⁷Fe Mössbauer spectroscopy. The latter gave the Fe³⁺ content of the garnet at the measured P-T-fO₂ conditions. Approach to equilibrium was checked by observed shifts in Fe³⁺ content and by the approach of garnet-olivine Fe?Mg partitioning to the expected value. The compositions of the phases were combined with mixing properties and thermodynamic data to calculate an apparent fO₂ from two possible garnet oxybarometers:- (1) $\begin{gathered} 2Ca_3 Fe_2 Si_3 O_{12} + 2Mg_3 Al_2 Si_3 O_{12} + 4FeSiO_3 = 2Ca_3 Al_2 Si_3 O_{12} \hfill \\ Gt Gt Opx Gt \hfill \\ + 8FeSi_{0.5} O_2 + 6MgSiO_3 + O \hfill \\ Ol Opx \hfill \\ \end{gathered} $ and (2) $\begin{gathered} 2Fe_3 Fe_2 Si_3 O_{12} = 8FeSi_{0.5} O_2 + 2FeSi_3 O_2 \hfill \\ Gt Ol Opx \hfill \\ \end{gathered} $ Comparison of calculated fO₂s with those measured by the Pt-Fe sensors demonstrated that either barometer gives the correct answer within the expected uncertainty. Data from the first (Luth et al. 1990) has an uncertainty of about 1.6 logfO₂ units, however, while that from equilibrium (2) (Woodland and O'Neill 1993) has an error of +/- 0.6 log units, comparable to that of the spinel peridotite oxybarometer. We therefore conclude that equilibrium (2) may be used to calculate the fO₂ recorded by garnet peridotites with an uncertainty of about +/- 0.6 log units, providing the potential to probe the oxidation environment of the deep continental lithosphere. Preliminary application based on data from Luth et al. (1990) indicates that garnet peridotite xenoliths from Southern Africa record oxygen fugacities about 3.0 log units below the FMQ (fayalite-magnetite-quartz) buffer. These are substantially more reducing conditions than those recorded by continental spinel lherzolites which typically give oxygen fugacities close to FMQ (Wood et al. 1990).