Deep Mantle Component Research Articles

Evidence for a primitive deep mantle component in the source of Marquesas Islands Lavas from Os isotope and highly siderophile element abundance systematics

The Marquesas islands are an age-progressive volcanic chain in French Polynesia formed by partial melting above a mantle plume. Previous studies have demonstrated the presence of an enriched mantle (EM2) endmember mixed with a high-3He/4He (up to 14.4 RA) component in Marquesas post-shield lavas. In contrast, Marquesas shield lavas sample a dilute high-μ (HIMU) type component with lower 3He/4He (∼8 RA). We report new incompatible trace element and highly siderophile element (HSE: Os, Ir, Ru, Pt, Pd, Re) abundances and osmium isotope data for bulk rocks from the islands of Fatu Hiva, Hiva Oa and Nuku Hiva, for which He, Sr, Nd and Pb isotope compositions have been reported previously (Castillo et al., 2007). As the HSE are chiefly hosted in sulfide phases in the mantle and all but Re are compatible during partial melting, they offer information on both sulfide-rich and melt-depleted mantle lithologies within the Marquesas plume. Primitive mantle-normalized HSE abundances show depletions in Ru and Pd, and fractionated (Ir/Os)N in fourteen of the seventeen analyzed samples. Ruthenium depletions and Ir-Os fractionations are likely caused by parental magma fractional crystallization. Palladium depletions are not easily explained by magmatic or secondary alteration processes and are instead likely to be a feature of the source mantle which was partially depleted in the more incompatible HSE by previous melting events within the Marquesas plume. Osmium isotopic systematics are mildly radiogenic relative to chondrites and are identical within uncertainties for shield-building Fatu Hiva (γOs = +2.9 ± 1.6; 2 SD; n = 7) and post-shield Hiva Oa (+2.7 ± 2.2; n = 9), showing no evidence for significant assimilation of radiogenic oceanic crust. The osmium budget of Marquesas lavas is controlled by a volumetrically dominant peridotite endmember that is incompatible element depleted while the Sr, Nd, and Pb budgets are strongly influenced by a volumetrically minor, incompatible element enriched EM2 component. The Nd-Sr-Os isotope systematics of the lavas can be explained by mixing of 0.3–0.7% of an EM2 type melt into the Marquesas source. Marquesas lavas offer further support that the high-3He/4He “focus zone” (FOZO) mantle has 187Os/188Os mildly more radiogenic than the depleted upper mantle, similar to estimates of the primitive mantle. Osmium isotope systematics offer a useful tool with which to characterize more depleted components within the ocean island basalt-source mantle that may be difficult to detect using lithophile isotope signatures.

Estimating the carbon content of the deep mantle with Icelandic melt inclusions

Earth's carbon budget is central to our understanding of the long-term co-evolution of life and the planet. Direct observations of surface reservoirs allow for the detailed quantification of their carbon content. However, the carbon content of Earth's deep interior remains poorly constrained. Here we study olivine-hosted melt inclusions from two Icelandic eruptions, with those from the Miðfell eruption allowing us to investigate the carbon content of the deep mantle. Comparison with the previously studied Borgarhraun eruption highlights the presence of deep, plume-sourced mantle material within the Miðfell source region. Miðfell contains trace element-depleted melt inclusions undersaturated in CO2, which have high CO2/Ba (=396±48) and CO2/Nb (=1832±316), though some inclusions preserve even greater relative carbon enrichment. These observations allow us to reconstruct the CO2 content of the bulk Miðfell source as being >690 ppm. By identifying that Miðfell is a mixture of depleted and deep mantle components, we can estimate a CO2 content for the deep mantle component of 1350±350 ppm; a concentration that is over ten times higher than depleted MORB mantle estimates. Assuming that the deep mantle component identified in Miðfell is representative of a global reservoir, then with our new CO2 estimate and by considering a range of representative mantle fractions for this reservoir, we calculate that it contains up to 14 times more carbon than that of the atmosphere, oceans, and crust combined. Our result of elevated CO2/Ba and CO2/Nb ratios, and carbon enrichment support geochemical bulk Earth carbon models that call for the presence of carbon-rich deep mantle domains to balance Earth's relatively carbon-poor upper mantle and surface environment.

Interaction between the supercontinent cycle and the evolution of intrinsically dense provinces in the deep mantle

AbstractShear wave travel times in the Earth's deep mantle reveal broad steep‐sided seismologically distinct provinces lying on the core‐mantle boundary. The longevity and permanence of the two large principal provinces, located below the sites of present‐day Africa and the Pacific Ocean, have become a matter of great interest. Examination of the flood basalt record and kimberlite eruption dating suggests the presence of these provinces may disclose a deep mantle component with a compositionally distinct origin that plays a role in the generation of mantle plumes at preferred locations. By extension, the presence of these provinces may affect the supercontinent cycle. Implementing a mantle convection model featuring distinct continental lithosphere and a compositionally anomalous and intrinsically dense (CAID) component, we study the distribution and mobility of naturally forming, compositionally distinct provinces and their impact on model supercontinent assembly. We employ 2‐D Cartesian geometry calculations of thermochemical convection with extremely low compositional diffusion to model Earth‐like convective vigor on a global scale and find that an intrinsically dense mantle component generally aggregates into one or two broad provinces. The positions of the provinces are time dependent, but in many of our calculations the province locations are characterized by periods of fixity that reach several hundred million years. Eras of province and associated plume fixity are punctuated by periods of relatively rapid migration. A correlation between supercontinent position and the locations of CAID provinces is not supported by our findings. However, we find that the frequency of supercontinent assemblies increases when CAID provinces are present.

Helium isotopic composition of hydrothermal fluids from the Manus back-arc Basin, Papua New Guinea

We present isotopic analysis of helium in hydrothermal fluids from the Manus back-arc basin, collected during the French-Japanese Manusflux and Manaute cruises in 1995 and 2000, respectively. The helium isotope composition of the fluids is comparable to that measured elsewhere in similar subduction environments, with 3He/4He values (7.4 to 7.9 Ra) slightly lower than typical MORB values, except on the Manus Spreading Center (MSC) where higher values (12 ± 0.1 Ra) are observed. The causal link between the narrow spatial extent of the 3He/4He anomaly at the MSC and a 3He-rich deep mantle component remains elusive, based on available geophysical data. Thus, the Manus Spreading Center results may be an interesting example of small scale upper-mantle heterogeneity.

Melt-generation processes associated with the Tristan mantle plume: Constraints on the origin of EM-1

The enriched mantle 1 (EM-1) component in ocean–island basalts (OIB, e.g., Kerguelen, Pitcairn and Walvis Ridge) has been attributed to melting in upwelling mantle plumes of either: (i) shallow-recycled delaminated subcontinental-lithospheric-mantle or crust; or (ii) deep-recycled metasomatised lithosphere, oceanic plateau or oceanic crust plus a few percent of pelagic sediment. We present new geochemical data for OIB samples from the central South Atlantic; these include 100 to 30 Ma alkali and tholeiitic basalts from the Walvis Ridge and Rio Grande Rise and < 3 Ma basanites and basalts from Tristan da Cunha, Inaccessible and Gough. Additionally, we have analysed Cretaceous mafic–potassic magmas from south-west Africa and eastern South America in order to establish the compositional variation of metasomatised lithospheric mantle that may have been delaminated during Gondwana break-up. The results of our rare-earth-element inversion and Sr-, Nd- and Pb-isotopic mixing models suggest that the ‘depleted’ mantle plume component resembles FOZO and that the composition of the ‘enriched’ mantle component in central South Atlantic OIB has varied both spatially and temporally. At least three different enriched mantle end-members are required to explain the compositional range of 100 to 30 Ma OIB magmas. These resemble the source regions of mafic–potassic magmas from: (i) the Congo craton and Damara belt of south-west Africa; (ii) the São Francisco craton and Brasilia belt of south-east Brazil; and (iii) the Rio Apa–Luis Alves craton of southern Brazil and Paraguay. The most isotopically enriched EM-1 basalts ( ɛNd = − 0.8 to − 4.5), generated on the Walvis Ridge and Rio Grande Rise between 89 and 78 Ma, appear to contain a 10% to 15% contribution from a melt source region with low ɛNd, 206Pb / 204Pb and high [La / Nb] n , similar in composition to metasomatised subcratonic lithospheric mantle beneath southern Brazil and Paraguay. Reconstructions of plate motions indicate that, at the time of continental break-up, this would have been located above the sites of subsequent EM-1 melt generation. Shallow-mantle recycling of metasomatised lithosphere delaminated from south-west Africa and south-east Brazil may explain the further variations in trace-element and isotopic ratios of 100 and 80 to 30 Ma basalts from the Walvis Ridge, respectively. Recent magmas from Tristan, Inaccessible and Gough contain a melt contribution from an enriched mantle source with high 206Pb / 204Pb, 208Pb / 204Pb and 87Sr / 86Sr ratios and low ɛNd, relative to bulk-Earth. This does not resemble the isotopic composition of mafic–potassic magmas from continents adjacent to the South Atlantic and we therefore propose that it is a deep-recycled mantle component. The high Pb-isotopic ratios of recent Tristan plume magmas are inconsistent with a melt contribution from recycled oceanic crust plus a few percent pelagic sediments. We suggest that the deep mantle component is recycled metasomatised lithospheric mantle. Recycled metasomatised lithospheric mantle, of variable composition, only appears to have been an intrinsic part of the sub-oceanic Tristan mantle plume and the bulk-rock composition of present-day Tristan plume related melts are therefore inappropriate for modelling the contribution of plume-derived melts to the Paraná–Etendeka continental flood-basalts. This relationship may also apply to other large igneous provinces and ocean islands along their associated hot spot tracks.

Noble gas isotopes suggest deep mantle plume source of late Cenozoic mafic alkaline volcanism in Europe

Late Cenozoic mafic alkaline volcanism associated with lithospheric thinning is a widespread phenomenon in Europe. However, there is no consensus about its geodynamical causes, e.g., if active shallow or deep mantle-derived plume upwelling has driven this activity. Seismic tomography demonstrated 400–500 km deep plume-like structures under the Massif Central and Eifel volcanic fields, but cannot prove compellingly a lower mantle root. Here, we present high-precision helium, neon and argon isotope data on ultramafic xenoliths from mafic alkaline volcanic formations from the Pannonian basin and the Eifel region. Noble gas extraction by stepwise crushing allowed a successful separation of variably fractionated atmosphere- and mantle-derived components, and the detailed evaluation of helium–neon and neon–argon isotope mixing relationships. The mantle xenoliths trapped mantle fluids with two distinct noble gas components: one component is derived from the local lithospheric and/or a HIMU mantle source with 21Ne/22Nemantle=0.07 and 4He/3He of about 120,000, and dominates helium and part of the neon. The second component is similar to hot spot-type noble gases, e.g., Hawaii or Réunion. It is clearly resolvable in neon isotopes and dominates the argon budget. Its isotopic structure is determined to 40Ar/36Ar=15,000±1200, 21Ne/22Nemantle=0.042±0.005 and 4He/3He of about 36,000. Our results are consistent with previous suggestions that a deep mantle component is involved in European intraplate volcanism.

Search for a deep-mantle component in mafic lavas using a Nb–Y–Zr plot

Concentrations of trace elements with bulk distribution coefficients &lt; 1 have been utilized to discriminate between mafic magmas erupted in different tectonic settings. Herein, the Nb–Y–Zr contents of mafic lavas will be utilized to pinpoint derivation from a deep-mantle source. Magmas derived from the Icelandic plume exhibit ΔNb &gt; 0, where ΔNb = log (Nb/Y) + 1.74 – 1.92 log (Zr/Y) (Fitton et al. 1997), whereas melts derived from depleted sections of the mantle and the crust show ΔNb &lt; 0. Regardless of the nature of the melting event, it was suggested the ΔNb value of magmas were essentially identical to that of the source material. Theoretical modeling of a garnet lherzolite source suggests very small partial melts may show elevated ΔNb values, yielding a spurious plumelike signature. For mafic rocks from different areas, with hypothesized hot spot derivation, ΔNb is calculated to search for a deep-mantle signature (ΔNb &gt; 0). The signature of the Réunion hot spot, from ~65–0 Ma, is seen in the Deccan Traps, India, lavas from the Mascarene Plateau, and Réunion Island. The signature of the Kerguélen plume is detected in rocks from Kerguélen Island and the Ninetyeast and Broken Ridges, Indian Ocean, a few lavas from the Rajmahal province, India, but none in the Bunbury Basalt, Australia. Some rocks from continental flood basalt provinces show a hot spot component, in agreement with conclusions based on 3He/4He studies, others show ΔNb &lt; 0, due to considerable subcontinental contamination. For lavas from Heard Island and Grand Comore, Indian Ocean, the technique pinpoints those showing a plume component. In an area of plate convergence in the Pacific Ocean, lavas with very low (&lt;1 ppm) Nb contents, serve as a critical test for this method. Only rocks close to the Samoan hot spot show ΔNb &gt; 0, whereas the others show the Nb depletion typical of arc-related material. In conjunction with trace-element and isotopic discriminants, this appears to be a useful tool in identifying a deep-mantle component in mafic lavas from a variety of plate tectonic settings.

Search for a deep-mantle component in mafic lavas using a Nb–Y–Zr plot

Search for a deep-mantle component in mafic lavas using a Nb–Y–Zr plot

Noble gas behaviour and composition in the mantle: constraints from the Iceland Plume

Current geochemical models envisage that noble gases in the upper mantle source region of Mid-Ocean Ridge Basalts (MORB) arise from three sources; an elementally unfractionated component, introduced by bulk transfer from the deeper mantle via plumes; a radiogenic/nucleogenic component arising from in-situ radiogenic decay of U, Th and K, and accumulated within the upper mantle over a residence time of approximately 1.4 Ga; a subducted atmospheric component, possibly restricted to the heavy noble gases. Our new data from Icelandic sub-glacial basaltic glasses have the highest 20Ne/ 22Ne so far reported in terrestrial samples, indistinguishable from the solar wind value of 13.8. Inferred 4He/ 3He, 21Ne/ 22Ne and 40Ar/ 36Ar ratios in the source are much lower than MORB values, reflecting the presence of a deep mantle component. Measured 22Ne/ 3He ratios are higher than upper mantle values, but correlate with 21Ne/ 4He, implying a recent localised enrichment of Ne in the magma. The implied 22Ne/ 3He ratio prior to this enrichment is identical to MORB estimates. This enrichment and the parallel enrichment in Ne/He appears to require a multi stage process during partial melting, in the magma chamber, and/or during eruption which fortuitously left the ratio of mantle He to Ar and Xe unchanged. 36Ar/ 3He ratios, estimated for the source region, are significantly higher than MORB values implying the introduction of atmospheric 36Ar to the magma at some stage. The samples exhibit only a small 129Xe excess, however, the implied 129Xe */ 3He is indistinguishable from the MORB value. Overall the elemental and isotopic ratios in the Iceland samples support mantle models which predict similar primordial noble gas ratios in the MORB and OIB source regions.