CONTRIBUTIONS TO GLOBAL GEOCHEMISTRY

Osmium Isotope Geochemistry

Harvey et al. (2006) present new Os isotopic data, together with other new geochemical data, on residual peridotites recovered from Site 1274. They also provide an excellent summary of Os isotope data from previous studies of oceanic peridotites, sulfides in peridotites, and MORB glasses. All of the 20 peridotite samples analyzed record 187Os/188Os less radiogenic than proposed for "primitive upper mantle" (PUM), indicating an ancient depletion event, probably more than 1 billion years ago. Although abyssal peridotites with 187Os/188Os lower than PUM have been observed previously, the mean of previous data is close to PUM, whereas the Site 1274 data are consistently lower than PUM. The unradiogenic Os isotope ratios in Site 1274 samples range as low as 0.117, slightly lower than the previous minimum value of 0.118 observed in mid-ocean-ridge peridotites (Standish et al., 2002).

Sulfide grains from the least radiogenic peridotite sample from Site 1274 generally have 187Os/188Os consistent with the whole-rock value. A few radiogenic grains have low Os and high Re. A few single sulfide grains show lower 187Os/188Os ratios than the most depleted whole-rock values, as low as 0.114, indicating that—whereas overall Os isotope ratios may have been raised by melt-rock or seawater-rock interaction—some samples record an ancient Re depletion event. This event can be modeled as having occurred more than 2 billion years ago. These data confirm the observation of 187Os/188Os as low as 0.110 in hydrothermal fluid from the Juan de Fuca Ridge (Sharma et al., 2000), indicating a highly depleted, ancient mantle component in the mantle source of mid-ocean-ridge basalts.

It remains unclear why there is so much variability in 187Os/188Os in the Site 1274 peridotites (0.11691–0.12665). The least radiogenic sample is only 12 m from the sample with the third highest 187Os/188Os and <30 m from the most radiogenic sample. Either (1) the mantle in the melting region beneath the Mid-Atlantic Ridge is variable in 187Os/188Os on a ~10-m scale and/or (2) tectonic processes have juxtaposed peridotites from larger heterogeneous domains in the melting region and/or (3) melt/rock reaction during MORB transport from source to surface has created small-scale 187Os/188Os variation in shallow mantle peridotites and/or (4) interaction with seawater has substantially modified whole-rock Os isotope values in Site 1274 peridotites. The fact that no observed MORB 187Os/188Os values extend significantly below those proposed for the primitive upper mantle suggests that source rocks rich in radiogenic Os play an important role in mantle melting, and thus that the variability in Os isotope ratios in Site 1274 peridotite is due mainly to igneous processes (1, 2, and 3) and not to hydrothermal alteration of homogeneous residues with low Os isotope ratios.

Trace Element Geochemistry and the Pb Paradox

As illustrated in Figure F7, Godard et al. (2005) find consistent high Pb/Ce in inductively coupled plasma–mass spectroscopic (ICP-MS) data for a compilation of the least altered and/or impregnated residual peridotite compositions from Leg 209 (Paulick et al., 2006). High Pb/Ce is also observed in abyssal peridotites from mid-ocean ridges worldwide (Niu, 2004), together with Godard and Kelemen's unpublished data on peridotites from the Oman, Josephine, and Trinity ophiolites and the Jurassic Talkeetna arc (Leg 209 and Oman ICP-MS data from the Université de Montpellier; Josephine, Trinity, and Talkeetna from Washington State University; some Pb concentrations checked by isotope dilution at Woods Hole Oceanographic Institution). Although this work is not yet published, we expand upon these observations and their interpretation here.

The samples in the Godard et al. (2005) compilation have average Pb/Ce ~10 times higher than primitive mantle (Hofmann, 1988), with only 3 of 180 samples having Pb/Ce less than that in primitive mantle. REE abundance, and Ce concentration specifically, is less than that in primitive mantle in 165 of 180 samples, consistent with depletion via melt extraction, modified by some magmatic refertilization. High Pb concentrations could be due to (1) retention of Pb in residual sulfide, (2) addition of Pb in sulfide and plagioclase during "impregnation" by crystallizing melt, and/or (3) addition of Pb in sulfide and carbonate during alteration.

There is little doubt that Pb is an incompatible element during melting of the mantle to form mid-ocean-ridge basalt with a bulk rock/melt distribution coefficient similar to that of Ce, and thus "unadulterated" residues should be depleted in Pb relative to primitive mantle and should have Pb/Ce approximately equal to Pb/Ce in mid-ocean-ridge basalts. Therefore, retention of Pb in residual sulfide during melt extraction cannot be the main cause of high Pb/Ce in residual peridotites. Additionally, Pb concentration is not well correlated with compatible and moderately incompatible elements such as Ni, Cr, Ti, and heavy REE.

Pb concentration is strongly positively correlated with Th, Nb, and light REE. These elements are commonly considered "immobile" during hydrothermal alteration (e.g., Winchester and Floyd, 1977) but mobile during melt migration and igneous metasomatism. Thus, high Pb concentrations and Pb/Ce are most likely due to magmatic impregnation—"reactive fractionation" in the shallow mantle—whereas metasomatism during hydrothermal alteration probably does not play an important role in Pb enrichment.

All of the residual peridotite samples in the Godard et al. (2005) compilation, except those from the Talkeetna arc section, have Th/Pb and Th/Nb less than those in primitive mantle. This suggests that relatively high Th concentration may distinguish arc from ridge (and ophiolite) peridotites. More importantly from the perspective of global geochemistry, this indicates that recycled mid-ocean-ridge peridotites will evolve to low 208Pb/204Pb compared to the primitive mantle over long residence times in the mantle.

Most dredged mid-ocean-ridge peridotites worldwide (Niu, 2004) have high U concentrations and U/Pb higher than in primitive mantle, but most other samples in our compilation have U/Pb less than in primitive mantle. Three shallow, oxidized residual peridotites recovered via drilling during Leg 209 have high U concentrations; U concentrations in other Leg 209 residual peridotites are lower than in MORB and primitive mantle (Fig. F8). Thus, high U in dredged mid-ocean-ridge peridotites can most likely be attributed to oxidizing seafloor weathering. Given that oxidized weathering only extends tens of meters below the seafloor, Godard et al. (2005) inferred that most mid-ocean-ridge peridotites have Th/Pb and U/Pb less than in primitive mantle. If residual peridotites form with Pb isotope ratios similar to mid-ocean-ridge basalts, these rocks will evolve to 206Pb/204Pb and 207Pb/204Pb ratios less radiogenic than primitive mantle.

The effect of subduction modification on Th/Pb and U/Pb is unclear. If Pb is immobile during hydrothermal alteration—as suggested by the positive correlation of Pb concentration with Th, Nb, and light REE—then subduction modification of Pb concentrations is likely to be minor as well. If U is fluid-mobile in subduction zones and is removed by fluids evolved from subducting peridotite, this will lower U/Pb and modified residues will evolve to still lower 206Pb/204Pb and 207Pb/204Pb ratios.

As discussed elsewhere in this paper, crystallization of igneous phases from cooling melt migrating along peridotite grain boundaries may be common in the thick conductive boundary layer beneath slow-spreading ridges. These crystallizing phases likely include sulfide and plagioclase with abundant Pb. Beneath the Mid-Atlantic Ridge, such reactive fractionation of sulfide and plagioclase is likely to occur within a 20-km-thick conductive boundary layer. Based on this estimate, together with the observation that residual peridotite samples in the Godard et al. (2005) compilation contain Pb concentrations similar to those in primitive mantle, tens of percent of mantle Pb could be sequestered in a high-Pb residual peridotite reservoir over geologic time. Evolution of strongly unradiogenic Pb isotope ratios in such a reservoir, in refractory peridotites that contribute little to subsequent melting, offers a potential solution to the "first lead paradox," in which observed mid-ocean-ridge and ocean-island lavas are systematically enriched in radiogenic Pb compared to meteorites and the inferred bulk earth composition (Allègre, 1969).

Zircon Provenance (Ridge vs. Arc Zircons)

Grimes et al. (in press) report on a new method of trace element discrimination for distinguishing zircons formed at mid-ocean ridges versus those formed in arc environments. This method will be useful for determining the provenance of detrital zircons in the geologic record, including the Archean and Hadean detrital zircons that provide virtually the only samples of Earth's crust prior to 3.8 billion years ago. Grimes et al. (in press) found that, as for ridge versus arc lavas, zircons from ridge environments have low fractionation-corrected U and Th concentrations compared to zircons from arc environments. This result contrasts with that of several previous studies, most recently that of Coogan and Hinton (2006), which used REE and Ti concentration data to argue that zircons from different tectonic environments could not be distinguished using trace element data. Using the new discrimination diagrams, ancient zircons from the Jack Hills and Acasta localities show clear affinities with arc zircons and are distinct from zircon generated in a mid-ocean-ridge environment.

NEXT