Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
16. X. Yin, Z. Ye, J. Rho, Y. Wang, X. Zhang, Science 339,1405–1407 (2013).
17. J. Lin et al., Science 340, 331–334 (2013).18. X. Zambrana-Puyalto, X. Vidal, G. Molina-Terriza, Nat.
Commun. 5, 4922 (2014).19. N. Yu et al., Science 334, 333–337 (2011).20. M. Pu et al., Sci. Adv. 1, e1500396 (2015).21. A. T. O’Neil, I. MacVicar, L. Allen, M. J. Padgett, Phys. Rev. Lett.
88, 053601 (2002).22. P. Genevet, J. Lin, M. A. Kats, F. Capasso, Nat. Commun. 3,
1278 (2012).23. Materials and methods and supplementary text are
available as supplementary materials on Science Online.24. I. Fernandez-Corbaton, X. Zambrana-Puyalto, G. Molina-Terriza,
Phys. Rev. A 86, 042103 (2012).
25. I. Fernandez-Corbaton et al., Phys. Rev. Lett. 111, 060401(2013).
26. N. Zhao, X. Li, G. Li, J. M. Kahn, Nat. Photonics 9, 822–826(2015).
ACKNOWLEDGMENTS
We thank X. Li for help with the ion beam lithography,G. Gervinskas and F. Eftekhari from the Melbourne Centre forNanofabrication for their fabrication efforts, H. Lu for technicalassistance with the waveguide calculation, and J. Storteboomfor assistance with using the Spectra-Physics Inspire lasersystem. This work was supported under the AustralianResearch Council Laureate Fellowship program (grantFL100100099). M.G. acknowledges support from the AustralianResearch Council Centre for Ultrahigh Bandwidth Devices for
Optical Systems (CUDOS) (project CE110001018). X.L.acknowledges support from the Australian Research Council(grant DE150101665). All data related to the experimentsdescribed in this manuscript are archived on a lab computer atSwinburne University of Technology.
SUPPLEMENTARY MATERIALS
www.sciencemag.org/content/352/6287/805/suppl/DC1Materials and MethodsSupplementary TextFigs. S1 to S14References (27–30)
18 December 2015; accepted 9 March 2016Published online 7 April 201610.1126/science.aaf1112
GEOCHEMISTRY
Preservation of Earth-forming eventsin the tungsten isotopic compositionof modern flood basaltsHanika Rizo,1* Richard J. Walker,2 Richard W. Carlson,3 Mary F. Horan,3
Sujoy Mukhopadhyay,4 Vicky Manthos,4 Don Francis,5 Matthew G. Jackson6
How much of Earth's compositional variation dates to processes that occurredduring planet formation remains an unanswered question. High-precision tungstenisotopic data from rocks from two large igneous provinces, the North AtlanticIgneous Province and the Ontong Java Plateau, reveal preservation to thePhanerozoic of tungsten isotopic heterogeneities in the mantle. These heterogeneities,caused by the decay of hafnium-182 in mantle domains with high hafnium/tungstenratios, were created during the first ~50 million years of solar system history,indicating that portions of the mantle that formed during Earth’s primary accretionaryperiod have survived to the present.
Four and a half billion years of geologic ac-tivity have overprintedmuch of the evidencefor the processes involved in Earth’s forma-tion and initial chemical differentiation.High-precision isotopic measurements of
the decay products of short-lived radionuclidesthat were presentwhen Earth formed can providea view of events that occurred during the first tensto hundreds ofmillion years of Earth history. Datafromboth the 146Sm-142Nd (half-life, t1/2 = 103millionyears) and 129I-129Xe (t1/2 = 15.7 million years) sys-tems show the importance of early mantle dif-ferentiation and outgassing events but provideconflicting evidence about the preservation ofearly-formed mantle reservoirs to the presentday (1–4). Of the short-lived systems, the 182Hf-182W (t1/2 = 8.9 million years) system is distinc-tively sensitive to metal-silicate separation andhas been used effectively to trace the timing and
processes of core formation (5), which is arguablythe most important chemical differentiation eventto occur on a rocky planet. Only recently, however,have measurement techniques improved to the
point of resolving 182W/184W variability in an-cient (>2.7 billion years old) terrestrial rocks;such variability reflects the preservation of com-positionally distinct domains in Earth’s interiorthat were probably created during Earth’s for-mation (6–10). Young mantle-derived rocks ex-amined to date have shown neither 142Nd nor182W isotopic heterogeneity, suggesting that theearly-formed compositional domains in Earth’sinterior were largely destroyed bymantle-mixingprocesses during the first half of Earth history(1–4, 6–10). Here we report 182W/184W ratios inPhanerozoic flood basalts from Baffin Bay andthe Ontong Java Plateau, some of which areamong the highest ever measured in terrestrialrocks. These results document the preservationof regions within Earth’s interior whose compo-sitions were established by events that occurredwithin the first ~50 million years of solar systemhistory. This study, consequently, provides newinsights into the processes at work during planetformation, the chemical structure of Earth’s in-terior, and the interior dynamics that allowedthe preservation of chemical heterogeneities for4.5 billion years.Flood basalts are the largest volcanic eruptions
identified in the geological record. These typesof eruptions created both the North Atlantic
SCIENCE sciencemag.org 13 MAY 2016 • VOL 352 ISSUE 6287 809
Fig. 1. m182W values measuredfor the Baffin Bay and OntongJava Plateau samples, thegeological reference materialsVE-32 and BHVO-1, and theAlfa Aesar W standard.Thevalues are expressed as devia-tions, in parts per million (ppm),from the average valuemeasured for the W standard.The gray shaded area represents2s for the average W standardvalue. Errors for each data pointare 2s.
1Geotop, Département des Sciences de la Terre et del’Atmosphère, Université du Québec à Montréal, Montreal, Canada.2Department of Geology, University of Maryland, College Park, MD,USA. 3Carnegie Institution for Science, Washington, DC, USA.4Department of Earth and Planetary Sciences, University ofCalifornia–Davis, Davis, CA, USA. 5Department of Earth andPlanetary Sciences, McGill University, Montreal, Canada.6Department of Earth Science, University of California–SantaBarbara, Santa Barbara, CA, USA.*Corresponding author. Email: [email protected]
RESEARCH | REPORTS
on
May
15,
201
6ht
tp://
scie
nce.
scie
ncem
ag.o
rg/
Dow
nloa
ded
from
Igneous Province, which hosts the Baffin Baylocale (11), and the Ontong Java Plateau in thewestern Pacific Ocean (12). We studied pillowlavas with high-MgO picritic compositions (tableS1) from Padloping Island, Baffin Bay (samplesPi-23 and Pd-2). We targeted these rocks becausesome Baffin Bay lavas contain the highest 3He/4He ratios ever measured (13), along with Pbisotopic compositions (14) and D/H ratios (15)that indicate that their mantle source wasrelatively primitive and undegassed, consistentwith it being isolated since shortly after Earth’sformation. Ontong Java is Earth’s largestknown volcanic province and shares chemicaland isotopic similarities with the Baffin Baylavas, consistent with a primitive mantle source(16). The Ontong Java sample (192-1187A-009R-04R) is a basalt (table S1) collected from theplateau’s eastern flank by Ocean Drilling ProjectLeg 192.We present data from the short-lived 182Hf-
182W and 146Sm-142Nd systems, because these twosystems are variably sensitive to the core forma-tion and mantle differentiation processes that oc-curred early in Earth history. We compare thesedata with data from the long-livedU-Th-He, 147Sm-143Nd, and 187Re-187Os isotope systems, togetherwithWandhighly siderophile element (HSE; Re,
Os, Ir, Ru, Pt, and Pd) concentrations, to betterdistinguish early differentiation events from thoseoccurring over all of Earth history.Glassy rim and core pieces of sample Pi-23 (Pi-
23a and Pi-23b, respectively), a bulk sample ofPd-2, and a bulk sample of 192-1187A-009R-04Rare characterized by high 182W/184W ratios thatare well resolved from standards, with m182Wvalues ranging from +10 to +48 {where m182W =[(182W/184W)sample/(
182W/184W)standard – 1] × 106}(Fig. 1 and Table 1) (17). The m182W values forsamples Pi-23a and Pi-23b are in good agree-ment. This rules out the influence of stable Wisotope fractionation through interaction ofseawater with the pillow rim in creating themeasured 182W values (17). Samples 192-1187A-009R-04R and Pd-2 had the lowest W concen-trations and the highest m182W values (Table 1).The W concentrations of these two samples [23and 26 parts per billion (ppb), respectively] arebroadly consistent with those ofmagmas derivedby 15-to-20% partial melting of a mantle sourcewith ~5 ppbW, indicating a primitive source freeofW-rich recycled crust (fig. S2) (17). The geologicalreference materials VE-32 (mid-ocean ridge glass)and BHVO-1 (Hawaiian basalt) weremeasured atthe same time as the Baffin Bay and Ontong Javasamples and yielded m182W values of –0.8 ± 4.5
and –2.3 ± 7.7, respectively (Table 1). These m182Wvalues are indistinguishable from the terrestrialAlfa Aesar W standard (m182W = 0) and othermodern rocks (6–10). The 3He/4He ratios mea-sured in olivines from the Baffin Bay samples(table S3) (17) yielded values up to 48.4 timeshigher than the 3He/4He ratio of the atmosphere(1.39 × 10−6), in agreement with previous findings(13), indicating that the source of these lavasis relatively undegassed and possibly has beenisolated since Earth’s formation. The HSE abun-dances and initial 187Os/188Os ratios (which pro-vide a record of the long-term Re/Os ratio) fromthe Baffin Bay andOntong Java Plateau samplesare indistinguishable from other modern mantle-derived lavas with similar MgO abundancesthat do not show elevated m182W (Fig. 2 and tableS4) (18).Variability in 182W/184W ratios reflects Hf/W
fractionation while 182Hf was extant. Hf/W frac-tionation has been observed in early solar systemmaterials (5), so variable W isotopic compositionsin terrestrial samples can reflect the imperfectmix-ing of late additions of such materials (6, 9). Them182W value of +48 for Baffin Bay sample Pd-2,however, is larger than can be accounted for bythis process, and so this possibility is discounted(supplementary text). Hf/W fractionation can alsooccur as the result of endogenous Earth differen-tiation processes, such as magma ocean crystalli-zation (7) and core formation (9). However, silicatefractionation processes cannot be responsible forthe generation of the anomalous 182W in the sourcesof the Baffin Bay and Ontong Java lavas. If the highm182Wwas due to silicate fractionation in a magmaocean while 182Hf was extant, then m182W shouldpositively correlatewith m142Nd, the decay productof the short-lived 146Sm (t1/2 = 103 million years)isotope system. Instead, the m142Nd values ofthe samples are indistinguishable from all other
810 13 MAY 2016 • VOL 352 ISSUE 6287 sciencemag.org SCIENCE
Fig. 2. HSE abundances forthe Baffin Bay and OntongJava Plateau (OJP) samples.Abundances are normalizedto the HSEs of carbonaceouschondrites (CI group) from (25).The gray shaded areashows the range of HSE abun-dances for type-2 Hawaiian pic-rites (26).
Fig. 3. Model for the creation of mantle reservoirs with distinct Wisotopic compositions. (A) Early core formation leaves the proto-Earth’smantle with a high Hf/W ratio that, with time, evolves to a high m182W value (i).(B) The impact of a largebodyaffects theHf/Wratio andW isotopic compositionof a portion of the proto-Earth’smantle. (C) Evolution of the portion of themantle(ii) affected by the impact of a large body, involving some degree of isotopicequilibration between the impactor materials and the mantle. The core ofthe impactor subsequently merges with the core of the proto-Earth. (D) Pos-
sible scenario after isostatic adjustment (27) and creation of a mantle withheterogeneous m182W through impacts of large bodies.Mantle domains affectedby impacts that occur after the extinction of 182Hf no longer generate ra-diogenic 182W, so their 182W/184W ratios can change only by mixing with otherterrestrial reservoirs orwith late-accreted chondriticmaterial. (E) Late accretion,representing ~0.5% of Earth’s mass, decreases the 182W/184W ratio of all theearlier-formed reservoirs by ~15 ppm. This last accretion is responsible forendowing themodernmantle with chondritic relative abundances of the HSEs.
RESEARCH | REPORTS
on
May
15,
201
6ht
tp://
scie
nce.
scie
ncem
ag.o
rg/
Dow
nloa
ded
from
modern basaltsmeasured so far (fig. S3 and tableS5) (17).This leaves Hf/W fractionation resulting from
metal-silicate segregation accompanying core for-mation as the probable cause of the observedanomalies in the Phanerozoic samples. Metal-silicate segregation is the process that can frac-tionate Hf/W most effectively, because Hf is astrongly lithophile trace element, whereas W ismoderately siderophile. The lowW concentrationsestimated for the mantle source of the floodbasalts described here are consistent with man-tle domains that experienced metal-silicate seg-regation (Table 1 and supplementary text).Repeated metal-silicate segregation events dur-ing planet formation (19) could create one ormore mantle domains with distinct m182W, with-out affecting the Sm-Nd system. Such eventswould result in variable m182W, due to metal-silicate segregation events occurring at differ-ent times (Fig. 3), or, alternatively, different Hf/Wratios in the resulting mantle reservoirs thatreflect different oxidation states and, hence,different partitioning of W into metal (supple-mentary text).The key observation to consider, which is sup-
ported by the results reported here, is that ter-restrial samples with 182W excesses do not seemto derive from sources depleted in HSEs (Fig. 2)(6–10). HSEs have partition coefficients betweenmetal and silicate of >104 (20); thus, their con-centrations in metal-depleted mantle domainsare expected to be very low. Evolving oxidationstates during Earth accretion might explain thedecoupling of 182W and HSEs, because althoughW becomes less siderophile under oxidizingconditions, the HSEs, even at high oxidationstates, are not soluble in silicates (20). Thismodel, however, would require subsequent late-accreted HSEs to have been mixed into dif-ferent mantle domains without themixing awayof W isotopic heterogeneity. Alternatively, theobserved decoupling could be explained if somemetal from the core of the Moon-forming giantimpactor was retained in the mantle, followedby a minor amount of late accretion (9). Thismodel would require a substantial mass of high-
density metal to have been retained in Earth'smantle after the impact, when the mantle waspartially or wholly molten, and this retainedmetal would have to have contained chondriticrelative abundances of the HSEs. These modelsare summarized in the supplementary text, alongwith a few additional potential explanations forthe apparent decoupling of W isotopic composi-tions and HSE abundance variations.Regardless of the origin of the 182W variability,
what is arguably more unexpected than the factthat Earth experienced such early differentiationevents is that reservoirs formed by these earlyprocesses remain in the mantle today. This con-clusion is now supported by isotopic variabilityin both W and Xe (1, 21), but not in 142Nd, whichsuggests that the observed heterogeneity in142Nd/144Nd ratios was reduced to an unobserva-ble level by the end of the Archean, probablythrough the mixing caused by mantle convection.Perhaps the key to reconciling these observationsis that the 129I-129Xe system primarily reflects man-tle outgassing, and the 182Hf-182W system reflectsmetal-silicate separation,whereas the 146Sm-142Ndsystem is controlled by internal mantle differ-entiation. For both W and Xe, one componentof the complementary chemical differentiation(the core for W and the atmosphere for Xe)may not be available for effective recycling andmixing in the mantle (22). In contrast, for Nd,the main reservoirs created during early Earthdifferentiation may have been in a portion of themantle that has since been effectively mixed bymantle circulation. Estimates for how much ofthe mantle can remain unmixed depend on therheological properties assigned to the variousmaterials involved. Some models (23) show thatas much as 20% of the mantle may remain iso-lated as distributed masses. An important aspectof the results presented here is that both 182Wanomalies and elevated 3He/4He ratios (table S3)appear in at least two major flood basalts. Floodbasalt eruptions require the melting of large vol-umes of mantle during unusual thermal eventsin the history of mantle circulation. The large sizeand sporadic nature of flood basalt eruptions areperhaps indicative of a layer in the mantle whose
density or rheological properties keep it from ef-fectively mixing with the rest of the mantle. Thelarge low–shear velocity provinces (LLSVPs) thathave been imaged at the base of the mantle (24)may constitute such a reservoir. These regions ap-pear to be warmer and compositionally differentfrom the surrounding mantle. Estimates of theirvolume range to as high as 7% of the mantle, oron the order of 6 × 1010 km3. If the LLSVPs areremnants of early differentiation events on Earth,they must have a delicately balanced density con-trast relative to the surrounding mantle to allowtheir survival through 4.5 billion years of dy-namic Earth history.
REFERENCES AND NOTES
1. S. Mukhopadhyay, Nature 486, 101–104 (2012).2. M. Boyet, M. O. Garcia, R. Pik, F. Albarede, Geophys. Res. Lett.
32, L04306 (2005).3. V. C. Bennett, A. D. Brandon, A. P. Nutman, Science 318,
1907–1910 (2007).4. H. Rizo, M. Boyet, J. Blichert-Toft, M. T. Rosing, Earth Planet.
Sci. Lett. 377–378, 324–335 (2013).5. T. Kleine et al., Geochim. Cosmochim. Acta 73, 5150–5188
(2009).6. M. Willbold, T. Elliott, S. Moorbath, Nature 477, 195–198
(2011).7. M. Touboul, I. S. Puchtel, R. J. Walker, Science 335, 1065–1069
(2012).8. M. Touboul, J. Liu, J. O'Neil, I. S. Puchtel, R. J. Walker, Chem.
Geol. 383, 63–75 (2014).9. M. Willbold, S. J. Mojzsis, H.-W. Chen, T. Elliot, Earth Planet.
Sci. Lett. 419, 168–177 (2015).10. H. Rizo et al., Geochim. Cosmochim. Acta 175, 319–336
(2016).11. A. D. Saunders, J. G. Fitton, A. C. Kerr, M. J. Norry, R. W. Kent,
Geophys. Monogr. Ser. 100, 45–94 (1997).12. C. R. Neal, J. J. Mahoney, L. W. Kroenke, R. A. Duncan,
M. G. Petterson, Geophys. Monogr. Ser. 100, 183–216(1997).
13. N. A. Starkey et al., Earth Planet. Sci. Lett. 277, 91–100(2009).
14. M. G. Jackson et al., Nature 466, 853–856 (2010).15. L. J. Hallis et al., Science 350, 795–797 (2015).16. M. G. Jackson, R. W. Carlson, Nature 476, 316–319 (2011).17. Materials and methods are available as supplementary
materials on Science Online.18. C. W. Dale et al., Earth Planet. Sci. Lett. 278, 267–277 (2009).19. D. Rubie, F. Nimmo, H. J. Melosh, “Formation of Earth’s core,”
in Treatise on Geophysics, G. Schubert, Ed. (Elsevier, ed. 2,2015), pp. 51–90.
20. H. St. C. O’Neill, D. B. Dingwell, A. Borisov, B. Spettel, H. Palme,Chem. Geol. 120, 255–273 (1995).
SCIENCE sciencemag.org 13 MAY 2016 • VOL 352 ISSUE 6287 811
Table 1. Tungsten concentrations and isotopic compositions. Included are data from the Baffin Bay samples Pi-23a, Pi-23b, and Pd-2; the Ontong Java Plateau
sample 192-1187A-009R-04R; the mid-ocean ridge glass sample VE-32; and the BHVO-1 basalt standard. Uncertainties are ±2s. More details are given in (17) and
table S2.
Locality Geodynamic
context
Sample m182W
(ppm)
±2s
(ppm)
W
(ppb)
Baffin Bay Flood basalt Pi-23a 11.9 5.9 62... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ...
Baffin Bay Flood basalt Pi-23b 8.3 5.6 62... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ...
Baffin Bay Flood basalt Pd-2 48.4 4.6 26... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ...
Ontong Java
PlateauFlood basalt
192-1187A-
009R-04R23.9 5.3 23
... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ...
East Pacific RiseMid-ocean
ridge basaltVE-32 –0.8 4.5 54
... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ...
Hawaii Ocean island basalt BHVO-1 –2.3 7.7 274... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... .. ... ... .. ... ... .. ... ... .. ... .. ... ... .. ...
RESEARCH | REPORTS
on
May
15,
201
6ht
tp://
scie
nce.
scie
ncem
ag.o
rg/
Dow
nloa
ded
from
21. M. K. Pető, S. Mukhopadhyay, K. A. Kelley, Earth Planet. Sci.Lett. 369–370, 13–23 (2013).
22. M. G. Jackson, S. B. Shirey, E. H. Hauri, M. D. Kurz, H. Rizo,Geochim. Cosmochim. Acta 10.1016/j.gca.2016.02.011 (2016).
23. J. P. Brandenburg, E. H. Hauri, P. E. van Keken, C. J. Ballentine,Earth Planet. Sci. Lett. 276, 1–13 (2008).
24. E. J. Garnero, A. K. McNamara, Science 320, 626–628(2008).
25. M. Horan, R. J. Walker, J. W. Morgan, J. N. Grossman,A. E. Rubin, Chem. Geol. 196, 27–42 (2003).
26. T. J. Ireland, R. J. Walker, M. O. Garcia, Chem. Geol. 260,112–128 (2009).
27. W. B. Tonks, H. J. Melosh, J. Geophys. Res. Planets 98,5319–5333 (1993).
ACKNOWLEDGMENTS
We thank T. Mock for assistance with mass spectrometry.This work was improved by helpful discussions with J. O’Neil,S. Shirey, and B. Wood. We thank S. Shirey, who providedsample VE-32. We also appreciate the helpful comments andsuggestions from three anonymous reviewers. We thankM. Garçon, who developed the four-step 142Nd acquisitionmethod. This work was supported by NSF grant EAR-1265169(Cooperative Studies of the Earth’s Deep Interior program) to
R.J.W. and grant EAR-1250419 to S.M. All data are available inthe main manuscript and supplementary materials.
SUPPLEMENTARY MATERIALS
www.sciencemag.org/content/352/6287/809/suppl/DC1Materials and MethodsFigs. S1 to S6Tables S1 to S5References (28–56)
12 November 2015; accepted 5 April 201610.1126/science.aad8563
SLEEP RESEARCH
Causal evidence for the role of REMsleep theta rhythm in contextualmemory consolidationRichard Boyce,1 Stephen D. Glasgow,2 Sylvain Williams,2*† Antoine Adamantidis2,3*†
Rapid eye movement sleep (REMS) has been linked with spatial and emotional memoryconsolidation. However, establishing direct causality between neural activity during REMSand memory consolidation has proven difficult because of the transient nature of REMSand significant caveats associated with REMS deprivation techniques. In mice, weoptogenetically silenced medial septum g-aminobutyric acid–releasing (MSGABA) neurons,allowing for temporally precise attenuation of the memory-associated theta rhythmduring REMS without disturbing sleeping behavior. REMS-specific optogenetic silencing ofMSGABA neurons selectively during a REMS critical window after learning erasedsubsequent novel object place recognition and impaired fear-conditioned contextualmemory. Silencing MSGABA neurons for similar durations outside REMS episodes had noeffect on memory. These results demonstrate that MSGABA neuronal activity specificallyduring REMS is required for normal memory consolidation.
The physiological function of rapid eyemove-ment sleep (REMS) is unclear (1). Evidencelinking REMS to aspects of memory con-solidation in mammals has been obtainedusing techniques such as statistical corre-
lation, pharmacology, and REMS deprivation(2, 3). However, whether REMS has a directrole in learning and memory remains contro-versial; correlative studies are not definitive,REMS has a transient pattern of occurrencethat prevents REMS-selective pharmacologicalmanipulation, and REMS deprivation has crit-ical caveats that are difficult to fully control for(4, 5).During REMS in mice and rats, a promi-
nent ~7-Hz theta oscillation is observed in lo-cal field potential (LFP) recordings from corticalstructures, including the hippocampus (6, 7).Hippocampal theta rhythms during REMS may
contribute to memory consolidation by pro-viding a mechanism for strengthening placecells formed during prior wakefulness (8, 9).Theta rhythm generation requires an intactmedial septum (MS) (10, 11), although the MSis not involved in REMS generation itself (12, 13).MS g-aminobutyric acid–releasing (MSGABA)neurons project to the hippocampus, prob-ably pacing the hippocampal theta rhythmduring REMS (14–16). In mice, we thereforeused optogenetics to silence MSGABA neuronsand reduce theta activity selectively duringREMS, without disturbing sleeping behavior,to determine whether intact MSGABA neuralactivity during REMS is important for memoryconsolidation.Adeno-associated virus (AAVdj) encoding Arc-
haerhodopsin fused to an enhanced yellow fluo-rescent protein (eYFP) reporter (ArchT-eYFP)was injected into the MS of VGAT::Cre mice (Fig.1A, top). The resulting ArchT-eYFP expressionwas ~95% specific for MSGABA neurons (Fig. 1A,bottom, and fig. S1A), stable, and localized to theMS and the diagonal band of Broca (DBB) regionfor at least 3months after injection.MSGABA neuralprojections were observed throughout the hippo-campus (Fig. 1B).
Whole-cell voltage and current clamp record-ings of ArchT-eYFP–expressing MS neurons inacute brain slices (fig. S1B) revealed hyperpolar-ization (−39.9 ± 6.6 mV) and outward current(293.9 ± 69.2 pA) upon 594-nm light exposure(fig. S1B). Single-unit recordings in behavingtransfected mice (fig. S1C) confirmed that photo-inhibitionduringREMS,non-REMsleep (NREMS),and wakefulness rapidly produced a potent andreversible reduction in spiking of putative MSGABA
neurons (fig. S1D).We next tested the effect of silencing MSGABA
neurons during REMS in freely behaving mice.Photoinhibition with constant light pulses deliv-ered to the MS in ArchT-eYFP–expressing mice(ArchTmice) resulted in significantly (65.3 ± 5.6%)reduced theta power measured from dorsal hippo-campal area CA1 LFP (CA1LFP) recording (Fig. 1D,top). No other frequency bands were affected, andthe spectral profile of the CA1LFP returned tobaseline levels almost immediately upon releaseof MSGABA neurons from photoinhibition (Fig.1D, top, and Fig. 2A). Current source density(CSD) analysis revealed that reduced theta powerwas present in all layers of dorsal hippocampalCA1 (Fig. 2B). Light pulses delivered to the MSof mice only expressing eYFP inMSGABA neurons(YFP control mice) did not affect CA1LFP power(Fig. 1D, bottom), ruling out light as a potentialconfounding factor in these results. Inhibition ofMSGABA neurons did not perturb sleeping behav-ior (Fig. 1D, top), and the probability of statetransition during REMS in ArchT mice was un-altered relative to YFP control mice (n = 30 ArchTmice,n = 19 YFP control mice; P = 0.63, unpairedStudent’s t test).We optogenetically silenced MSGABA neurons
selectively during each REMS episode after ac-quisitionof anovel object place recognition (NOPR)task (Fig. 3A). Mice showed no preference foreither object during initial object habituation[day 1 (D1), task acquisition] (Fig. 3A, right).After acquisition, EEG/CA1LFP/EMG (EEG, elec-troencephalogram; EMG, electromyogram) ac-tivity wasmonitored for 4 hours. Upon entry intoREMS, mice in the ArchT or YFP control grouphad light continuously delivered to the MS untiltransition to another state occurred, at whichtime light delivery ceased until subsequent REMSwas detected (Fig. 3B). A group of ArchT-eYFPexpressing mice that never received light (ArchTcontrol) served as a baseline control for ArchT-eYFP transfection. To determine whether REMSwas a critical factor in our results, a final group
812 13 MAY 2016 • VOL 352 ISSUE 6287 sciencemag.org SCIENCE
1Integrated Program in Neuroscience, McGill University,Montreal, Quebec, Canada. 2Department of Psychiatry, McGillUniversity, Montreal, Quebec, Canada. 3Department ofNeurology and Department of Clinical Research, InselspitalUniversity Hospital, University of Bern, Bern, Switzerland.*Corresponding author. Email: [email protected](S.W.); [email protected] (A.A.) †These authorscontributed equally to this work.
RESEARCH | REPORTS
on
May
15,
201
6ht
tp://
scie
nce.
scie
ncem
ag.o
rg/
Dow
nloa
ded
from
INSIGHTS | PERSPECTIVES
768 13 MAY 2016 • VOL 352 ISSUE 6287 sciencemag.org SCIENCE
GR
AP
HIC
: V
. A
LT
OU
NIA
N/SCIENCE
By Tais W. Dahl
The chemical composition of Earth’s
mantle can tell us how our planet
formed and how subsequent mantle
dynamics have since homogenized the
mantle through convective processes.
Most terrestrial rocks have a similar
tungsten (W) isotope composition (1), but
some rocks that have been dated at 2.8 Ga
(billion years old) (2), 3.8 Ga (3),
and 3.96 Ga (4) have elevated 182W/184W ratios. This is reported
as µ182W, in parts per million
(ppm) deviation from the bulk
silicate Earth. Until now, the out-
liers have included only these an-
cient rock samples with a small
µ182W excess (≤15 ppm) that can
be attributed to the final ~0.5%
of Earth’s mass that accreted
late in its accretion history. On
page 809 of this issue, Rizo et al.
(5) report W isotope data from
young mantle-derived rocks with
µ182W excesses of 10 to 48 ppm.
This result is spectacular be-
cause the range of µ182W values
in mantle-derived rocks is larger
than can be accommodated by
late accretion; the implication is
that remnants of Earth’s earliest
mantle have been preserved over
the entirety of Earth’s history.
Two decades ago, geochemists
started measuring W isotopes
in terrestrial and meteoritic
samples to estimate the tim-
ing of Earth’s core formation
(6). During core formation, it is
presumed that W preferentially
partitioned into the metal melt
and sank to the core, whereas
hafnium (Hf) remained in the
silicate mantle. Therefore, the
silicate mantle has a higher
Hf/W ratio. As 182Hf decayed to 182W with a half-life of 8.9 mil-
lion years, the silicate mantle
developed 182W excesses relative
to its source material. Most of
the µ182W variability in the solar system is
produced by the separation of metal from
silicate during the first 60 million years of
solar system history. As a point of reference
for calibration, mantle material would have
gained a µ182W excess of >1000 ppm if Earth
and its core formed at the very beginning of
our solar system. The mantle, however, dis-
plays a nearly uniform µ182W excess of +200
ppm relative to primitive meteorites, which
suggests that the last metal-silicate equilibra-
tion event occurred ~30 million years into
the history of our solar system.
The Moon formed after 182Hf went extinct
~4.51 billion years ago. This inference is
based on the tiny µ182W excess in the Moon
relative to Earth’s mantle, +20 ppm (7). The
small 182W excess may well have developed by
disproportionate late accretion on Earth and
the Moon (7), where the addition of nonra-
diogenic material had a greater
effect on Earth.
The new data from flood ba-
salt lavas that erupted into the
North Atlantic Igneous Prov-
ince (Baffin Bay locale) and
the Ontong Java Plateau (west-
ern Pacific Ocean) show larger
µ182W variability than between
the Moon and Earth’s mantle.
In fact, the mantle source of the
Ontong Java and Baffin Bay la-
vas prior to late accretion prob-
ably had a radiogenic W excess
of +70 ppm. This exceeds what
can be accomplished by late ac-
cretion. By implication, these
parts of Earth’s mantle predate
the Moon and did not chemi-
cally equilibrate metal and
silicate during the giant impact
that formed the Moon.
W isotope heterogeneity in the
mantle is likely to arise during
the late stages of planet accre-
tion. The planet accreted from
dozens of collisions between dif-
ferentiated bodies with metallic
cores and silicate mantles, all
with potentially unique W iso-
tope compositions (8). The cores
contained nonradiogenic W,
whereas the radiogenic counter-
part was hosted in their mantles.
As the planet accreted, Earth’s
silicate mantle would have lost
both radiogenic and nonradio-
genic W to its core, assuming
that an appreciable portion of
each impactor’s metal core disin-
tegrated into smaller blobs and
mixed down to diffusion-length
scales where chemical equilibra-
tion occurred.
Several factors may lead to
mantle heterogeneity. First, the
GEOCHEMISTRY
Identifying remnants of early EarthIsotope analysis reveals portions of Earth that have remained the same since accretion
Upper mantle
Inner core
Outer core
Lower mantleAfrica
Low–shear velocity provinces
South America
North Atlantic Igneous Province
Ontong Java Plateau
182Hf
184W
182W
W sequestered into the core with sinking metal, although some stayed in the mantle because of incomplete metal-silicate equilibration
Hf and W began in a fxed proportion in the mantle
Hf decayed into W
Equatorial cross section of Earth. Large low–shear velocity provinces (LLSVPs)
occur at the base of the mantle under Africa and the Pacific Ocean. Flood basalts
erupted at the North Atlantic Igneous Province and the Ontong Java Plateau show
substantial radiogenic W excess relative to rest of the mantle. This compositional
heterogeneity may preserve remnants after incomplete metal-silicate equilibration
during Earth’s core formation in the first 60 million years of solar system history. The
chemical heterogeneity may have been preserved in LLSVPs at the base of the mantle
for more than 4.5 billion years.
Natural History Museum of Denmark, University of Copenhagen, DK-1350 Copenhagen K, Denmark. Email: [email protected]
Published by AAAS
on
May
15,
201
6ht
tp://
scie
nce.
scie
ncem
ag.o
rg/
Dow
nloa
ded
from
13 MAY 2016 • VOL 352 ISSUE 6287 769SCIENCE sciencemag.org
far-side effect generates heterogeneity on a
hemispherical scale, because an impact hits
only one side of the planet. Second, portions
of the mantle may have been isolated from
mixing with the rest of the mantle. Perhaps
such a hidden reservoir has existed at the
base of the mantle (see the figure). Alter-
natively, self-gravitating cores may have
plunged through the mantle without ap-
preciable emulsification, thus preventing
enough metal to mix with the silicate mantle
down to small enough length scales for chem-
ical equilibration to occur (9).
What is surprising is that such an initial
W isotope heterogeneity could be preserved
in the mantle today. The mantle source res-
ervoir for the flood basalts is a matter of
contention. One candidate is the large low–
shear velocity provinces (LLSVPs) at the
base of the mantle (see the figure). These
regions consist of dense, hot material (10)
and are thought to be stable at their pres-
ent antipodal position on the core-mantle
boundary close to the equator (11). Rizo et
al. suggest that these regions have existed
for essentially all of Earth’s history.
Previous analyses of oceanic basalts
thought to be derived from deep mantle
plumes, including samples from the On-
tong Java Plateau, found no 182W excesses
(±5 ppm) (2). If we accept both studies at
face value, then the chemical heterogeneity
apparently exists within a single source do-
main, excluding W heterogeneity created by
the far-side effect.
Future studies must first rule out the pos-
sibility of analytical artifacts, because sev-
eral processes may obscure high-precision
isotope analyses of W poor samples (for
example, lavas), including isotope fraction-
ation in the laboratory, contamination, and
interferences. In the end, identifying primi-
tive signatures in Earth’s mantle today eluci-
dates the style of Earth’s core formation and
requires certain parts of the mantle to stay
sufficiently buoyant to escape convective
mixing throughout essentially all of Earth’s
geodynamical history. j
REFERENCES
1. D. C. Lee, A. N. Halliday, Science 274, 1876 (1996). 2. M. Willbold, T. Elliott, S. Moorbath, Nature 477, 195 (2011). 3. M. Touboul, I. S. Puchtel, R. J. Walker, Science 335, 1065
(2012). 4. M. Willbold, S. J. Mojzsis, H. W. Chen, T. Elliott, Earth Planet.
Sci. Lett. 419, 168 (2015). 5. H. Rizo et al., Science 352, 809 (2016). 6. A. Halliday, M. Rehkamper, D. C. Lee, W. Yi, Earth Planet. Sci.
Lett. 142, 75 (1996). 7. M. Touboul, I. S. Puchtel, R. J. Walker, Nature 520, 530
(2015). 8. J. E. Chambers, Earth Planet. Sci. Lett. 223, 241 (2004). 9. T. W. Dahl, D. J. Stevenson, Earth Planet. Sci. Lett. 295, 177
(2010). 10. R. G. TrØnnes, Mineral. Petrol. 99, 243 (2009). 11. K. Burke, Annu. Rev. Earth Planet. Sci. 39, 1 (2011).
10.1126/science.aaf2482
EVOLUTION
Toward a prospective molecular evolution
By Xionglei He and Li Liu
The field of molecular evolution is con-
cerned with evolutionary changes in
genes and genomes and the underlying
driving forces behind those changes.
Current studies in molecular evolution
are almost entirely retrospective, with
a focus on the mutations that were fixed dur-
ing evolution, and the conclusions are often
explanatory, offering no predictive insights.
Because only a tiny fraction of all mutations
that have ever occurred during evolution
have been fixed, the “successes” that we see
today provide an incomplete or even biased
understanding of the evolutionary process.
One way to circumvent this problem is to ob-
tain the whole fitness landscape of a gene to
understand, prospectively, chance and neces-
sity in evolution (see the figure). Two studies
in this issue, by Li et al. on page 837 (1) and
Puchta et al. on page 840 (2), each take on
this challenge by characterizing the in vivo
fitness landscape of two RNA genes.
The enormous mutational space of a typi-
cal gene poses a considerable challenge to
the characterization of fitness landscapes.
For example, there are a total of 4100 (or
1060) possible variants for an RNA gene of
100 nucleotides, and 20100 (or 10130) for a
protein of 100 amino acids. The advent of
second-generation DNA sequencing goes
some way toward addressing this problem.
For example, in 2010 a pioneering study
used second-generation DNA sequencing to
measure the in vitro biochemical activities
of millions of variants of a gene (3). Li et al.
and Puchta et al. both go one step further
than this in vitro study. Li et al. generated
a library comprising >65,000 mutant al-
leles of a 72-nucleotide transfer RNA (tRNA)
gene of the yeast Saccharomyces cerevisiae.
They then competed all the yeast strains—
each carrying a different allele of the tRNA
gene—in a liquid coculture. The frequency
of each allele was determined by sequencing
the amplicons of the target tRNA gene pool.
The increase or decrease in the frequency of
a mutant allele relative to the wild-type al-
lele during the competition represents the
relative fitness of the allele. In the second
study, Puchta et al. adopted largely the same
strategy and estimated the relative fitness of
~60,000 mutant alleles of a 333-nucleotide
small nucleolar RNA (snoRNA) gene, also in
yeast. Although the number of mutants they
examined is still a small fraction of all pos-
sible variants of the genes, most of the possi-
ble genotypes that differ from the wild-type
by one or two point mutations were char-
acterized. Thus, a high-quality local fitness
landscape of a gene has been constructed.
The capability to map large fitness land-
scapes opens the door to study gene evolu-
tion prospectively. Both studies used the
landscape constructed to understand the
various conservation levels of different sites
in the two RNA genes. The same idea could
be applied to the study of among-gene differ-
ences in evolutionary rate. An outstanding
question here is why expression level acts as
the most important determinant of sequence
conservation (4). Mapping fitness landscapes
of the same gene expressed at different levels,
or different genes expressed at the same level,
by manipulating promoter activity would
help test many competing hypotheses (5).
One of the most intriguing aspects of a
fitness landscape is epistasis, or the interac-
tions seen between mutations. Both studies
observed widespread epistasis, especially
negative epistasis, meaning that the com-
bined deleterious effect of two harmful muta-
tions is greater than that expected from the
individual mutations (6). Sign epistasis (7),
where a deleterious mutation becomes ben-
eficial in the presence of another mutation,
is of special interest because it can generate
fitness peaks that trap an adapting popula-
“One of the most intriguing aspects of a fitness landscape is…the interactions seen between mutations.”
Fitness landscapes provide a prospective understanding of chance and necessity in evolution
The State Key Laboratory of Biocontrol, School of Life Sciences, Sun Yat-sen University, Guangzhou 510275, China. Email: [email protected]; [email protected]
Published by AAAS
on
May
15,
201
6ht
tp://
scie
nce.
scie
ncem
ag.o
rg/
Dow
nloa
ded
from
2
Materials and Methods
1. Major and minor element concentrations
Whole rock major element compositions were determined by X-ray fluorescence (XRF) on fused disks made from powders at Franklin and Marshall College (see ref. 28 for procedure). Totals of major element oxides were 100.8 ± 0.2 wt.%. Analytical error was usually better than 1% (1σ) for the major elements with concentrations greater than 1%, and better than 5% (1σ) for the remaining major and minor elements. The results are shown in Table S1.
2. Sample preparation, W concentrations and isotopic measurements
Rock samples were carefully crushed in ceramic jaw crushers to avoid any metal contamination. The rock chips were then inspected under a binocular microscope, and only the freshest pieces of rock sample were handpicked for study. The rock chips were then leached with ~ 1M HCl for ~ 30 minutes, rinsed with ultraclean water, and then dissolved. Tungsten concentrations were determined by isotope dilution, using between 100 to 200 mg of sample that was spiked with 186W tracer and dissolved in concentrated HF-HNO3. After complete dissolution, tungsten was separated from the bulk sample using an anion column filled with 1ml of AG1-X8 resin. Tungsten concentrations were measured using a Thermo ICAP Q at the Department of Terrestrial Magnetism (DTM) of the Carnegie Institution for Science. Total chemistry blanks were also determined by isotope dilution in the ICAP Q, and were ~ 0.25 ng, which represents ~ 7% of the quantity of W analyzed by isotope dilution for the sample with the lowest W concentration.
For high precision W isotope analyses, between 11 to 44 g of rock chips were dissolved to enable the extraction of at least 500 ng of W from each sample. The samples were dissolved in batches of ~ 6 g in closed 120 ml Savillex digestion vessels using 5 ml concentrated HNO3 and 25 ml concentrated HF, and placed on a hotplate at ~120 °C for 4 days. The samples were then dried down and re-dissolved several times in concentrated HNO3 to remove fluorides until the samples provided clear solutions when dissolved in 6M HCl. The chemistry protocol followed here was the one developed by Touboul and Walker (29) that uses first a cation resin column step to extract the W together with high field strength elements from the bulk sample, followed by four steps using anion resin columns to purify the W fraction and ensure the removal of Ti, which inhibits W ionization during mass spectrometry. The W recovery after chemistry was between 60% and 90%. Samples with the lowest yields were Pd-2 and OJP, where 34 g and 44 g of sample powder had to be processed. The low yields could have been caused by the loss of W through the leaching. Approximately 1600 ng, 685 ng, 1215 ng, 440 ng and 544 ng were collected from samples VE-32, Pi-23a, Pi-23b, Pd-2 and OJP, respectively. The total chemistry blank was measured by isotope dilution in the Thermo ICAP Q and was a
3
maximum of 14 ng, which represents less than 3% of the total W analyzed for the samples.
Tungsten isotope compositions were measured on Re filaments as WO3-, with La acting
as electron emitter. Rhenium, which causes oxide interference on 186WO3, was monitored during measurements to allow correction for the interference. The average 185Re/184W signal intensity ratio of the runs was 0.1 for both samples and standards. The exception is sample Pi-23a that ran with an average 185Re/184W of 0.3, but this sample was duplicated (Pi-23b, with a 185Re/184W of 0.06) and μ182W values results agree with the first analysis within error. Osmium, which has an isotope (184Os, 0.02% abundance) that could cause interference on 184W was also monitored, but was always below the limit of detection.
When measuring W isotopic composition by TIMS, 182W/184W vs. 183W/184W ratios of the standards and samples show a positive correlation (Fig. S1 and refs. 7, 8, 10, 29). This correlation appears after correction of W isotope ratios for instrumental fractionation and oxygen isotopic composition, and is attributed to the effects of mass dependent oxygen isotope fractionation on the W ratios (29). Figure S1 shows the 182W/184W vs. 183W/184W correlations obtained for the Alfa Aesar W standard solution for each of the three analytical sessions performed for this study: B587, B594, and B622. The regression slopes yielded the following values, respectively: 2.0804, 2.090, and 2.1558. These slopes and a ratio of 183W/184W= 0.416729 were used to correct the 182W/184W of the samples and the standards for the second-order oxygen isotope fractionation effect. After this correction, the 182W/184W ratio for the standards was 0.864766 ± 0.000003 (2σ, n=4) 0.864769 ± 0.000004 (2σ, n=5), and 0.864765 ± 0.000003 (2σ, n=10) for sessions B587, B595 and B622, respectively. Detailed results are shown in Table S2.
Willbold et al. (6) reported W isotope data for a sample from the Ontong-Java Plateau and the related Manihiki Plateau. Neither sample showed a resolvable anomaly, but both were measured only twice using the MC-ICPMS technique employed in that study, whereas the majority of samples analyzed in that study were measured at least 5 times in order to improve the measurement precision. Figure 1 of Willbold et al. (6) shows what appears to be estimated external precisions of the limited number of measurements of the order ±12 ppm, and so should have resolved a value as high as measured here for the one Ontong-Java sample analyzed in our study. As the Willbold et al. (6) paper reports only W isotopic composition and not W or HSE concentrations for the samples, we cannot make a direct comparison with the different sample from Ontong-Java that was analyzed in our study.
3. Sample preparation and He concentrations and isotopic compositions
Olivine grains free from adhering matrix were handpicked under a binocular microscope and then leached in 1% oxalic acid for 60 minutes at 90 °C to remove any surficial alteration. The crystals were then ultrasonically cleaned in distilled water and acetone at room temperature. Helium concentrations and isotopic ratios were measured at the noble gas laboratory at UC Davis. The olivine grains were crushed under vacuum with a piston
4
crusher (30). The evolved gases were sequentially exposed to a hot and cold SAES getter. Helium was separated from neon at 32 K and inlet into a Nu Noblesse noble gas mass spectrometer. Helium isotopic ratios and concentrations were determined by normalizing to a standard with a 3He/4He of 8.8 RA (where RA is the 3He/4He ratio normalized to the atmospheric ratio of 1.39×10−6). 4He was measured with a Faraday cup while 3He was measured using a discrete dynode multiplier operating in pulse counting mode. Procedural blanks were run before and after crushing the samples. Blanks averaged 6 × 10-11 cc STP of 4He and varied by ~15% (range 5-7 × 10-11 cc STP of 4He). The reproducibility of standard 3He/4He ratios similar in size to the samples ranged from 2-4%. The uncertainties in the computed ratios are based on the reproducibility of standards and the uncertainties associated with the blank correction. The results for He concentrations and isotopic compositions are shown in Table S3. Please note that the samples analyzed here are different from those reported in Starkey et al. (13).
4. Sample preparation, Re-Os isotope systematics and highly siderophile elementconcentrations
Enriched tracer solutions of 185Re, 190Os, and mixed 191Ir-99Ru-194Pt-105Pd spikes were added to ~1 g of sample powder that was then sealed in borosilicate Carius tubes with 4 ml of concentrated HCl and 5 ml of concentrated HNO3. The Carius tubes were placed in an oven at 270°C for six days. Osmium was extracted from the acid solution into carbon tetrachloride (31), then back-extracted into 4 ml of concentrated HBr, and further purified using a micro-distillation technique (32). Rhenium, Ir, Ru, Pt and Pd were separated and purified using anion exchange column chromatography (33). Osmium isotope analyses of the ultramafic samples were performed at UMD using a Thermo Fisher Triton thermal-ionization mass spectrometer (TIMS) as OsO3
- ions. Ion beams were quantified via peak hopping using a single electron multiplier. Typical ion intensities for 192OsO3
- were ~ 150,000 cps. Instrumental mass bias effects on Os were corrected using an exponential law and a value for 192Os/188Os of 3.083. The internal precision for 187Os/188Os of all analyses was better than ± 0.1% (2 s.e.). The Os blank contribution was < 1 pg, and insignificant for all of the rocks analyzed here. The 187Os/188Os of the UMD J-M Os standard gave a value of 0.11381 ± 13 (n=3) during the period of analysis of these samples. Concentrations of Os were determined with an uncertainty better than 0.2% (2 r.s.d.).
Rhenium, Ir, Ru, Pt and Pd concentrations were determined using the Nu Plasma MC-ICP-MS at UMD, with a triple electron multiplier configuration in static mode. Highly siderophile element concentrations were determined with ± 2% of uncertainty on the Nu Plasma for Ru, Pt, and Pd, and ± 0.5% for Re and Ir. The accuracy of the data on this instrument was periodically assessed by comparing the results for the reference materials UB-N and GP-13 obtained during the ongoing analytical campaigns (e.g., 34) with the results from other laboratories. Concentrations of all HSE and Os isotopic compositions obtained at the UMd were in good agreement with the other laboratories. Diluted spiked aliquots of iron meteorites were run during each analytical session as secondary standards. The results from these runs agreed within 0.5% for Re and Ir, and within 2%
5
for Ru, Pt, and Pd, with fractionation-corrected values obtained from measurements of undiluted iron meteorites using Faraday cups of the same instrument with a signal of >100 mV for the minor isotopes.
Blanks were determined during the course of the different analyses and gave on average (2σ): for Re 1.5 ± 0.5 pg, for Ir 0.9 ± 0.5 pg, for Ru 20 ± 10 pg, for Pt 147 ± 50 pg and for Pd 40 ± 10 pg. These correspond to blank contributions of less than 2% for all elements.
These data are given in Fig. 3 and Table S4.
5. Sample preparation and Nd concentrations and isotopic measurements
Between 250 and 350 mg of powdered whole rock samples were digested using HF-HNO3 following ref. 35. To eliminate insoluble fluorides, which can contain REE, each sample was re-dissolved first in concentrated HNO3 and then in concentrated HCl to which 40-60 mg of boric acid was added. Rare earth elements were separated as a group by cation exchange. Preliminary separation of Nd from the other REE was accomplished by one pass through a column with 2-methyllactic acid. Cerium was oxidized to Ce4+ and was further removed by two passes through a column with Eichrom LN resin using 8M HNO3-20 mM NaBrO3 (36, 37). Samarium was removed using LN resin in 0.175M HCl.
Mass spectrometric procedures at both DTM and Geotop followed those described in ref. 38. Neodymium was measured as Nd+ in a four-step dynamic routine that consists of 270cycles with 8 second integrations per mass step and one 30-second baseline measurement at the start of the run. Each cycle provides static measurements of all Nd isotope ratios and of 140Ce, 141Pr, and 147Sm, and also yields two dynamic measurements each for 142Nd/144Nd, 143Nd/144Nd and 148Nd/144Nd. When possible, samples were analyzed twice on the same filament and those results were combined into a single dataset. Data reduction was accomplished offline. Instrumental mass fractionation of 142Nd/144Nd and 148Nd/144Nd were corrected using 146Nd/144Nd = 0.7219 and the exponential law. These data are given in Figure S3 and Table S5.
Supplementary Text - Models to explain the observed μ182W variability
1. Models for which μ182W and HSE abundances should be coupled
1.1. Late accretion
Abundances of highly siderophile elements (HSE: Re, Os, Ir, Ru, Rh, Pt, Pd, Au) in the mantle provide important clues regarding whether W isotopic anomalies are the result of primordial terrestrial differentiation processes or subsequent late accretionary additions. Given the high metal-silicate partition coefficients of the HSE (e.g. 20), their abundances in the mantle should be several orders of magnitude lower than observed if the mantle
6
were in equilibrium with the core. Accretion of ~ 0.5 wt.% material with bulk chondritic composition after chemical communication between core and mantle ceased can account for the observation that HSE mantle abundances are only ~ 200 times lower than in chondritic meteorites, and that their relative abundances are roughly chondritic (39).
As discussed in Willbold et al. (6), ~ 0.5 wt.% of late accretion would have lowered the μ182W of Earth’s mantle by 10-30, given that chondrites have μ182W values of ~ -200 and W concentrations between 100-200 ppb (5). Chondrites have similar Nd abundances as Earth’s mantle and Nd isotopic distinctions from the mantle are too small to be affected by this amount of late addition of chondritic material. Mantle domains having elevated μ182W, uncorrelated with μ142Nd, could thus reflect their isolation from late accretionary additions. However, rocks derived from such mantle domains should show evidence of originating from sources characterized by low HSE abundances and fractionated Re/Os ratios.
The Baffin Bay and Ontong Java Plateau samples have HSE abundances and initial 187Os/188Os ratios (providing a record of long-term Re/Os ratio) indistinguishable from other mantle-derived lavas with similar MgO abundances that do not show elevated μ182W (Figure 3, Table S4). Furthermore, Baffin Bay sample Pd-2 yielded a μ182W value of +48 that is significantly higher than the range of values that variable late accretion could produce (10-30) if late accretion is responsible for the general modern mantle μ182W = 0. Thus, the data for the flood basalts studied here do not support the idea that they were derived from a mantle reservoir that was isolated from late accretionary additions.
1.2. Metal-silicate segregation
Metal-silicate segregation is the most effective process for fractionating Hf/W ratios, because Hf is a strongly lithophile trace element, while W is moderately siderophile. Core formation and accompanying metal-silicate segregation increases the Hf/W ratio of planetary mantles, leaving them with Hf/W ratios of over 10 times the chondritic value. Core formation while 182Hf was extant is why planetary mantles, (e.g., Earth, Mars, Moon) have 182W/184W ratios > 200 ppm higher than primitive meteorites (e.g., ref. 5).
Sizable impacts during terrestrial planet growth likely led to repeated metal-silicate segregation events (19), creating mantle domains with different Hf/W ratios. The fractionation of the Sm/Nd ratio of these mantle domains through this process would have been negligible because the solubility of both elements in metal is extremely low (40). If these impacts did not entirely melt the Earth’s mantle, or result in extensive mixing that obliterated the evidence of past differentiation events, then each of these events could have produced discrete mantle domains with different Hf/W ratios. The Hf-W systematics of the mantle domain affected by each impact will depend on the time and size of the impact, how the impactor’s core interacts with the surrounding mantle materials, and the oxidation state of the mantle at the time of impact. Mantle domains generated early, under reduced conditions, will have the highest Hf/W ratios and, therefore, the highest 182W/184W ratios ever present in the Earth’s interior (Figure S4).
7
This is because of both the comparatively high abundance of 182Hf early in Solar System history, and the fact that W is more siderophile under reducing conditions (41). Incomparison, mantle domains generated at later times under oxidizing conditions will be characterized by lower Hf/W ratios and less 182Hf, and hence, evolve to lower 182W/184W ratios.
Figure S4 shows possible ranges of μ182W values of mantle domains created after Hf/W fractionation caused by metal-silicate segregation. The figure shows the schematic evolution of these mantle domains formed at different times (0 Ma to 30 Ma after Solar System formation) and at different oxidation states that translate into different partitioning of W between metal and silicate (for metal/silicate distribution coefficients of W ( D W) between 30 and 50, ref. 41). These curves do not take into account any mantle mixing during Earth’s growth, and assume that the impactor’s core sinks through the proto-Earth’s mantle without chemical equilibration. The high 182W/184W mantle domain(s) (μ182W between 10 and 50) sampled by Baffin Bay and Ontong Java lavas most likely formed during an early phase of the giant impact stage of terrestrial growth.
A problem with the model of metal-silicate segregation being responsible for the variable 182W is that HSE abundances would be expected to correlate with μ182W. High Hf/W domains (evolving towards high μ182W values) should show depletions in HSE. As discussed in the main manuscript, however, all the terrestrial samples that have been studied that are characterized by 182W excesses appear to derive from mantle sources that have HSE abundances similar to other modern rocks, with chondritic relative abundances of the HSE, and Os isotopic compositions that are evidence of long-term chondritic Re/Os (6-10). Changing oxidation states during Earth accretion might explain the decoupling of 182W and HSE, because while W becomes less siderophile under oxidizing conditions, the HSE even at high oxidation states, are not soluble in silicates (20). In any case, the addition of a late accreting, HSE-rich, component would have to be mixed into the mantle in a way that did not also mix away any W isotope heterogeneity present when the mixing takes place. One way around this constraint is if the late accreted component is contributed shortly after each major impact. For example, the chondritic relative abundances of HSE and chondritic Os isotopic compositions of the HED meteorites believed to derive from the asteroid Vesta (42) suggest that the late accretion that mayestablish mantle HSE abundances happens very quickly after segregation of metal into the core is complete. How this is done remains unclear, and is at least part of the motivation to consider other ways the mantle could have obtained its HSE abundance pattern.
1.3. Partial core re-mixing
If there is some mechanism by which metal-silicate separation allows the mantle to routinely retain on order 0.1 to 0.2% of metal during core-forming events, the portions of the mantle affected by different core forming events could end up with similar, comparatively high HSE abundances, but the W isotopic composition of each reservoir would depend upon when the metal segregation event occurred and at what depth. This type of model has traditionally been referred to as inefficient core formation (43). There
8
are two problems with this type of model. First, the chondrite normalized mantle abundances of moderately siderophile elements (MSE), including W, are less depleted in the mantle compared to the HSE. This means an additional process to add to, or retain MSE in the mantle is required. The process that established MSE abundances in the mantle has been argued to be metal-silicate segregation in a magma ocean (e.g., 45). Yet this process would lead to a loss of the retained metal to the core. Second, even if some metal can be retained in the mantle, solid metal-liquid metal fractionation is likely to occur, as noted by Jones and Drake (43). Meteorite analogues of inefficient core-formation are characterized by fractionated HSE abundances (44).
A modification of this model was used by Willbold et al. (9) to explain their µ182W values of +6 to +15 for samples of the 3.96 Ga Acasta Gneiss. Their model consists of the following: 1) subsequent to the Moon-forming giant impact, a small fraction of metal from the impactor’s core is retained and re-mixed into Earth’s mantle, driving its HSE contents up to ~80% of current levels and leaving a mantle with a μ182W value of +15; 2) subsequent late accretion of materials with chondritic bulk composition drives the μ182W value of the upper mantle down to 0 and its HSE concentrations up to their current concentrations; 3) samples with elevated 182W abundances are then derived from portions of the lower mantle that do not yet have the final late accretionary materials mixed into them.
The advantage of this model is that it can produce a 182W-enriched mantle reservoir with near normal HSE abundances, because it derives most of the HSE, but only a minor fraction of the W, from retention of the impactor’s core metal in Earth’s mantle. This model differs in two primary ways from the canonical inefficient core formation model of Jones and Drake (43). First, it envisions retention of metal in the mantle, as well as noHSE fractionation resulting from metal crystallization and separation of S-rich liquids from S-poor solids. Second, it requires an additional stage of late accretion to produce the final, desired HSE abundances and W isotopic composition. The chief shortcoming of this model is that it requires the retention of high density metal in a mantle that is partially or wholly melted by the giant impact.
2. Models where high μ182W and HSE need not be correlated
2.1. Early silicate differentiation
As discussed in detail in ref. 7, μ182W variability detected in terrestrial rock samples can be created through crystal-liquid fractionation that occurs, for example, during the crystallization of a magma ocean. Tungsten is more incompatible than Hf, and therefore, Hf/W fractionation is expected to occur by this process. If the fractionation occurs while 182Hf is extant (during the first 50 Ma of Solar System history), then the early-differentiated mantle reservoirs will evolve to variable μ182W values.
The μ182W that results from silicate differentiation depends on the degree of Hf/W fractionation and time of the differentiation event. For example, μ182W values of ~+50
9
can be created in a mantle reservoir that underwent 10% melt extraction 30 Ma after Solar System formation (Fig. S5A). Besides affecting Hf/W ratios, however, melt removal will also affect lithophile element ratios, such as Sm/Nd, because Nd is more incompatible than Sm. This would leave a melt residue (or initial magma ocean precipitate) with high Sm/Nd ratio. This fractionation would, therefore, be traceable in both the short-lived 146Sm→142Nd (t½ = 103 Ma) and the long-lived 147Sm→143Nd (t½ =106 Ga) isotopic systems. As shown in Figure S5A, removal of 10% melt at 30 Ma, would result in a present day mantle having 147Sm/144Nd of 0.216, μ142Nd = +7 and ε143Nd +10. The μ142Nd values measured for Baffin Bay and Ontong Java Plateau rocks average +4 (Table S5), and are not resolvable from either the terrestrial standard or the value predicted for an early formed residue of 10% partial melt extraction. The ε143Nd value for the same samples average of +6 (Table S5), which is in the range of values obtained for other Baffin Bay (e.g. 13) and Ontong Java Plateau rocks (46). This isotopic composition is marginally lower than predicted for the early-formed residue of melt extraction. Thus, while it would be expected that μ182W and μ142Nd correlate in the products of magma ocean differentiation, the resolution provided by the 146Sm-142Nd system may be insufficient to identify the expected correlation.
A mantle reservoir formed even later, e.g., 60 Ma, would require extreme Hf/W fractionation to produce μ182W values of +50 in the melt-depleted reservoirs (Fig. S5B). This magnitude of differentiation would produce reservoirs with modern ε143Nd and μ142Nd values of +30 and +58, respectively, that are well outside the range observed in the modern mantle. This result suggests that if silicate differentiation is responsible for the W isotope variation, it must have occurred early (< 30 Ma after Solar System formation) and involved relatively small degrees of fractionation.
2.2. Binary mixing models
Touboul et al. (7) and Willbold et al. (9) proposed binary mixing models between early-differentiated deep silicate mantle domains, characterized by depletions in HSE and high μ182W values, and a primitive upper mantle-like composition (PUM) that has μ182W=0 and modern-mantle HSE abundances.
Figure S6A shows such a model that assumes chondrite normalized Os, Pd, Re abundances in the early-differentiated HSE depleted mantle of 2x10-6, 0.004 and 0.0006, respectively (based on high pressure and temperature metal-silicate partition coefficients for these elements, 47, 48) a W concentration of 2 ppb (49) and a μ182W value of +130 (calculated for a silicate reservoir differentiated 30 Ma after Solar System formation, with a DW=50). Material from this early-differentiated mantle mixes with PUM-like mantle, having chondrite normalized HSE abundances of 0.008, a W concentration of ~ 13 ppb (49, 50) and μ182W=0. This model can explain a range in μ182W values, but it would take very large amounts (>80%, Figure S6A) of PUM-like mantle in the mixture before the HSE patterns come close to matching those of the modern mantle. Additionally, the resulting μ182W value (~ +5) would be much lower than the value measured in the flood basalts.
10
An alternative to consider is that the high μ182W values measured in the flood basalts come from recycling small fractions of ancient crust into the source of these lavas. Although one would expect crust to have low Hf/W, and hence unradiogenic W, some ancient crustal rocks have been shown to have positive μ182W as high as +20 (6-10). These high μ182W values imply the genesis of this crust from mantle sources that had high Hf/W while 182Hf was still extant. If crust formation occurred after the extinction of 182Hf, as it clearly did in Greenland, Acasta and Nuvvuagittuq, the measured μ182W in these ancient crustal rocks must have been inherited from their sources. Figure S6B shows the consequences of recycling such ancient crustal material into PUM-like mantle assumed to have the HSE and W concentrations of 0.008 times chondritic HSE and W concentrations ~ 13 ppb W (49, 50), and a μ182W= 0 (6-10). We have used crustal amphibolites from Greenland (10) as the hypothetical recycled crust, and adjusted their μ182W to +50. The assumed recycled crust has Os, Pd, and Re concentrations of 0.0001, 0.006 and 0.1 times chondritic with W concentration of 300 ppb (10). Recycling of this material will have relatively little effect on the HSE pattern of the mantle, but will change its W concentration and μ182W values significantly because of the high W concentration in the recycled crust. In the example chosen, the recycled crust also results in a significant increase in the Re/Os ratio of the mantle into which the crust is recycled. Even with only 4% of such crust present in the mantle source domain, the Re/Os ratio will be increased by 50% over that of uncontaminated mantle, which would result in elevated 187Os/188Os that is not observed in the modern flood basalts (this study, and 18). The recycled crustal component would have to be extremely radiogenic (μ182W > 250) so that less than 1% of this recycled material would be required in order to maintain low W concentrations and Re/Os ratios within ±10% of PUM, in the portion of the mantle affected by this crustal recycling. Similarly, crustal recycling of this nature also would have consequences for the Sr, Nd, Hf and Pb isotopic composition of the source, leading to the expectation of a negative correlation between μ182W and ε143Nd, for example, but the magnitude of the correlation depends on the Sm and Nd concentrations of the crustal component. The composition of crust can be changed dramatically during the subduction process, but because W is one of the fluid-mobile elements during plate subduction, subduction processing of the crust will result in a recycled component that likely has very low W concentrations, which would make this by of crustal recycling model unlikely to explain the data.
11
Fig. S1 182W/184W vs. 183W/184W of the Alfa Aesar W standard solution and the samples measured in this study. The correlation of these two ratios is attributed to mass dependent oxygen isotope fractionation during the analysis of the WO3
- signals.
0.86488
0.86492
0.86496
0.86500
0.86504
0.46718 0.46720 0.46722 0.46724 0.46726
W StandardsBaffin Bay VE-32 Ontong Java Plateau
182 W
/184 W
183W/184W
Figure S1
12
Fig. S2
Tungsten concentration of a magma after different percentages of partial melting (F) of a mantle source with ~ 5 ppb W. These magmas are estimated to derive from 15-20% partial melting (51) and will thus contain ~ 25-27 ppb W. The partition coefficient for W (DW) between solid and melt used in this model is 0.02, for an upper mantle phase assemblage composed of 60% Ol + 20% Opx + 20% Cpx (53).
1
10
100
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
W (p
pb)
F
Figure S2
magma
13
Fig. S3
μ142Nd values measured for Baffin Bay and Ontong Java Plateau samples. The values are expressed as parts in 1,000,000, or ppm deviations from the average measured from the JNdi standard (n=7). Errors for each data point are 2σ-mean. The light grey shaded area represents the 2σ-mean for the JNdi standard measurements obtained at the Department of Terrestrial Magnetism, and the dark shaded area represents the 2σ for the JNdi standard measurements obtained at Geotop.
-5.0 0.0 5.0 10.0 15.0 -10.0 ȝ142Nd (ppm)
Baffin BayPd-2
Baffin BayPi-23
OJP192-1187A-009R-04R
14
Fig. S4 Top: Schematic evolution of mantle domains differentiated from a chondritic composition at different times and at different oxidation states that translate into different partitioning of W between metal and silicate (DW between 30 and 50, ref. 41). Bottom: The evolution for the same mantle domains, showing their respective μ182W values and W concentrations in function of the different DW between metal and silicate.
-350
-150
50
250
450
650
850
0 10 20 30 40 50 60 70 80
Chondritic evolution
Time after Solar System formation
ȝ182
W (p
pm)
DW = 50DW = 45DW = 40DW = 35DW = 30
80
0
500
1000
1500
2000
2500
3000
3500
30 32 34 36 38 40 42 44 46 48 50 0
2
4
6
8
10
12
14
16
18
20
0 Ma
10 Ma
20 Ma
30 Ma
W concentration
DW metal-silicate
W concentration of silicate
ȝ182
W (p
pm)
15
Fig. S5 Evolution of μ142Nd (grey dotted curves) and μ182W (black solid curves) for depleted reservoirs through time after melt extraction. (A) 10% melt extraction that occurs 30 Ma after Solar System formation. (B) 70% melt extraction that occurs after 60 Ma of Solar System formation. Partition coefficients for Sm, Nd, Hf and W are those for a melt residue composed of 85% MgPv, 5% CaPv and 5% Mw (52). Models involving shallowmantle phases yield even stronger fractionations (using partition coefficients from ref. 53)and therefore are not shown here. The μ142Nd and μ182W values are calculated using half-lives of 103 Ma and 8.9 Ma for 146Sm and 182Hf, respectively (54, 55). The initial 182Hf/180Hf and 146Sm/144Sm ratios used are those of ref. 5 and ref. 54, respectively. μ142Nd values are calculated comparing the evolution of the depleted reservoir to that of a non-chondritic silicate Earth (56).
0
10
20
30
40
50
60
0 100 200 300 400 500 Time (Ma)
ȝ��SSP
�
0
10
20
30
40
50
60
0 100 200 300 400 500 Time (Ma)
ȝ��SSP
�
A Bȝ182:��SSP��
ȝ1421G��SSP��
ȝ182:��SSP��
ȝ1421G��SSP��
16
Fig. S6 Mixing models between an early differentiated deep silicate mantle domain, characterized by depletions in HSE based on high P-T partition coefficients from ref. 47 and 48 and high μ182W values, and a PUM-like mantle that has μ182W=0 and modern-mantle HSE abundances (Panel A). Panel B shows the HSE patterns created by recycling continental crust similar in composition to an Isua amphibolite (10) with μ 182W artificially increased to +50 into a PUM-like mantle with μ 182W = 0. The legends show the percentage of mixed PUM-like material or ancient crust, with its μ 182W and W concentration shown to the side of each HSE pattern.
0.005Os Pd ReA
bund
ance
/ C
hond
rites
Element
0.010
0.015
0.020 % of ancient crust
ȝ182W [W]36 42
25 2516 19
1-10% 10 16
B
0.010
Os Pd Re
Abu
ndan
ce /
Cho
ndrit
es
Element
% of PUM-likemantle ȝ182W [W]
50
24
50
4
6
1113100
80
40
200.002
0.003
0.0040.005
A
17
Table S1. Major and minor element concentrations for the Baffin Bay and Ontong Java plateau samples
Ontong Java 192-1187A-9R-4 Baffin Bay Pd-2 Baffin Bay Pi-23SiO2 (wt%) 48.87 46.06 45.66TiO2 0.81 0.94 0.80Al2O3 14.84 12.90 11.97Fe2O3T 11.37 11.92 11.42MnO 0.17 0.19 0.18MgO 9.37 15.20 18.45CaO 12.40 10.95 9.75Na2O 1.81 1.49 1.27K2O 0.063 0.021 0.047P2O5 0.09 0.09 0.08Total 99.79 99.76 99.63LOI 1.68 0.84 0.60
Sr (ppm) 127 146 149Zr 68 74 72V 260 298 258Cr 496 965 1462
18
Tabl
e S2
Tung
sten
isot
ope
resu
lts fo
r Baf
fin B
ay a
nd O
nton
g Ja
va P
late
au s
ampl
es, t
he g
eolo
gica
l ref
eren
ce m
ater
ial V
E-4
3 an
d th
e A
lfa A
esar
W s
tand
ard*
Sam
ple
Mag
azin
e18
5R
e/18
4W
188O
s/18
4W
180W
/184W
± 2σ
182W
/184W
± 2σ
183W
/184W
± 2σ
182W
/184W
**±
2σµ
182 W
(ppm
)±
2σ (p
pm)
Pi-2
3aB
587
0.30
-0.0
0013
0.00
3873
0.00
0001
0.86
4981
0.00
0005
0.46
7226
0.00
0003
0.86
4776
00.
0000
051
11.9
5.9
Pi-2
3b
B59
50.
06-0
.000
100.
0038
740.
0000
010.
8649
500.
0000
050.
4672
110.
0000
020.
8647
762
0.00
0004
88.
35.
6P
i-23
aver
age
0.86
4776
10.
0000
004
10.1
5.0
Pd-
2B
622
0.06
-0.0
0009
0.00
3873
0.00
0001
0.86
5049
0.00
0004
0.46
7241
0.00
0002
0.86
4806
50.
0000
040
48.4
4.6
OJP
B62
20.
16-0
.000
150.
0038
740.
0000
010.
8650
180.
0000
050.
4672
370.
0000
020.
8647
854
0.00
0004
623
.95.
3
VE
-32
B59
50.
12-0
.000
030.
0038
720.
0000
010.
8649
020.
0000
00.
4671
920.
0000
00.
8647
680.
0000
04-0
.84.
5
W s
tdB
587
0.17
-0.0
0018
0.00
3876
0.00
0001
0.86
5003
0.00
0006
0.46
7242
0.00
0003
0.86
4765
0.00
0006
-0.8
6.4
W s
tdB
587
0.07
-0.0
0023
0.00
3873
0.00
0001
0.86
5028
0.00
0004
0.46
7254
0.00
0002
0.86
4766
0.00
0004
0.1
5.1
W s
tdB
587
0.07
-0.0
0016
0.00
3873
0.00
0001
0.86
4974
0.00
0006
0.46
7229
0.00
0003
0.86
4764
0.00
0006
-1.8
6.6
W s
tdB
587
0.10
-0.0
0015
0.00
3874
0.00
0001
0.86
4984
0.00
0004
0.46
7232
0.00
0002
0.86
4768
0.00
0004
2.5
5.1
W s
td a
vera
ge (n
=4)
0.86
4766
0.00
0003
0.0
3.7
W s
tdB
595
0.02
-0.0
0006
0.00
3873
0.00
0001
0.86
4914
0.00
0004
0.46
7198
0.00
0002
0.86
4767
0.00
0004
-1.9
4.5
W s
tdB
595
0.03
-0.0
0007
0.00
3872
0.00
0001
0.86
4921
0.00
0004
0.46
7201
0.00
0002
0.86
4768
0.00
0004
-1.0
4.6
W s
tdB
595
0.05
-0.0
0003
0.00
3872
0.00
0001
0.86
4890
0.00
0004
0.46
7186
0.00
0002
0.86
4770
0.00
0004
0.8
5.2
W s
tdB
595
0.06
-0.0
0020
0.00
3873
0.00
0001
0.86
5017
0.00
0004
0.46
7247
0.00
0002
0.86
4768
0.00
0004
-1.4
5.0
W s
tdB
595
0.03
-0.0
0015
0.00
3876
0.00
0001
0.86
4968
0.00
0005
0.46
7222
0.00
0002
0.86
4772
0.00
0005
3.4
6.0
W s
td a
vera
ge (n
=5)
0.86
4769
0.00
0004
0.0
4.3
W s
tdB
622
0.14
-0.0
0013
0.00
3874
694
0.00
0001
0.86
4978
0.00
0004
0.46
7227
0.00
0002
0.86
4764
0.00
0004
-1.6
4.7
W s
tdB
622
0.09
-0.0
0014
0.00
3873
187
0.00
0001
0.86
4977
0.00
0004
0.46
7226
0.00
0002
0.86
4766
0.00
0004
0.7
4.8
W s
tdB
622
0.15
-0.0
0014
0.00
3875
919
0.00
0002
0.86
4976
0.00
0007
0.46
7224
0.00
0003
0.86
4768
0.00
0007
3.4
7.6
W s
tdB
622
0.04
-0.0
0009
0.00
3873
430.
0000
010.
8649
340.
0000
040.
4672
070.
0000
020.
8647
640.
0000
04-1
.64.
6W
std
B62
20.
09-0
.000
200.
0038
7300
20.
0000
010.
8650
100.
0000
040.
4672
420.
0000
020.
8647
630.
0000
04-2
.75.
1W
std
B62
20.
07-0
.000
170.
0038
7355
60.
0000
010.
8650
040.
0000
040.
4672
380.
0000
020.
8647
670.
0000
041.
95.
2W
std
B62
20.
07-0
.000
150.
0038
7327
80.
0000
010.
8649
780.
0000
040.
4672
280.
0000
020.
8647
640.
0000
04-1
.14.
9W
std
B62
20.
05-0
.000
120.
0038
7243
0.00
0001
0.86
4958
0.00
0005
0.46
7218
0.00
0002
0.86
4765
0.00
0005
0.6
5.6
W s
tdB
622
0.14
-0.0
0019
0.00
3872
907
0.00
0001
0.86
5008
0.00
0004
0.46
7241
0.00
0002
0.86
4766
0.00
0004
1.2
5.0
W s
tdB
622
0.15
-0.0
0019
0.00
3873
289
0.00
0001
0.86
5008
0.00
0004
0.46
7242
0.00
0002
0.86
4764
0.00
0004
-0.7
4.7
W s
td a
vera
ge (n
=10)
0.86
4765
0.00
0003
0.0
3.8
*The
W s
tand
ard
used
in th
is s
tudy
is th
e A
lfa A
esar
tung
sten
pla
sma
stan
dard
(LO
T 32
-438
793F
), ce
rtiife
d at
1.0
19 ±
0.0
02g/
ml,
in 5
% H
NO
3/tr
HF.
**C
orre
cted
for s
econ
d-or
der o
xyge
n is
otop
e fra
ctio
natio
n ef
fect
. See
Mat
eria
ls a
nd M
etho
ds s
ectio
n 2
for m
ore
deta
ils
19
Table S3. Helium concentration and isotopic ratios in Baffin Bay picrites
Sample Weight,(g)4He
1039,cc,STP,g31 3He/4He,RA ±1σ
Baffin&Bay&Pi)23 0.27 1.3 43.2 2.2Baffin&Bay&Pi)24 0.47 0.4 24.1 1.5Baffin&Bay&Pi)26 0.58 1.7 48.4 1.5
20
Sam
ple
Os
IrR
uPt
PdR
e18
7 Os/
188 O
s18
7 Re/
188 O
sA
ge (M
a)18
7 Os/
188 O
(t)γO
sP
i-23
Baf
fin B
ay1.
124
0.68
71.
076.
142.
840.
226
0.12
757
0.96
960
0.12
660.
0P
d-2
Baf
fin B
ay0.
439
0.33
62.
025.
347.
540.
416
0.13
161
4.56
600.
1270
0.3
Ont
ong
Java
192
-118
7A-9
R-4
0.18
70.
148
0.65
6.82
12.0
1.11
20.
1790
228
.812
00.
1214
-3.8
Tabl
e S4
. Hig
hly
side
roph
ile e
lem
ent a
bund
ance
s an
d R
e-O
s is
otop
e sy
stem
atic
s fo
r the
Baf
fin B
ay a
nd O
nton
g Ja
va P
late
au s
ampl
es
21
Tabl
e S5
. µ14
2 Nd
valu
es a
nd 14
7 Sm-14
3 Nd
for B
affin
Bay
and
Ont
ong
Java
Pla
teau
sam
ples
T (M
a)14
2 Nd/
144 N
dµ14
2 Nd
(ppm
)2S
E (p
pm)
Sm (p
pm)
Nd
(ppm
)14
3 Nd/
144 N
d2S
E (p
pm)
147 Sm
/144 N
dε14
3 Nd(
T)B
affin
Bay
PD
-2*
601.
1418
463.
82.
50.
5130
040.
0000
01B
affin
Bay
PD
-2 re
plic
ate
1*60
1.14
1846
3.9
2.1
0.51
2966
0.00
0001
Baf
fin B
ay P
D-2
repl
icat
e 2*
*60
1.14
1840
3.9
1.3
2.11
16.
657
0.51
2998
0.00
0001
0.19
177.
2B
affin
Bay
PD
-2 re
plic
ate
3**
601.
1418
412.
03.
62.
117
6.66
70.
5129
980.
0000
010.
1919
7.2
Baf
fin B
ay P
i-23*
601.
1418
529.
44.
71.
781
5.84
90.
5129
250.
0000
020.
1841
5.8
Baf
fin B
ay P
i-23
repl
icat
e 1*
*60
1.14
1842
0.9
3.1
1.81
95.
848
0.51
2926
0.00
0001
0.18
815.
8B
affin
Bay
Pi-2
3 re
plic
ate
2**
601.
1418
423.
85.
71.
806
5.81
10.
5129
210.
0000
030.
1879
5.7
Baf
fin B
ay, A
vera
ge1.
1418
444.
05.
41.
96.
20.
5129
630.
0000
020.
1887
6.4
Ont
ong
Java
Pla
teau
*12
01.
1418
463.
93.
01.
481
4.39
30.
5129
660.
0000
010.
2038
6.4
JNdi
ave
rage
(n=7
)*1.
1418
410.
08.
6(2
SD
, ppm
)JN
di a
vera
ge (n
=4)*
*1.
1418
370.
05.
4(2
SD
, ppm
)
Rep
licat
es a
re s
epar
ate
diss
olut
ions
*M
easu
red
at th
e D
epar
tmen
t of T
erre
stria
l Mag
netis
m o
f Car
negi
e In
stitu
tion
for S
cien
ce, W
ashi
ngto
n D
.C.
** M
easu
red
at th
e G
eoto
p re
sear
ch c
ente
r, U
nive
rsité
du
Qué
bec à
Mon
tréal
, Mon
tréal
.
References and Notes 1. S. Mukhopadhyay, Early differentiation and volatile accretion recorded in deep-mantle neon
and xenon. Nature 486, 101–104 (2012). Medline doi:10.1038/nature11141
2. M. Boyet, M. O. Garcia, R. Pik, F. Albarede, A search for 142Nd evidence of primordial mantle heterogeneities in plume basalts. Geophys. Res. Lett. 32, L04306 (2005). doi:10.1029/2004GL021873
3. V. C. Bennett, A. D. Brandon, A. P. Nutman, Coupled 142Nd-143Nd isotopic evidence for Hadean mantle dynamics. Science 318, 1907–1910 (2007). Medline doi:10.1126/science.1145928
4. H. Rizo, M. Boyet, J. Blichert-Toft, M. T. Rosing, Early mantle dynamics inferred from 142Nd variations in Archean rocks from southwest Greenland. Earth Planet. Sci. Lett. 377–378, 324–335 (2013). doi:10.1016/j.epsl.2013.07.012
5. T. Kleine, M. Touboul, B. Bourdon, F. Nimmo, K. Mezger, H. Palme, S. B. Jacobsen, Q.-Z. Yin, A. N. Halliday, Hf–W chronology of the accretion and early evolution of asteroids and terrestrial planets. Geochim. Cosmochim. Acta 73, 5150–5188 (2009). doi:10.1016/j.gca.2008.11.047
6. M. Willbold, T. Elliott, S. Moorbath, The tungsten isotopic composition of the Earth’s mantle before the terminal bombardment. Nature 477, 195–198 (2011). Medline doi:10.1038/nature10399
7. M. Touboul, I. S. Puchtel, R. J. Walker, 182W evidence for long-term preservation of early mantle differentiation products. Science 335, 1065–1069 (2012). Medline doi:10.1126/science.1216351
8. M. Touboul, J. Liu, J. O'Neil, I. S. Puchtel, R. J. Walker, New insights into the Hadean mantle revealed by 182W and highly siderophile element abundances of supracrustal rocks from the Nuvvuagittuq Greenstone Belt, Quebec, Canada. Chem. Geol. 383, 63–75 (2014). doi:10.1016/j.chemgeo.2014.05.030
9. M. Willbold, S. J. Mojzsis, H.-W. Chen, T. Elliot, Tungsten isotope composition of the Acasta Gneiss Complex. Earth Planet. Sci. Lett. 419, 168–177 (2015). doi:10.1016/j.epsl.2015.02.040
10. H. Rizo, R. J. Walker, R. W. Carlson, M. Touboul, M. F. Horan, I. S. Puchtel, M. Boyet, M. T. Rosing, Early Earth differentiation investigated through 142Nd, 182W, and highly siderophile element abundances in samples from Isua, Greenland. Geochim. Cosmochim. Acta 175, 319–336 (2016). doi:10.1016/j.gca.2015.12.007
11. A. D. Saunders, J. G. Fitton, A. C. Kerr, M. J. Norry, R. W. Kent, The North Atlantic igneous province. Geophys. Monogr. Ser. 100, 45–94 (1997).
12. C. R. Neal, J. J. Mahoney, L. W. Kroenke, R. A. Duncan, M. G. Petterson, The Ontong Java Plateau. Geophys. Monogr. Ser. 100, 183–216 (1997).
13. N. A. Starkey, F. M. Stuart, R. M. Ellam, J. G. Fitton, S. Basu, L. M. Larsen, Helium isotopes in early Iceland plume picrites: Constraints on the composition of high 3He/4He mantle. Earth Planet. Sci. Lett. 277, 91–100 (2009). doi:10.1016/j.epsl.2008.10.007
14. M. G. Jackson, R. W. Carlson, M. D. Kurz, P. D. Kempton, D. Francis, J. Blusztajn, Evidencefor the survival of the oldest terrestrial mantle reservoir. Nature 466, 853–856 (2010). Medline doi:10.1038/nature09287
15. L. J. Hallis, G. R. Huss, K. Nagashima, G. J. Taylor, S. A. Halldórsson, D. R. Hilton, M. J.Mottl, K. J. Meech, Evidence for primordial water in Earth’s deep mantle. Science 350, 795–797 (2015). Medline doi:10.1126/science.aac4834
16. M. G. Jackson, R. W. Carlson, An ancient recipe for flood-basalt genesis. Nature 476, 316–319 (2011). Medline doi:10.1038/nature10326
17. Materials and methods are available as supplementary materials on Science Online.
18. C. W. Dale, D. G. Pearson, N. A. Starkey, F. M. Stuart, R. M. Ellam, L. M. Larsen, J. G.Fitton, C. G. Macpherson, Osmium isotopes in Baffin Island and West Greenland picrites: Implications for the 187Os/188Os composition of the convecting mantle and the nature of high 3He/4He mantle. Earth Planet. Sci. Lett. 278, 267–277 (2009). doi:10.1016/j.epsl.2008.12.014
19. D. Rubie, F. Nimmo, H. J. Melosh, “Formation of Earth’s core,” in Treatise on Geophysics,G. Schubert, Ed. (Elsevier, ed. 2, 2015), pp. 51–90.
20. H. St. C. O’Neill, D. B. Dingwell, A. Borisov, B. Spettel, H. Palme, Experimentalpetrochemistry of some highly siderophile elements at high temperatures, and some implications for core formation and the mantle's early history. Chem. Geol. 120, 255–273 (1995). doi:10.1016/0009-2541(94)00141-T
21. M. K. Pető, S. Mukhopadhyay, K. A. Kelley, Heterogeneities from the first 100 million yearsrecorded in deep mantle noble gases from the Northern Lau Back-arc Basin. Earth Planet. Sci. Lett. 369–370, 13–23 (2013). doi:10.1016/j.epsl.2013.02.012
22. M. G. Jackson, S. B. Shirey, E. H. Hauri, M. D. Kurz, H. Rizo, Peridotite xenoliths from thePolynesian Austral and Samoa hotspots: Implications for the destruction of ancient 187Os and 142Nd isotopic domains and the preservation of Hadean 129Xe in the modern convecting mantle. Geochim. Cosmochim. Acta 10.1016/j.gca.2016.02.011 (2016). doi:10.1016/j.gca.2016.02.011
23. J. P. Brandenburg, E. H. Hauri, P. E. van Keken, C. J. Ballentine, A multiple-system study ofthe geochemical evolution of the mantle with force-balanced plates and thermochemical effects. Earth Planet. Sci. Lett. 276, 1–13 (2008). doi:10.1016/j.epsl.2008.08.027
24. E. J. Garnero, A. K. McNamara, Structure and dynamics of Earth’s lower mantle. Science320, 626–628 (2008). Medline doi:10.1126/science.1148028
25. M. Horan, R. J.. Walker, J. W. Morgan, J. N. Grossman, A. E. Rubin, Highly siderophileelements in chondrites. Chem. Geol. 196, 27–42 (2003). doi:10.1016/S0009-2541(02)00405-9
26. T. J. Ireland, R. J. Walker, M. O. Garcia, Highly siderophile element and 187Os isotopesystematics of Hawaiian picrites: Implications for parental melt composition and source heterogeneity. Chem. Geol. 260, 112–128 (2009). doi:10.1016/j.chemgeo.2008.12.009
27. W. B. Tonks, H. J. Melosh, Magma ocean formation due to giant impacts. J. Geophys. Res.Planets 98, 5319–5333 (1993). doi:10.1029/92JE02726
28. F. R. Boyd, S.A. Mertzman, in Magmatic Processes: Physicochemical Principles, B. O. Mysen, Ed. (Special Publication, Geochemical Society, 1987), pp. 13–24.
29. M. Touboul, R. J. Walker, High precision tungsten isotope measurement by thermal ionization mass spectrometry. Int. J. Mass Spectrom. 309, 109–117 (2012). doi:10.1016/j.ijms.2011.08.033
30. E. Gayer, S. Mukhopadhyay, B. J. Meade, Spatial variability of erosion rates inferred from the frequency distribution of cosmogenic 3He in olivines from Hawaiian river sediments. Earth Planet. Sci. Lett. 266, 303–315 (2008). doi:10.1016/j.epsl.2007.11.019
31. A. S. Cohen, F. G. Waters, Separation of osmium from geological materials by solvent extraction for analysis by thermal ionisation mass spectrometry. Anal. Chim. Acta 332, 269–275 (1996). doi:10.1016/0003-2670(96)00226-7
32. J. L. Birck, M. R. Barman, F. Capman, Re‐Os isotopic measurements at the femtomole level in natural samples. Geostand. Newsl. 21, 19–27 (1997). doi:10.1111/j.1751-908X.1997.tb00528.x
33. M. Rehkämper, A. N. Halliday, Accuracy and long-term reproducibility of lead isotopic measurements by multiple-collector inductively coupled plasma mass spectrometry using an external method for correction of mass discrimination. Int. J. Mass Spectrom. 181, 123–133 (1998). doi:10.1016/S1387-3806(98)14170-2
34. I. S. Puchtel, R. J. Walker, M. Touboul, E. G. Nisbet, G. R. Byerly, Insights into early Earth from the Pt–Re–Os isotope and highly siderophile element abundance systematics of Barberton komatiites. Geochim. Cosmochim. Acta 125, 394–413 (2014). doi:10.1016/j.gca.2013.10.013
35. M. Boyet, R. W. Carlson, 142Nd evidence for early (>4.53 Ga) global differentiation of the silicate Earth. Science 309, 576–581 (2005). Medline doi:10.1126/science.1113634
36. H. Tazoe, H. Obata, T. Gamo, Determination of cerium isotope ratios in geochemical samples using oxidative extraction technique with chelating resin. J. Anal. At. Spectrom. 22, 616–622 (2007). doi:10.1039/b617285g
37. Y Y. Hirahara, Q. Chang, T. Miyazaki, T. Takahashi, J.-I. Kimura, Improved Nd chemical separation technique for 143Nd/144Nd analysis in geological samples using packed Ln resin columns. JAMSTEC Rep. Res. Devel. 15, 27–33 (2012). doi:10.5918/jamstecr.15.27
38. R. W. Carlson, M. Boyet, M. Horan, Chondrite barium, neodymium, and samarium isotopic heterogeneity and early Earth differentiation. Science 316, 1175–1178 (2007). Medline doi:10.1126/science.1140189
39. R. J. Walker, Highly siderophile elements in the Earth, Moon and Mars: Update and implications for planetary accretion and differentiation. Chem. Erde Geochem. 69, 101–125 (2009). doi:10.1016/j.chemer.2008.10.001
40. M. A. Bouhifd, M. Boyet, C. Cartier, T. Hammouda, N. Bolfan-Casanova, J. L. Devidal, D. Andrault, Superchondritic Sm/Nd ratio of the Earth: Impact of Earth's core formation. Earth Planet. Sci. Lett. 413, 158–166 (2015). doi:10.1016/j.epsl.2014.12.054
41. J. Wade, B. J. Wood, J. Tuff, Metal–silicate partitioning of Mo and W at high pressures and temperatures: Evidence for late accretion of sulphur to the Earth. Geochim. Cosmochim. Acta 85, 58–74 (2012). doi:10.1016/j.gca.2012.01.010
42. J. M. D. Day, R. J. Walker, L. Qin, D. Rumble III, Late accretion as a natural consequence of planetary growth. Nat. Geosci. 5, 614–617 (2012). doi:10.1038/ngeo1527
43. J. H. Jones, M. J. Drake, Geochemical constraints on core formation in the Earth. Nature 322, 221–228 (1986). doi:10.1038/322221a0
44. J. M. D. Day, R. J. Walker, R. D. Ash, Y. Liu, D. Rumble III, A. J. Irving, C. A. Goodrich, K. Tait, W. F. McDonough, L. A. Taylor, Origin of felsic achondrites Graves Nunataks 06128 and 06129, and ultramafic brachinites and brachinite-like achondrites by partial melting of volatile-rich primitive parent bodies. Geochim. Cosmochim. Acta 81, 94–128 (2012). doi:10.1016/j.gca.2011.12.017
45. J. Li, C. B. Agee, Geochemistry of mantle–core differentiation at high pressure. Nature 381, 686–689 (1996). doi:10.1038/381686a0
46. M. L. G. Tejada, J. J. Mahoney, P. R. Castillo, S. P. Ingle, H. C. Sheth, D. Weis, Pin-pricking the elephant: Evidence on the origin of the Ontong Java Plateau from Pb-Sr-Hf-Nd isotopic characteristics of ODP Leg 192 basalts. Geol. Soc. Lond. Spec. Publ. 229, 133–150 (2004). doi:10.1144/GSL.SP.2004.229.01.09
47. J. M. Brenan, W. F. McDonough, Core formation and metal-silicate fractionation of osmium and iridium from gold. Nat. Geosci. 2, 798–801 (2009). doi:10.1038/ngeo658
48. K. T. Wheeler, D. Walker, W. F. McDonough, Pd and Ag metal-silicate partitioning applied to Earth differentiation and core-mantle exchange. Meteorit. Planet. Sci. 46, 199–217 (2011). doi:10.1111/j.1945-5100.2010.01145.x
49. R. Arevalo Jr., W. F. McDonough, Tungsten geochemistry and implications for understanding the Earth's interior. Earth Planet. Sci. Lett. 272, 656–665 (2008). doi:10.1016/j.epsl.2008.05.031
50. S. König, C. Münker, S. Hohl, H. Paulick, A. R. Barth, M. Lagos, J. Pfänder, A. Büchl, The Earth’s tungsten budget during mantle melting and crust formation. Geochim. Cosmochim. Acta 75, 2119–2136 (2011). doi:10.1016/j.gca.2011.01.031
51. D. Francis, The Baffin Bay lavas and the value of picrites as analogues of primary magmas. Contrib. Mineral. Petrol. 89, 144–154 (1985). doi:10.1007/BF00379449
52. A. Corgne, C. Liebske, B. J. Wood, D. C. Rubie, D. J. Frost, Silicate perovskite-melt partitioning of trace elements and geochemical signature of a deep perovskitic reservoir. Geochim. Cosmochim. Acta 69, 485–496 (2005). doi:10.1016/j.gca.2004.06.041
53. K. Righter, C. K. Shearer, Magmatic fractionation of Hf and W: Constraints on the timing of core formation and differentiation in the Moon and Mars. Geochim. Cosmochim. Acta 67, 2497–2507 (2003). doi:10.1016/S0016-7037(02)01349-2
54. M. Boyet, R. W. Carlson, M. Horan, Old Sm–Nd ages for cumulate eucrites and redetermination of the solar system initial 146Sm/144Sm ratio. Earth Planet. Sci. Lett. 291, 172–181 (2010). doi:10.1016/j.epsl.2010.01.010
55. C. Vockenhuber, M. Bichler, R. Golser, W. Kutschera, A. Priller, P. Steier, S. Winkler, 182Hf, a new isotope for AMS. Nucl. Instrum. Methods Phys. Res. B 223–224, 823–828 (2004). doi:10.1016/j.nimb.2004.04.152
56. G. Caro, B. Bourdon, Non-chondritic Sm/Nd ratio in the terrestrial planets: Consequences for the geochemical evolution of the mantle–crust system. Geochim. Cosmochim. Acta 74, 3333–3349 (2010). doi:10.1016/j.gca.2010.02.025