Roadmap | The review process | Home
   Osmium & Tungsten
Re-Os, Pt-Os and Hf-W isotopes and tracing the core in mantle melts

Anders Scherstén

Danmarks & Grønlands Geologiske Undersøgelse (GEUS) , Øster Voldgade 10 , DK-1350 København K , Denmark,


For a brief and simple introduction, see "Journey from the Centre of the Earth?" by Anders Scherstén, Tim Elliott & Chris Hawkesworth; a NERC article.


Positive 187Os/188Os – 186Os/188Os correlations in Hawaiian picrites and Gorgonan komatiites have been interpreted as representing a small contribution from the Earth’s core to their mantle sources [1-3]. If correct, this chemical core probe could help to constrain core crystalisation history [2]. Furthermore, tracing the core in oceanic lavas such as Hawaiian picrites (“plume lavas” hereafter) may provide a geochemical argument for the existence of mantle plumes if they are predicted to originate at the core-mantle boundary. Hf-W isotope systematics provide a test of the core contribution model and core-mantle admixtures are predicted to show weak but resolvable W-isotope anomalies. Data from Hawaii and South African kimberlites yield no anomalies and do not support the core contrbution model put forward on the basis of Os-isotope systematics.


Planetary core formation fractionates iron from the silicate mantle, and siderophile (iron loving) elements are strongly enriched in the core while lithophile elements remain in the primitive silicate mantle. Core formation thus causes extreme concentration contrasts between the mantle and the core and even a very small contribution of core material into the mantle will control the siderophile element budget of the admixture.

If it is assumed that the bulk Earth is chondritic, then the bulk core should have near chondritic relative abundances of the highly siderophile elements. The reason is simply that nearly 100% of the highly siderophile elements partitioned into the core. For example, Pt/OsCore ~ Pt/OsChondrite. The isotopic composition of osmium (Os; non-radiogenic isotopes are, e.g., 188Os and 189Os; click here for further information), which changes due to the decay of rhenium (Re; 187Re → 187Os; half-life 41.6 Ga) and platinum (Pt; 190Pt → 186Os; half-life ~490 Ga), will thus be indistinguishable from a chondritic reference or the Primitive Mantle (PM) e.g., [4], which appears to be near-chondritic in its platinum group element (PGE) and Re budget. Discussion of the cause of the PM PGE composition is beyond the scope of this webpage, but one hypothesis stipulates a late veneer of chondritic material.

Crystalisation of the inner core might, by analogy with siderophile element fractionation in iron meteorites, fractionate the siderophile elements [5]. Although this may be a matter of debate, it is assumed here that Os is more compatible in the inner core than Re, which is more compatible than Pt. Inner core crystalisation would thus lead to an outer core that has an excess of Pt and Re relative to Os, i.e., suprachondritic Re/Os and Pt/Os ratios. Because 187Re and 190Pt decay to form 187Os and 186Os, depending on the timing and extent of inner core fractionation, the 187Os/188Os and 186Os/188Os isotope ratios in the outer core may have become more radiogenic than the chondritic reference or the mantle. It thus follows that if the interpretation of Brandon et al. [1;2;6] is correct, that the outer core has suprachondritic Os isotope ratios, then inferences about the crystalisation history and composition of the Earth’s core can possibly be made. One question is; when might the inner core have started to crystalise? Given the very long half-life of 190Pt (~490 Gyr) and the very low abundance of all isotopes of Pt (~0.013%), the inner core must have begun fractionating early in the Earth's history in order to produce more radiogenic 186Os/188Os in the outer core, and in "plume lavas" containing outer core material.

An underlying assumption is, of course, that Pt/Os and Re/Os were indeed fractionated by inner core crystallisation such that the outer core has suprachondritic ratios. This assumption may be at odds with some recent experimental data [7] but supported by others [8].

Brandon et al. [2] infer an outer core 186Os/188Os ratio of 0.119870. This value for the outer core can be achieved for many different Pt/Os ratios and inner core crystalisation timescales. Simply put, the earlier and faster the inner core crystalises, the lower the Pt/Os ratio must have been for the inferred present outer-core ratio of 0.119870 to have been achieved:

  • early, fast inner core crystalisation – low outer core Pt/Os with slow 186Os/188Os and 187Os/188Os increase (for a present-day outer core 186Os/188Os ratio of 0.119870), or
  • late, slow inner core crystalisation – high outer core Pt/Os with fast 186Os/188Os and 187Os/188Os increase.

For a fixed Pt partition coefficient, a lower Pt/Os ratio naturally means a higher Os concentration in the outer core. A higher Os concentration in the outer core translates to higher Os in a fixed core-mantle admixture with a given percentage of core material. Alternatively, from the point of view of surface observations, for any fixed 186Os/188Os ratio that is interpreted as a core-mantle admixture, the core contribution will be smaller if the outer core Os-concentration is higher. It should be noted, however, that the higher the Os concentration postulated for the outer core, the earlier inner core crystalisation is required to begin as the Pt/Os ratio of the outer core decreases.

If mantle plumes start at the base of the mantle and the core began crystalising early enough to develop suprachondritic 186Os/188Os and 187Os/188Os ratios in the outer core, then a small core-mantle admix erupted at the surface would manifest itself in a suprachondritic 186Os/188Os ratio. It would be expected that other siderophile elements would also provide evidence for such an admixture.


Hafnium (Hf) and tungsten (W) are refractory elements with a limited range in chondritic Hf/W ratios [9;10]. It therefore seems justified to assume that the bulk Earth Hf/W ratio is also chondritic. W is siderophile and most of the Earth’s W is in the core [11]. Hf, on the other hand, is lithophile and absent in the Earth’s core. Planetary core formation thus strongly fractionates Hf/W ratios such that they are suprachondritic in silicate mantles and zero in metallic cores. Whilst core formation depletes the mantle of W, partial mantle melting will cause further depletion as W behaves as an incompatible element during mantle partial melting, much like thorium (Th) and lanthanum (La). Like the core, the Earth's crust is therefore also strongly enriched in W relative to the PM (Figure 1).

Figure 1: A cartoon of the Earth and its Hf-W budget. The primitive mantle (PM) W/Th ratio of 0.19 +7/-5 [11] forms the basis for the concentration calculations, which translates into 15 +5/-4 and ~500 ppb W in PM and the bulk core (BC) respectively. Continental crust (CC) generation depletes the mantle further. Averaging this amount over the entire mantle volume produces an 8 ppb W concentration (right half of figure), while a layered mantle scenario produces a strongly depleted upper mantle (DUM) and a virtually undepleted lower mantle (LM) for WPM = 20 ppb (left half of figure).

182Hf decays to 182W with a half-life of ~9 Myr [12], which means that only Hf/W fractionation processes that occurred within the first ~60 Myr of solar system evolution can produce non-chondritic W isotope compositions through 182Hf decay. The W isotope composition is expressed as ε182W, where silicate Earth ε182W = 0 (Figure 2). The current best estimate for the solar system initial value is ε182W = -3.5 [10;13]. The present silicate Earth 182W/184W ratio is 3.5 parts in ten thousand higher than this. Chondritic meteorites have a relatively uniform ε182W of -1.9. The evolution curve of W is shown in Figure 2. If it is assumed that the bulk Earth composition is chondritic, core ε182W is calculated to be about -2 by mass balance. Importantly, the ~2 ε182W difference between the core and the silicate Earth (or PM) will remain constant throughout Earth's history as 182Hf had essentially all decayed away by ~60 Myr after solar system formation. Later mantle fractionation will have had no effect, and core contribution to the mantle is probably the only process that could significantly alter the ε182W composition of the PM later, towards lower ε182W.

Figure 2: Tungsten evolution diagram where t is time in Myr after the condensation of calcium-aluminium rich inclusions and ε182W is the deviation from silicate Earth (or PM) in parts per ten thousand (ε182W = {[182W/184W]spl/[182W/184W]PM–1} x 10,000). Thus, compositions with high 182W/184W plot above the blue line and compositions with low 182W/184W plot below. A chondritic unfractionated planet (brown) would evolve along the chondritic evolution line (purple). Core formation fractionates Hf and W such that the Hf/W ratio of the core is zero (because it contains essentially no Hf) and therefore remains unchanged with time (black horizontal line). The silicate mantle will have a suprachondritic Hf/W ratio and grow increasingly more radiogenic ε182W depending on the timing and extent of Hf/W fractionation. The bulk silicate Earth (BSE; or primitive mantle; blue) has an ε182W that is zero by definition.

The debate regarding a possible Hf-W contribution from the core

Lavas from Hawaii and South African kimberlites do not display resolvable negative anomalies as would be predicted from a core contribution (Figure 3). The immediate question is whether this is due to insufficient sensitivity.

Figure 3: ε182W values for Group I and Group II kimberlites and Hawaiian picrites. For no sample is ε182W resolved from the silicate Earth standard NIST SRM3163. A more detailed discussion of this figure is available in [14].

Mixing models

The sensitivity of Hf-W for detecting a core contribution to proposed plume lavas depends on the isotope composition and concentration contrasts of the mixing reservoirs. The main variables that affect the core-mantle contrast are the amount of W mantle depletion due to core formation, whether mantle depletion due to the formation of the continental crust should be averaged over the entire mantle or only the upper mantle, and whether there are significant amounts of W-rich recycled crust in the source. Recycled crust would be expected to have the same ε182W as the mantle with which it mixes, but a larger absolute concentration. A contribution from the crust would thus act to decrease the concentration contrast between the core and the mantle, thus decreasing the sensitivity of the Hf-W system to a core contribution (c.f. [15]).

Scherstén et al. [14] tested two end-member models for core-mantle mixing (Figure 4). In Model 1, the average W depletion due to core formation (PM = 15 ppb) was combined with average depletion due to continental crust formation over the entire mantle volume. They adopted an outer core Os concentration of 300 ppb, which is the average of inner core crystalisation Models 3 and 4, as described in [2]. In Model 2, a minimum value for depletion due to core formation was used (PM = 20 ppb) and it was assumed that the continental crust was almost entirely drawn from the upper mantle, while the lower mantle that mixed with the core was virtually undepleted. The inner-core-crystalisation Model 1 of Brandon et al. [2] was used to maximise the outer core concentration at 660 ppb. Thus, Model 1 suggests a large contrast in W concentration between the outer core and the mantle, and a high ε182W whereas Model 2 suggests the reverse.

Figure 4: ε182W – 186Os/188Os mixing models for two core-contribution scenarios. The blue line represents Model 1 while the dashed black line represents Model 2 (see text for explanation). Hawaiian picrites are plotted with 2σ error bars.

Scherstén et al. [14] preferred Model 1 and argued that the more extreme Model 2, which was developed to force a fit with their Hawaiian data, is implausible given the large number of end-member assumptions that have to be made. As a result, they concluded that there is no detectable contribution from the Earth’s core in hotspot lavas. However, a recycled crustal component enriched in W may increase the W content of the source such that the sensitivity of W isotopes to a core component is lost [14].

This possibility can be assessed using concentration data, and using Th as a proxy for W. Tungsten behaves as an incompatible element, like Th, in the mantle silicate system. Melting thus depletes the mantle of W. W/Th ratios in rocks from all settings and age have very similar values, with an average of 0.19 ± 0.6 [11]. This shows that W/Th is not significantly fractionated by most geological processes and that the observed ratio of 0.19 is likely to be representative of the primitive mantle. A core contribution would increase the observed W/Th ratio in a melt by a small but insignificant amount (a 0.5% admix would increase the W/Th ratio from 0.19 to 0.25), but the measured Th-concentration of the melts can nevertheless be used to estimate the W concentration of the source.

Using Th as a proxy for W (DTh ~ DW), and assuming a source W/Th ratio of 0.25 for the data from Norman & Garcia [16] (a generous estimate that maximises the estimated W in the source and minimises the effect on the W isotopes of a core contribution), the Hawaiian source is predicted to be moderately depleted with ~10 ppb W, and W isotopes should detect a core contribution if it is present, according to the preferred mixing curve shown in Figure 4 (Model 1).

If the conclusion of Scherstén et al. [14] is correct, that there is no contribution from the Earth’s core to the Hawaiian source, then Os-isotope correlations cannot be used to argue for the existence of mantle plumes from the core-mantle boundary. On the other hand, W isotope systematics cannot disprove their existence.


  1. A.D. Brandon, R.J. Walker, J.W. Morgan, M.D. Norman, and H.M. Prichard, Coupled Os-186 and Os-187 evidence for core-mantle interaction, Science 280, 1570-1573, 1998.
  2. A.D. Brandon, R.J. Walker, I.S. Puchtel, H. Becker, M. Humayun, and S. Revillon, Os-186-Os-187 systematics of Gorgona Island komatiites: implications for early growth of the inner core, Earth and Planetary Science Letters 206, 411-426, 2003.
  3. A.D. Brandon, M.D. Norman, R.J. Walker, and J.W. Morgan, Os-186-Os-187 systematics of Hawaiian picrites, Earth and Planetary Science Letters 174, 25-42, 1999.
  4. T. Meisel, R.J. Walker, and J.W. Morgan, The osmium isotopic composition of the Earth's primitive upper mantle, Nature 383, 517-520, 1996.
  5. R.J. Walker, J.W. Morgan, and M.F. Horan, Os-187 Enrichment in some plumes - Evidence for core-mantle interaction, Science 269, 819-822, 1995.
  6. A.D. Brandon, Os-186-Os-187 systematics of Gorgona komatiites and Iceland picrites, Geochimica et Cosmochimica Acta 66, A100-A100, 2002.
  7. S. Keshav and J.A. Van Orman. Re-Os-Pt Partitioning in sulfur-bearing solid/molten iron metal at 3-22 GPa and 1300-1775 C: Is the Earth's outer core so giving? AGU Chapman Conference " The Great Plume Debate: The Origin and Impact of LIPS and Hot Spots", abstract, p 83-84, 2005.
  8. D. Walker, Core participation in mantle geochemistry: Geochemical Society Ingerson Lecture, GSA Denver, October 1999, Geochimica et Cosmochimica Acta, 64, 2897-2911, 2000.
  9. T. Kleine, K. Mezger, C. Munker, H. Palme, and A. Bischoff, Hf-182-W-182 isotope systematics of chondrites, eucrites, and martian meteorites: Chronology of core formation and early mantle differentiation in Vesta and Mars, Geochimica et Cosmochimica Acta 68, 2935-2946, 2004.
  10. Q.Z. Yin, S.B. Jacobsen, K. Yamashita, J. Blichert-Toft, P. Telouk, and F. Albarede, A short timescale for terrestrial planet formation from Hf-W chronometry of meteorites, Nature 418, 949-952, 2002.
  11. H.E. Newsom, K.W.W. Sims, P.D. Noll, W.L. Jaeger, S.A. Maehr, and T.B. Beserra, The depletion of tungsten in the bulk silicate Earth: Constraints on core formation, Geochimica et Cosmochimica Acta 60, 1155-1169, 1996.
  12. C. Vockenhuber, F. Oberli, M. Bichler, I. Ahmad, G. Quitte, M. Meier, A.N. Halliday, D.C. Lee, W. Kutschera, P. Steier, R.J. Gehrke, and R.G. Helmer, New half-life measurement of Hf-182: Improved chronometer for the early solar system, Physical Review Letters 93, 2004.
  13. T. Kleine, C. Munker, K. Mezger, and H. Palme, Rapid accretion and early core formation on asteroids and the terrestrial planets from Hf-W chronometry, Nature 418, 952-955, 2002.
  14. A.D. Brandon and R.J. Walker, The debate over core-mantle interaction, Earth and Planetary Science Letters 232, 211-225, 2005.
  15. M.D. Norman and M.O. Garcia, Primitive magmas and source characteristics of the Hawaiian plume: petrology and geochemistry of shield picrites, Earth and Planetary Science Letters 168, 27-44, 1999.
last updated 22nd December, 2005