378 MINERALOGICAL MAGAZINE, VOLUME 58A Extinct decay signatures and mantle evolution

C.L. Harper Jr. Department of Earth and Planetary Sciences, Harvard S.B. Jacobsen University, 20 Oxford Street, Cambridge, MA 02138, USA.

Introduction encouraged us to think optimistically about applications requiring precise measurement We are developing three promising extinct radio- capabilities of radiogenic isotope effects at the 5 nuclide systems for studies of early Earth to 10 ppm level. differentiation and geodynamics: 146Sm-142Nd (tl/2 = 103 Ma), lS2Hf2s2w (tl/2 = 9 Ma), and 244pu- 146Sm.142Nd Ru,Pd (tl/2 = 80 Ma). Each of these systems has potential for advancing understanding of different The presence of a small but significant abundance aspects of Earth history. The utility of an extinct of 146Sm in the early solar system provides a isotope system is determined by four factors. First, chronometry particularly well suited for dating the cosmo/geochemical properties of the parent/ very early episodes of differentiation in planetary daughter element pair determine the potential bodies (Harper and Jacobsen, 1992). Our advantages or disadvantages of the system for approach is based on the preservation of isotopic investigating different modes of planetary differ- signatures in large-scale subcrustal reservoirs and entiation and geodynamic transport. Second, the is advantageous in cases where early-formed half-life of the parent defines the timescale for crustal samples are either unavailable, as in the development of isotopic differences between case of Mars (for sampling reasons), and the Earth differentiated reservoirs at early times. Relevant (for tectonic reasons), or severely disturbed, as in 'boundary conditions' of Earth history (accretion the case of the Moon. In pn'nciple, this system is rate, giant impacts, late additions, geodynamic ideal for dating differentiation episodes in the mixing rates, etc.) are a third generic factor and silicate portions of planetary bodies because other set the stage for the geochemical behavior large scale cosmochemical processes such as associated with the events and processes one seeks volatile depletion or core formation do not to understand. The final factor is the range of fractionate Sm/Nd. Large-scale merger events are radiogenic isotope effects in rocks which determines expected to rehomogenize the silicate portion of the practicality of signal resolution by high the merged object and reset the isotopic clock. precision isotopic measurements. We have focused Consequently the age of the oldest differentiated recently on developing mass spectrometry for Nd mantle reservoir should provide a lower limit on and are now able to reproduce some isotope ratios cessation of hierarchical accretion series. These to better than __+4 ppm (___0.04 e-units) at the 26 ~4roperties and linkage to the long-lived level for single measurements. This experience has 7 Sin- 143 Nd system make the 146 Sm- 142 Nd system attractive. AT(Ma) Our first order question is to determine a lower 10 ...... T ,// 0 limit to the age of the last big merger event in the II Nakhla I / 50 accretion history of the Earth for comparison with

100 parallel efforts to determine the age of the Moon.

20O A summary of the most significant 142Nd data for 300 the Earth, Moon (and Mars) is shown in figure 1. 0.0 ...... The terrestrial datum is from the 3.81 Ga old Isua - ~----11g.,~/ Isua 715-28 sample 715-28, which appears to have been -o., derived from a mantle reservoir depleted very early in Earth history (Harper and Jacobsen, 1992; Sharma, et al., 1994). The lunar KREEP datum is -1.0 , , , J , I , , l , l 0.14 0.16 0.18 0.20 0.22 0.24 0.26 from Harvard measurements of two 3.85 Ga old 14~Sm/l~Nd KREEP basalts (14078 and 15386). The 142Nd shift in KREEP probably dates the late-stage FIG. 1. 146Sm-142Ndfossil isochron diagram. freeze-up of the lunar magma ocean. The Mars GOLDSCHMIDT CONFERENCE EDINBURGH 1994 379 datum is for Nakhla (Harper et aL, in prep.). Mars may be particularly useful in addressing the appears to have accreted rapidly (by 4.54 Ga). The question of whether or not contemporary-style inferred accretion interval may be a reasonable plate tectonics was operative in the Hadean. They approximation to the timescale for formation of may also provide a test between some of the Mars-sized objects in the Earth's accretion series. scenarios mentioned in the previous section. For Model ages for the Isua and lunar 'urKREEP" example, if the 715-28 source is a residue of partial sources agree at 4.47 Ga, delimiting the interval melting to form the protocrust, and if the veneer for formation of the Earth-Moon system to be < component was partly sequestered within the 100 Ma. protocrust, then ~n2Nd should correlate with Pd A second objective is to know why so few early and Ru isotopes. Archean samples, (many of which apparently came from much older depleted mantle sources), have 182I-ifol~ w normal 142Nd. Chondritic 142Nd in 3.7 to 4.0 Ga old rocks may be explained in two ways. Either the Hf is a refractory lithophile element, whereas W is actual initial ~143Nd for the source(s) is not high, refractory, moderately siderophile and depleted (ergo no problem to begin with), or very strongly- relative to Hf by a factor of 15 to 30 due to fractionated depleted source reservoirs formed at partitioning into the core. A reasonable assump- ~4.2 Ga. If we assume the former explanation, tion about most of the history of core growth is then we again have two different scenarios. The that it was rate-determined by accretion. chondritic ~42Nd sources were either parts of the Consequently the lS2Hf-182W system should mantle that remained undifferentiated through the provide a chronometer of both terrestrial accre- early Hadean, or products of geodynamic remixing. tion and core formation. In principle one would In the latter case, the non-chondritic 715-28 source want to determine the initial abundance of Lg2Hf was isolated (mantle lithosphere?). We also wish to in the early solar system first from internal address the question of the initial mode of origin of isochron meteorite studies. However, for tech- the high-142Nd Isua 715-28 mantle source. Did it nical reasons we are initially using the Hf/W form by partial melting to form 'primordial crust', fractionation in the silicate Earth to look for or by magma ocean differentiation (crystal evidence of live initial lS2Hf against W from large fractionation)? samples of the Toluca meteorite used as a proxy for the solar system initial composition. A W shelf standard represents the silicate Earth. We 2'Upu-Pd,Ru measure 182W/183W as WO 3", (normalized to 244pu-Pd,Ru chronometry is based on the 184W/183W), by static multicollection with simul- of 244pu. Pu is a refractory taneous 180/160 determination. The standard lithophile element expected to follow Nd in most exhibits an apparent radiogenic mass-182 shift of differentiation processes, and Pd and Ru are both "3__+1 e-units relative to Toluca, confirming refractory and highly siderophile. Equilibrium previous results (Harper et al., 1991). Neutron partitioning during segregation of metal and transport modeling (Masarik and Reedy, pers. sulfide into the core is expected to have depleted comm.) suggests this effect is too large to be due to Pd and Ru by factors > 106. Accretion of a 'late neutron capture burnout during space exposure. If (post-core formation) 'veneer' has been suggested our measurements are accurate (and we have no to explain the relatively small and similar mantle evidence of interferences), then the isotopic depletion factors (~300) observed for all the measurements indicate that 182Hf was live in the highly siderophile refracto2r~44elements. The radio- early solar system, and that the early stages of genic component from Pu in Ru and Pu terrestrial accretion proceeded rapidly (viz., within isotopes ranges from 3 to 10 ppm for/Pu/Ru'pd = a few 182Hf half-lives). 300. Notable features of the veneer scenario are: (i) initial deposition of highly siderophile element References mantle budgets at the early Earth's surface; and (ii) geodynamic mixing into the mantle over [1] Harper (C.L. Jr.) and Jacobsen, (S.B.), (1992) geological time. There is tentative evidence that Nature, 360, 728. the mixing remains incomplete today. The [2] Sharma (M.), Papanastassiou (D.A.), Wasserburg interesting prospect is that early source reservoirs (G.J.) and Dymek(R.F.) (1994) Lunar and may have had much larger j, Pu/Ru,Pd and Planetary Science, XXV, 1253. proportionally greater radiogenic effects. Pd and [3] Harper (C.L.Jr.), Vrlkening (J.) and Heumann, Ru isotopes might therefore provide a tracer of the (K.G.), (199I) Lunar and Planetary Science, XXII, tectonic/geodynamic cycle over Earth history and 515.