Geol. 656 Isotope Geochemistry

Transcription

1 ISOTOPE COSMOCHEMISTRY INTRODUCTION Meteorites are our primary source of information about the early Solar System. Chemical, isotopic, and petrological features of meteorites reflect events that occurred in the first few tens of millions of years of Solar System history. Observations on meteorites, together with astronomical observations on the birth of stars and the laws of physics, are the basis of our ideas on how the Solar System, and the Earth, formed. Meteorites can be divided into two broad groups: primitive meteorites and differenti- Figure Photograph of the meteorite Allende, which fell in Mexico in Circular/spherical features are chondrules. Irregular white patches are ated meteorites. The CAI s. chondrites constitute the primitive group: most of their chemical, isotopic, and petrological features resulted from processes that occurred in the cloud of gas and dust that we refer to as the solar nebula. All chondrites, however, have experienced at least some metamorphism on parent bodies, the small planets (diameters ranging from a few km to a few hundred km) from which meteorites are derived by collisions. The differentiated meteorites, which include the achondrites, stony irons, and irons, were so extensively processed in parent bodies, by melting and brecciation, that information about nebular processes has largely been lost. On the other hand, the differentiated meteorites provide insights into the early stages of planet formation. Chondrites are so called because they contain chondrules, small (typically a few mm diameter) round bodies that were clearly once molten (Figure 16.1). The other main constituents of chondrites are the matrix, which is generally very fine grained, amoeboid olivine aggregates (AOA s), and refractory calcium-aluminum inclusions (generally called CAI s). These last two groups formed by a variety of mechanisms, some of which appear to be complex, but we can generalize and say that all these are grains or aggregates of grains which are also grain that equilibrated with nebular gas at high temperature through condensation and/or evaporation. Most chondrites can be divided into carbonaceous (C), ordinary, and enstatite classes 1. The carbonaceous chondrites are, as their name implies, rich in carbon (as carbonate, graphite, organic matter, and, rarely, microdiamonds) and other volatiles and are further 1 In the last decade or two, additional classes have been added that are defined by rarer meteorites /23/7

2 divided into classes CI, CV, CM, CO, CR, CH, and CB. The CI chondrites lack chondrules and are considered the compositionally the most primitive of all objects. The ordinary chondrites are divided into classes H, L, and LL based on iron content, and enstatite chondrites can be subdivided into EH and EL, also based on iron content. Chondrites are further assigned a petrographic grade on the basis of the extent of metamorphism they have experienced in parent bodies. Grades 4, 5, and 6 have experienced increasing degrees of high-temperature metamorphism, while grades 1 and 2 experienced lowtemperature aqueous alteration. Grade 3 is the least altered. Achondrites are in most cases igneous rocks, some roughly equivalent to terrestrial basalt, others appear to be cumulates. Other achondrites are highly brecciated mixtures. Irons, as they name implies, consist mainly of Fe-Ni metal (Ni content around 5), and can also be divided into a number of classes. Stony-irons are, as their name implies, mixtures of iron metal and silicates. In these two lectures, we focus on the question of the age of meteorites and variations in their isotopic composition. Conventional methods COSMOCHRONOLOGY Meteorite ages are generally taken to be the age of Solar System. The oft cited value for this age is Ga. Before we discuss meteorite ages in detail, we need to consider the question of precisely what event is being dated by radiometric chronometers. Radioactive clocks record the last time the isotope ratio of the daughter element, e.g., 87 Sr/ 86 Sr, was homogenized. This is usually some thermal event. In the context of what we know of early Solar System history, the event dated might be (1) the time solid particles were removed from a homogeneous solar nebula, (2) thermal metamorphism in meteorite parent bodies, or (3) crystallization (in the case of chondrules and achondrites), or (4) impact metamorphism of meteorites or their parent bodies. In some cases, the nature of the event being dated is unclear. The oldest reliable high precision age is from CAI inclusions of Allende, a CV3 meteorite. These give a Pb isotope age of 4.566±.3 Ga. The matrix of Allende seems somewhat younger, although this is uncertain. Thus this age probably reflects the time of formation of the CAI s. Precise Pb-Pb ages of Ga have been reported by several laboratories for the St. Severin LL chondrite. The same age (4.552±.3 Ga) has been reported for 2 L5 chondrites. U-Pb ages determined on phosphates in equilibrated (i.e., petrologic classes 4-6) ordinary chondrites range from to 4.52 Ga. As these phosphates are thought to be secondary and to have formed during metamorphism, these ages apparently represent the age of metamorphism of these meteorites. Combined whole rock Rb-Sr ages for H, E, and LL chondrites are 4.498±.15 Ga. However, within the uncertainty of the value of the 87 Rb decay constant, this age could be Ga (uncertainties normally reported on ages are based only on the scatter about the isochron and the uncertainty associated with the analysis, they do not include uncertainty associated with the decay constant). The age of Allende CAI s thus seems 5 Ma older than the oldest ages obtained on ordinary chondrites. No attempt has been made at high-precision dating of CI chondrites as they are too fine-grained to separate phases. Pb isotope ages of the unusual achondrite Angra dos Reis, often classed by itself as an angrite but related to the Ca-rich achondrites, give a very precise age of ±.4 Ma. Ibitira, a unique unbrecciated eucrite (achondrite), has an age of 4.556±.6 Ga. Perhaps surprisingly, these ages are the same as those of chondrites. This suggests that the parent body of these objects formed, melted, and crystallized within a very short time interval. Not all achondrites are quite so old. A few other high precision ages (those with quoted errors of less than 1 Ma) are available and they range from this value down to 4.529±.5 Ga for Nueve Laredo. Thus the total range of the few high precision ages in achondrites is about 3 million years. K-Ar ages are often much younger. This probably reflects Ar outgassing as a result of collisions. These K-Ar ages therefore probably date impact metamorphic events rather than formation ages /23/7

3 The present state of conventional meteorite chronology may be summarized by saying that it appears the meteorite parent bodies formed around 4.56±.5 Ga, and there is some evidence that hightemperature inclusions (CAI's: calcium-aluminum inclusions) and chondrules in carbonaceous chondrites may have formed a few Ma earlier than other material. Resolving events on a finer time-scale than this has proved difficult using conventional techniques. There are, however, other techniques that help to resolve events in early solar system history, and we now turn to these. Initial Ratios Attempts have been made to use initial isotope ratios to deduce a more detailed chronology, but these have been only moderately successful. Figure 16.2 shows initial 87 Sr/ 86 Sr ratios of meteorites and lunar rocks and a time scale showing how 87 Sr/ 86 Sr should evolve in either a chondritic or solar reservoir. The reference 'initial' 87 Sr/ 86 Sr of the solar system is taken as.69897±3, based on the work of Papanastassiou and Wasserburg (1969) on basaltic achondrites (this value is known as BABI: basaltic achondrite best initial). Basaltic achondrites were chosen since they have low Rb/Sr and hence the initial ratio (but not the age) is well constrained in an isochron. Subsequent high precision analyses of individual achondrites yield identical results, except for Angra Dos Reis and Kapoeta, which have slightly lower ratios: This suggests their parent body(ies) were isolated from the solar system somewhat earlier. CAI's and Rb-poor chondrules from Allende have an even lower initial ratio:.69877±3. Allende chondrules appear to be among the earliest formed objects. The parent body of the basaltic achondrites appears to have formed 1 to 2 Ma later. Note there is no distinction in the apparent age of the oldest lunar rocks and the basaltic achondrites: from this we may conclude there was little or no difference in time of formation of the moon, and presumably the Earth, and the basaltic achondrite parent body. The initial 143 Nd/ 144 Nd ratio of the solar system is taken as.5669±8 (normalized to 143 Nd/ 144 Nd =.7219) based on the work on chondrites of Jacobsen and Wasserburg (198). Achondrites seem to have slightly higher initial ratios, suggesting they formed a bit later. The initial isotopic composition of Pb is taken from the work of Tatsumoto et al. (1973) on troilite from the Canyon Diablo iron meteorite as 26 Pb/ 24 Pb: 9.37, 27 Pb/ 24 Pb: 1.294, 28 Pb/ 24 Pb: These values are in agreement with the best initial values determined from chondrites, including Allende chondrules. More recent work by Chen and Wasserburg (1983) confirms these results, i.e.: 9.366, Figure Initial Sr isotope ratios plotted against a time scale for 87 Sr/ 86 Sr assuming a chondritic Rb/Sr ratio. After Kirsten (1978) /23/7

4 1.293, and respectively. EXTINCT RADIONUCLIDES There is evidence that certain short-lived nuclides once existed in meteorites. This evidence consists of the anomalous abundance of nuclides, for example, 129 Xe, known to be produced by the decay of short-lived radionuclides, e.g., 129 I, and correlations between the abundance of the radiogenic isotope and the parent element. Consider, for example, 53 Cr, which is the decay product of 53 Mn. The half-life of 53 Mn, only 3.7 million years, is so short that any 53 Mn produced by nucleosynthesis has long since decayed. If 53 Mn is no longer present, how do we know that the anomalous 53 Cr is due to decay of 53 Mn? We reason that the abundance of 53 Mn, when and if it was present, should have correlated with the abundance of other isotopes of Mn. 55 Mn is the only stable isotope of Mn. So we construct a plot similar to a conventional isochron diagram (isotope ratios vs. parent/daughter ratio), but use the stable isotope, in this case 55 Mn as a proxy for 53 Mn. An example is shown in Figure Starting from our basic equation of radioactive decay, we can derive the following equation: D = D + N (1 e t ) 16.1 This is a variation on the isochron equation we derived in lecture 4. Written for the example of the decay of 53 Mn to 53 Cr, we have: 53 Cr Cr = Cr 52 Cr + 53 Mn 52 Cr (1 ' e '(t ) 16.2 where the subscript naught denotes an initial ratio, as usual. The problem we face is that we do not know the initial 53 Mn/ 52 Cr ratio. We can, however, measure the 55 Mn/ 53 Cr ratio. Assuming that initial isotopic composition of Mn was homogeneous in all the reservoirs of interest; i.e., 53 Mn/ 55 Mn is constant, the initial 53 Mn/ 52 Cr ratio is just: 53 Mn 52 Cr = 55 Mn 52 Cr 53 Mn 55 Mn Of course, since 55 Mn and 52 Cr are both non-radioactive and non-radiogenic, the initial ratio is equal to the present ratio (i.e., this ratio is constant through time). Substituting 16.3 into 16.2, we have: 53 Cr Cr = Cr 52 Cr + 55 Mn 52 Cr 53 Mn 55 Mn 16.3 (1' e '(t ) 16.4 Finally, for a short-lived nuclide like 53 Mn, the term λt is very large after 4.55 Ga, so the term e λt is (this is equivalent to saying all the 53 Mn has decayed away). Thus we are left with: 53 Cr Cr = Cr 52 Cr Figure Correlation of the 53 Cr/ 52 Cr ratio with 55 Mn/ 52 Cr ratio in inclusions from the Allende CV3 meteorite. After Birck and Allegre (1985) Mn 52 Cr 53 Mn 55 Mn /23/7

5 On a plot of 53 Cr/ 52 Cr vs. 55 Mn/ 52 Cr, the slope is proportional not to time, as in a conventional isochron diagram, but to the initial 53 Mn/ 55 Mn ratio. Thus correlations between isotope ratios such as these is evidence for the existence of extinct radionuclides. In this way, many extinct radionuclides have been identified in meteorites from variations in the abundance of their decay products. The most important of these are listed in Table On a cosmic scale, nucleosynthesis is a more or less continuous process roughly Table Short-Lived Radionuclides in the Early Solar System Radio- Half-life Decay Daughter Abundance nuclide Ma Ratio 1 Be 1.5 β 1 B 26 Al.7 β 26 Mg 41 Ca.15 β 41 K 53 Mn 3.7 β 53 Cr 6 Fe 1.5 β 6 Ni 17 Pd 9.4 β 17 Ag 129 I 16 β 129 Xe 146 Sm 13 α 142 Nd 182 Hf 9 β 182 W 244 Pu 82 α, SF Xe 1 Be/ 9 Be ~ Al/ 27 Al ~ Ca/ 4 Ca Mn/ 55 Mn ~ Fe/ 56 Fe ~ Pd/ 18 Pd ~ I/ 127 I ~ Sm/ 144 Sm ~ Hf/ 18 Hf ~ Pu/ 238 U ~.5 once every second, a supernova explodes somewhere in the universe. So we might expect that interstellar dust might contain some of the longer-lived of these nuclides at low concentrations. However, such events are much rarer on a local scale (fortunately for us), and the shorter-lived of these nuclides must have been synthesized nearby shortly (on geological time scales) before the solar system formed. To understand why these short-lived radionuclides require a nucleosynthetic event, consider the example of 53 Mn. Its half-life is 3.7 Ma. Hence 3.7 Ma after it was created only 5 of the original number of atoms would remain. After 2 half-lives, or 7.4 Ma, only 25 would remain, after 4 half-lives, or 14.8 Ma, only of the original 53 Mn would remain, etc. After 1 half lives, or 37 Ma, only 1/2 1 (.1) of the original amount would remain. The correlation between the Mn/Cr ratio and the abundance of 53 Cr indicates some 53 Mn was present when the meteorite, or its parent body, formed. From this we can conclude that an event that synthesized 53 Mn occurred not more than roughly 3 million years before the meteorite formed. We will return to this issue in the next lecture. 129 I 129 Xe and 244 Pu Among the most useful of these shortlived radionuclides, and the first to be discovered, has been 53 I, which decays to 129 Xe. Figure 16.4 shows the example of the analysis of the meteorite Khairpur. In this case, the analysis in done in a manner very analogous to 4 Ar- 39 Ar dating: the sample is first irradiated with neutrons so that 128 Xe is produced by neutron capture and subsequent decay of 127 I. The amount of 128 Xe produced is proportional to the amount of 127 I present (as well as the neutron flux and reaction cross section). The sample is then heated in vacuum through a series of steps and the Xe released at each step analyzed in a mass spectrometer. As was the case in Figure 16.3, the slope is proportion to the 129 I/ 127 I ratio at the time the meteorite formed. Figure Correlation of 129 Xe/ 13 Xe with 128 Xe/ 13 Xe. The 128 Xe is produced from 127 I by irradiation in a reactor, so that the 128 Xe/ 13 Xe ratio is proportional to the 127 I/ 13 Xe ratio. Numbers adjacent to data points correspond to temperature of the release step /23/7

6 In addition to 129 Xe produced by decay of 129 I, the heavy isotopes of Xe are produced by fission of U and Pu. 244 Pu is of interest because it another extinct radionuclide. Fission does not produce a single nuclide, rather a statistical distribution of many nuclides. Each fissionable isotope produces a different distribution. The distribution produced by U is similar to that produced by 244 Pu, but the difference is great enough to demonstrate the existence of 244 Pu in meteorites, as is shown in Figure Fission tracks in excess of the expected number of tracks for a known uranium concentration are also indicative of the former presence of 244 Pu. These extinct radionuclides provide a Figure Variation of 134 Xe/ 132 Xe and 136 Xe/ 132 Xe in meteorites (5). The isotopic composition of fission products of man-made 244 Pu is shown as a star ( ). After Podosek and Swindle (1989). Figure Summary of I-Xe ages of meteorites relative to Bjurböle. After Swindle and Podosek (1989). means of relative dating of meteorites and other bodies. Of the various systems, the 129 I 129 Xe decay is perhaps most useful. Figure 16.6 shows relative ages based on this decay system. These ages are calculated from 129 I/ 127 I ratios, which are in turn calculated from the ratio of excess 129 Xe to 127 I. Since the initial ratio of 129 I/ 127 I is not known, the ages are relative to an arbitrary value, which is taken to be the age of the Bjurböle meteorite, a L4 chondrite. The ages date closure of the systems to Xe and I mobility, but it is not clear if this occurred at condensation or during metamorphism. Perhaps both are involved. The important point is that there is only slight systematic variation in age with meteorite types. Carbonaceous chondrites do seem to be older than ordinary and enstatite chondrites, while LL chondrites seem to be the youngest. Differentiated meteorites are generally younger. These are not shown, except for silicate in the El Taco iron, which is not particularly young. The bottom line here is that all chondrites closed to the I-Xe decay system within about 2 Ma /23/7

7 An interesting aspect of Figure 16.6 is that the achondrites, which are igneous in nature, and the irons are at most only slightly younger on average than the chondrites. Irons and achondrites are both products of melting on meteorite parent bodies. That they appear to be little younger than chondrites indicates that and melting and differentiation of those planetismals must have occurred very shortly after the solar system itself formed and within tens of millions of years of the synthesis of 129 I. 17 Pd 17 Ag The existence of variations in isotopic composition of silver, and in particular variations in the abundance of 17 Ag that correlate with the Pd/Ag ratio in iron meteorites indicates that 17 Pd was present when the irons formed. The half-life of 17 Pd is 9.4 million years; hence the irons must have formed within a few tens of millions of years of synthesis of the 17 Pd. This in turn implies that formation of iron cores within small planetary bodies occurred within a few tens of millions of years of formation of the solar system. Fractions of metal from the meteorite Gibeon (IVA iron) define a fossil isochron indicating an initial 17 Pd/ 18 Pd ratio of (Chen and Wasserburg, 199). Other IVA irons generally fall along the same isochron (Figure 16.7). IIAB and IIAB irons, as well as several anomalous irons show 17 Ag/ 19 Ag 18 Pd/ 19 Ag correlations that indicate 17 Pd/ 18 Pd ratios between 1.5 and Assuming these differences in initial 17 Pd/ 18 Pd are due to time and the decay of 17 Pd, all of these iron meteorites would have formed no more than 1 million years after Gibeon (Chen and Wasserburg, 1996). 26 Al 26 Mg Figure Correlation of 17 Ag/ 19 Ag with 18 Pd/ 19 Ag in Group IVA iron meteorites, demonstrating the existence of 17 Pd at the time these irons formed. After Chen and Wasserburg (1984). Another key extinct radionuclide has been 26 Al. Because of its short half-life (.72 Ma), it provides much stronger constraints on the amount of time that could have passed between nucleosynthesis and processes that occurred in the early solar system. Furthermore, the abundance of 26 Al was such that it s decay could have been a significant source of heat. 26 Al decays to 26 Mg; an example of the correlation between 26 Mg/ 24 Mg and Figure Al-Mg evolution diagram for Allende CAI WA. Slope of the line corresponds to an initial 26 Al/ 27 Al ratio of 27 Al/ 24 Mg is shown in Figure Al/ 27 Al ratio of After Lee et al. (1976). 12 2/23/7

8 Because of the relatively short half-life of 26 Al and its potential importance as a heat source, considerable effort has been devoted to measurement of Mg isotope ratios in meteorites. Most of this work has been carried out with ion microprobes, which allow the simultaneous measurement of 26 Mg/ 24 Mg and 27 Al/ 24 Mg on spatial scales as small as 1 µ. As a result, there are some 15 measurements on 6 meteorites reported in the literature, mostly on CAI s. The reason for the focus on CAI s is, of course, because their high Al/Mg ratios should produce higher 26 Mg/ 24 Mg ratios. Figure 16.9 summarizes these Figure Inferred initial 26 Al/ 27 Al for all available meteoritic data. After MacPherson et al. (1995). data. These measurements show a maximum in the 26 Al/ 27 Al ratio of around Significant 26 Mg anomalies, which in turn provide evidence of 26 Al, are mainly confined to CAI s. This may in part reflect the easy with the anomalies are detected in this material and the focus of research efforts, but it almost certainly also reflects real differences in the 26 Al/ 27 Al ratios between these objects and other materials in meteorites. This in turn probably reflects a difference in the timing of the formation of the CAI s and other materials, including chondrules. The evidence thus suggests that CAI s formed several million years before chondrules and other materials found in meteorites. Extinct Radionuclides in the Earth Several of the short-lived radionuclides listed in Table 16.1 have half-lives sufficiently long that they should have been present in the early Earth. Of greatest interest are 129 I, 182 Hf, and 146 Sm. The decay products of these nuclides are 129 Xe, 182 W, and 142 Nd, an atmophile, a siderophile, and a lithophile element respectively. Their distribution can tell us about the early evolution of the Earth s atmosphere, core, and mantle. Here we ll consider 182 Hf and 142 Sm. We ll discuss 129 I in the lecture on the origin and evolution of the atmosphere. 182 Hf 182 W and Core Formation The Hf-W pair is particularly interesting because Hf is lithophile while W is moderately siderophile. Thus the 182 Hf- 182 W decay system should be useful in dating silicate-metal fractionation, including core formation in the terrestrial planets and asteroids. Both are highly refractory elements, while has the advantage the one can reasonably assume that bodies such as the Earth should have a chondritic Hf/W ratio, but the disadvantage that both elements are difficult to analyze by conventional thermal ionization. These observations have led to a series of measurements of W isotope ratios on terrestrial materials, lunar samples, and a variety of meteorites, including those from Mars. The conclusions have evolved and new measurements have become available. Among other things, the story of Hf-W illustrates the importance of the fundamental dictum in science that results need to be independently replicated before they be accepted. Because the variations in 182 W/ 183 W ratio are quite small, they are generally presented and discussed in the same ε notation used for Nd and Hf isotope ratios. There is a slight difference, however; ε W is the deviation in parts per 1, from a terrestrial tungsten standard, and ƒ Hf/W is the fractional deviation of the Hf/W ratio from the chondritic value. Assuming that the silicate Earth has a uniform W isotope composition identical to that of the standard (an assumption which has not yet been proven), then the 121 2/23/7

9 silicate Earth has ε W of by definition. The basic question can posed this way: if the 182 W/ 183 W ratio in the silicate Earth is higher than in chondrites, it would mean that much of the Earth s tungsten had been sequestered in the Earth s core before 182 Hf had ε W Carbonaeous Chondrite Chondrite Achondrite (Eucrite) Initial 182 Hf/ 18 Hf = 1. x Hf/ 18 Hf = 1.1 x 1-5 (29.5 Ma) Silicate Earth entirely decayed. Since the half-life of 182 Hf is 9 Ma and using our rule of thumb that a radioactive nuclide is fully decayed in 5 to 1 half-lives, this would mean the core must have formed within 45 to 9 million years of the time chondritic meteorites formed (i.e., of the formation of the solar system). If on the other hand, the 182 W/ 183 W ratio in the silicate Earth was the same as in chondrites, which never underwent silicate-metal fractionation, this would mean that at least 45 to 9 million years must have elapsed (enough time for 182 Hf to fully decay) between the formation of chondrites and the formation of the Earth s core. Anomalous W isotopic compositions were first found in the IA iron Toluca by Harper et al. (1991). They found the 182 W/ 183 W ratio in the meteorite was 2.5 epsilon units (i.e., parts in 1,) lower than in terrestrial W. This value was revised to -3.9 epsilon units by subsequent, more precise, measurements (Jacobsen and Harper, 1996). Essentially, the low 182 W/ 183 W ratio indicates Toluca metal separated from Hf-bearing silicates before 182 Hf had entirely decayed. Because of the difference between terrestrial W, Jacobsen and Harper (1996) concluded the Earth s core must have segregated rapidly. At this point, however, no measurements had yet been made on chondritic meteorites, which never underwent silicate-iron fractionation, so the conclusion was tentative. Lee and Halliday (1995) reported W isotope ratios for 2 carbonaceous chondrites (Allende and Murchison), two additional iron meteorites (Arispe, IA, and Coya Norte, IIA) and a lunar basalt. They found the iron meteorites showed depletions in 182 W (ε W = -4.5 and -3.7 for Arispe and Coya Norte respectively) that were similar to that observed in Toluca reported by Jacobsen and Harper (1996). The chondrites, however, had ε w values that were only slightly positive, about +.5, and were analytically indistinguishable from terrestrial W, as was the lunar basalt. Lee and Halliday (1995) inferred an initial 182 Hf/ 18 Hf for the solar nebula of , much higher than assumed by Jacobsen and Harper. Based on this similarity of isotopic compositions of chondritic and terrestrial W, Lee and Halliday (1995) concluded that the minimum time required for formation of the Earth s core was 62 million years. Subsequently, Lee and Halliday (1998) reported ε W values of +32 and +22 in the achondrites Juvinas and ALHA These large differences in W isotopic composition meant that metal-silicate fractionation, i.e., core formation, occurred quite early in the parent bodies of achondritic meteorites; in other words, asteroids or planetismals must have differentiated to form iron cores and silicate mantles very early, virtually simultaneous with the formation of the solar system. This is consistent with other evidence discussed above for very little age difference between differentiated and undifferentiated meteorites. Lee and Halliday (1998) also reported ε W values in the range of +2 to +3 in 3 SNC meteorites f Hf/W Moon Figure W isotope ratios in meteorites, the Moon and the Earth reported by Yin et al. (22) /23/7

10 thought to have come from Mars. These data indicated that the Martian core formed relatively early. The heterogeneity in tungsten isotopes indicates in Martian mantle was never fully homogenized. Lee et al. (1997) reported that the W isotope ratio of the Moon was about 1 epsilon unit higher than that of terrestrial W. Thus at this point, the Earth appeared to be puzzlingly anomalous among differentiated planetary bodies in that silicatemetal differentiation appeared to have occurred quite late. In the latest chapter of this story, Yin et al. (22) reported W isotope measurements carried out in two laboratories, Harvard University and the Ecole Normale Supérieure de Lyon, which showed that the chondrites Allende and Murchison which showed that they had W isotope ratios 1.9 to 2.6 epsilon units lower than the terrestrial standard (Figure 16.1). In the same issue of the journal Nature, Kleine et al., (22) reported similarly low ε W (i.e., -2) for the carbonaceous chondrites Allende, Orgueil, Murchison, Cold Bokkeveld, Nogoya, Murray, and Karoonda measured Carbonaceous chondrites -6-5 Toluca ε W in a third laboratory (University of Münster). Furthermore, Kleine et al. (22) analyzed a variety of terrestrial materials and found they all had identical W isotopic composition (Figure 16.11). It thus appears that the original measurements of Lee and Halliday (1995) were wrong. The measurement error most likely relates to what was at the time an entirely new kind of instrument, namely the multicollector ICP-MS. Yin et al. (22) also analyzed separated metal and silicate fractions from two ordinary chondrites (Dhurmsala and Dalgety Downs) that allowed them to estimate the initial 182 Hf/ 18 Hf of the solar system as Yin et al. (22) considered two scenarios for the formation of the core (Figure 16.12). In the first, which they call the two-stage model in which the Earth first accretes (stage 1) and then undergoes core formation (stage 2), induced by the giant impact that forms the moon. In this scenario, core formation occurs 29 million years after formation of the solar system. In the second scenario that they believed more likely, metal segregates continuously from a magma ocean. In this continuous model, the mean age of core formation is 11 million years. In contrast, they concluded that the parent body of the eucrite class of achondrites (suspected to be the large asteroid Vesta) underwent core formation within 3 million years of formation of the solar system. Klein et al. (22) reached similar conclusions. -4 Karooonda Murray Nogoya Cold Bokkeveld Murchison Orgueil Allende -3 a bc a b IGDL-GD G1-RF BB BE-N Terrestrial samples Figure W isotope ratios measured in chondrites, the iron meteorite Toluca, and terrestrial materials by Kleine et al. (22) /23/7

11 146 Sm- 142 Nd As we have mentioned, geochemists generally assume that rare earth and other refractory elements have the same relative concentrations in the Earth as they have in chondrites. If so, the Sm/Nd ratio of the Earth should be chondritic. Thus the 147 Sm/ 144 Nd ratio of the present Earth should be chondritic and the 146 Sm/ 144 Nd of the early Earth should have been chondritic. Thus the 143 Nd/ 144 Nd and 142 Nd/ 144 Nd of the bulk earth should also be chondritic. However, recent studies of the 142 Nd/ 144 Nd ratio in chondrites and terrestrial materials suggest that this may not be the case, at least that part of the Earth accessible to sampling. This is surprising to say the least. These two elements are very similar to each other in chemical behavior, having identical configurations of electrons in bonding orbitals, and are both refractory lithophile elements. Indeed, Nd and Sm have 5 condensation temperatures of 162 and 159 K, respectively. It is difficult to see how processes operating in the solar nebula could have fractionated these elements significantly. The total range of high precision Sm/Nd ratio measurements in chondrites varies by less than 3, which would seem to confirm that these elements were not fractionated in the solar nebula. 142 Nd is the product of α- Δε W Magma ocean model 11±1 Myr 29.5±1.5 Myr Two-stage model Lee Halliday (1995) Mean time of core formation (Myr) Figure Models for timing of core formation in the Earth. The figure shows how the difference between the 182 W/ 183 W between the silicate Earth and chondrites, Δε W, declines as a function of time between formation of the chondrites and separation of the Earth s core. Yin et al. (22) considered two scenarios: a two-stage model in which Earth first accretes completely and then the core forms, and a model in which the core segregates progressively from a magma ocean as the Earth accretes. In the first scenario, the mean age of the core is about 3 million years, in the second it is 11 million years. These results are sharply different from those of Lee and Halliday (1995) who found only a small difference in ε W between the Earth and chondrites and consequently concluded the core formed later (at about 6 million years). decay of 146 Sm, a nuclide with a half-life of 13 million years. As Table 16.1 shows, the initial 146 Sm/ 144 Sm ratio of the solar system about.8, a value deduced from 142 Nd/ 144 Nd variations in meteorites using procedures discussed above. 144 Sm is the least abundant isotope of Sm, comprising only 3 of natural Sm, so even initially, 146 Sm would have only constituted.25 of Sm. Because of this and because the range of Sm/Nd ratios in nature is small, any variations in the 142 Nd/ 144 Nd ratio should be quite small, no more than a few 1 s of ppm. Detecting such small variations is analytically challeng /23/7

12 ing, and indeed was nearly impossible before about 15 years ago. Furthermore, because the half-life of 146 Sm is short, any variation in 142 Nd/ 144 Nd must be the result of fractionation occurring in at most the first few hundred million years of solar system or Earth history. However, considerable variation in the 142 Nd/ 144 Nd ratio had been detected in SNC meteorites, which suggested early mantle differentiation on Mars. It thus seemed reasonable to look for such variations on Earth. Geochemists focused their initial attention on early crustal rocks from the Isua area in Greenland. Some rocks from this area are as old as 3.8 Ga and have initial 143 Nd/ 144 Nd ratios several epsilon units above the chondritic value, suggested there were derived from an incompatible element-depleted mantle with high Sm/Nd. A study by Harper and Jacobsen (1992) reported a 33 ppm excess of 142 Nd in one 3.8 Ga old metavolcanic rock from Isua. This excess was based on a comparison between the rock and laboratory standards; the latter was assumed to have the same 142 Nd/ 144 Nd ratio as chondrites. Other workers failed to find any excesses in other rocks from Isua, so these results were controversial. More recent work using advanced mass spectrometers by Caro et al. (23) and Boyet et al. (23), however, has confirmed the original findings of Harper and Jacobsen. This means that these early parts of the crust formed from a mantle reservoir that had Sm/Nd ratios higher than the chondritic one and importantly, that this reservoir formed very early, most likely within the first 1 Ma. A yet more surprising result came when Boyet and Carlson (25) analyzed the 142 Nd/ 144 Nd ratios of meteorites and found that terrestrial rocks had 142 Nd/ 144 Nd ratios that average 2 ppm or.2 epsilon units higher than chondrites, and most eucrites as well (Figure 16.13). This implies that the accessible Earth has a significantly higher Sm/Nd ratio than chondrites. How much higher depends on when the increase occurred. If the increase occurred 5 million years after the beginning of the solar system (taken as the age of Figure Variation in ε 142Nd in the Earth and meteorites. Gray region is the range measurements of laboratory standards derived from terrestrial Nd. All other terrestrial materials plot within this range with the exception of some samples from Isua, Greenland. Chondrites have, on average, ε 142Nd of -.2 relative to the terrestrial standards. Data from Caro et al. (23), Boyet and Carlson (25), Boyet and Carlson (26). SNC data from the compliation of Halliday (21) /23/7

13 CAI s), the Sm/Nd ratio of the accessible Earth would have to be 8 higher than chondrites; if the increase occurred at 3 million years, it would have to be 1 higher. If the increase occurred later, the Sm/Nd ratio would have to be even higher. This increase in Sm/Nd might seem small; after all, we have already stated that the assumption that the Earth has chondritic abundances of refractory elements is probably only valid to 1. Yet this small difference is very important in interpretation of Nd isotope systematics. For the two scenarios above, 5 Ma and 3 Ma, the ε Nd of the accessible Earth would be +8.4 and +1.1 respectively. These values fall within the range of values of mid-ocean ridge basalts. Recalling that the Isua samples have a 3 ppm excess in 142 Nd relative to a terrestrial standard, this means that the Isua samples have a 5 ppm excess in 142 Nd relative to chondrites. How might the increase in Sm/Nd come about? First, we need to recall that meteorites come from the asteroid belt and their compositions might not be representative of the composition of the inner solar nebula from which the Earth and the other terrestrial planets formed. Its possible the inner solar nebula had a higher Sm/Nd ratio. That said, it is very difficult to see why this should be so. The observable fractionation in primitive meteorites relates to volatility and lithophile/siderophile tendency. As we have seen, Sm and Nd have quite high and very similar condensation temperatures and neither shows a significant siderophile tendency. Although the possibility cannot be excluded, there is simply no good reason to believe that the Sm/Nd ratio of the Earth should be different from chondrites. If the Earth does have the same Sm/Nd ratio as chondrites, then the Sm/Nd ratio of the accessible Earth must be higher than that of the Earth as a whole. Since neither Sm nor Nd should be present to any significant degree in the core, this implies there is an unsampled reservoir in the mantle with a lower than chondritic Sm/Nd ratio. Furthermore, since 146 Sm has a half-life of only 13 Ma, the differentiation that produced high and low Sm/Nd reservoirs must have occurred very early in Earth s history. Fractional crystallization of a magma ocean might seem an obvious candidate for this event. Indeed, Boyet and Carlson (25) suggested that crystallization of the terrestrial magma ocean left a layer of residual melt, similar to the KREEP source on the Moon. They termed this hypothetical reservoir the early enriched reservoir (EER) and its compliment the early depleted reservoir (EDR). The EER would be created in the upper mantle, but since it is unsampled by volcanism and tectonism, the unsampled mantle reservoir should be in the deep mantle. Boyet and Carlson (25) noted that if it were rich in Fe and Ti, as is the lunar KREEP reservoir is, once crystallized the EER may have sunk into the deep mantle, where it remains because if its high density. So in their scenario, the EDR forms in lower mantle but ends up becoming the part of the mantle that produced the continental crust and continues to be sampled by volcanism today. In other words, the EDR comprises the accessible mantle. The EER could be the product of fractional crystallization of a mantle that was initial entirely or largely molten. In that case, the principal crystallizing phases would be the deep mantle minerals, the two perovskite phases and magnesiowüstite. Judging from partition coefficients published by Corgne et al. (25), a cumulate layer formed of Mg-perovskite should have a Sm/Nd ratio over twice that of chondrites, far too high to be appropriate for the accessible mantle. This could be mixed with unfractionated mantle material. However, because Sm and Nd concentrations would be low in the perovskite cumulate, a great deal of it would be necessary: the EDR would be composed of about 75 Mgperovskite cumulate. This reservoir would be quite depleted in most incompatible elements, with Sm and Nd concentrations only about 1/3 those of the BSE and would have ratios of some refractory elements that are very different from chondritic. Another possibility is that Mg- and Ca-perovskite crystallized together to form the cumulate. This requires about 7-75 fractional crystallization, judging from the partition coefficients of Corgne et al. (25). Interestingly, because Sm and Nd partition into Ca-perovskite, an EDR created in this way would have concentrations of many lithophile trace elements, including the REE, U, and Th, that are close to or slightly higher than BSE. However, ratios of some elements, such as Sc/Sr and Ba/Sm, would be very different from chondritic. This problem The two standards commonly used in Nd isotope ratios measurements are the La Jolla standard and the Ames standard. Both are solutions created from industrially purified Nd /23/7

14 seems to preclude the possibility of creating the EDR by crystallization of a magma ocean that extended into the deep mantle. Boyet and Carlson (25) suggested instead that crystallization of the magma ocean involved purely upper mantle phases. This would be the case if the magma ocean were relatively shallow and did not extend substantially deeper than 66 km. The problem with this idea is that region above constitutes only 25 of the mass of the mantle and it is difficult to create an enriched reservoir within it with sufficiently low Sm/Nd that the remaining mantle would have Sm/Nd 1 greater than chondritic. In the upper mantle, the phase that is likely to fractionate Sm and Nd most is majorite garnet. However, even majorite does not seem to fractionate Sm and Nd enough to do this. Using partition coefficients published by Corgne and Wood (24), no extent of fractional crystallization of majorite from an upper mantle magma ocean produces a sufficiently enriched residual melt layer to leave the rest of the mantle with a Sm/Nd ratio that is 1 higher than chondritic. Yet another alterative is that a primordial crust was created by partial melting of an already solidified mantle. That crust would have been enriched in incompatible elements just as the modern crust is. The crust may have destabilized in some way and been recycled back into the deep mantle. Under certain circumstances, its density might be greater than that of ordinary mantle peridotite such that if forms a stable layer in the deep mantle, perhaps in D. Boyet and Carlson (25) calculated that if this Early Enriched Reservoir (EER) occupied the volume of D, it would have to be as nearly enriched in incompatible elements as the present continental crust. If the EER comprises the region deeper than 16 km, it need be only twice as enriched in incompatible elements as the bulk silicate Earth. There are significant problems with all of the scenarios; none seems entirely satisfactory. An additional problem with them all is that there is no seismic evidence for chemical layering in the deep mantle. Indeed, tomographic imaging of the mantle shows seismically fast regions extending from subduction zones at the surface to the deep mantle. These regions are presumably sinking oceanic lithosphere. Other images show mantle plumes rising from near the core-mantle boundary to the surface. Both suggest whole mantle convection and consequently, whole mantle mixing. It is difficult to see how any chemical layer could survive. Thus the discovery that the 142 Nd/ 144 Nd ratio of the accessible Earth is 2 ppm higher than chondrites presents a difficult and intricate problem for geochemistry: a true conundrum. It is certainly the result of something that happened very early. If it reflects fractionation is the solar nebula, then our understanding of nebular chemistry is weaker than we realize. If it is the product of early differentiation of the Earth, then our understanding of both early planetary processes and the chemistry of the Earth may be poorer than we had thought. In particular, it potentially invalidates our estimates of the composition of the Earth, and also models of its evolution. REFERENCES AND SUGGESTIONS FOR FURTHER READING Birck, J. L. and C. J. Allègre Evidence for the presence of 53 Mn in the early solar system. Earth Planet. Sci. Lett. Geophys. Res. Lett.: Boyet, M., J. Blichert-Toft, M. Rosing, M. Storey, P. Telouk and F. Albarede, 142 Nd evidence for early Earth differentiation, Earth Planet. Sci. Lett., 214: , 23. Boyet, M. and R. L. Carlson, 142 Nd evidence for early (>4.3 Ga) global differentiation of the silicate Earth, Science, 39: , 25. Caro, G., B. Bourdon, J.-L. Birck and S. Moorbath, 146 Sm- 142 Nd evidence from Isua metamorphosed sediments for early differentiation of the Earth's mantle, Nature, 423: , 23. Corgne, A., C. Liebske, B. J. Wood, D. C. Rubie and D. J. Frost, Silicate perovskite-melt partitioning of trace elements and geochemical signature of a deep perovskitic reservoir, Geochim Cosmochim Acta, 69: , 25. Corgne A. and Wood B. J. Trace element partitioning betweenmajoritic garnet and silicate melt at 25 GPa. Phys. Earth Planet. Inter., , , /23/7

15 Chen, J. H. and G. J. Wasserburg, The least radiogenic Pb in iron meteorites. Fourteenth Lunar and Planetary Science Conference, Abstracts, Part I, Lunar Planet Sci. Inst., Houston, pp Chen, J. H. and G. J. Wasserburg The presence of 17 Pd in the early solar system. Lunar Planet. Sci. Conf. Absts. 21: Chen, J. H. and G. J. Wasserburg Live 17 Pd in the early solar system and implications for planetary evolution. In Earth Processes: Reading the Isotope Code, Vol. 95, S. R. Hart and A. Basu. ed., pp Washington: AGU. Halliday, A. N., H. Wanke, J.-L. Birck and R. N. Clayton, The accretion, composition, and early differentiation of Mars, Space Sci. Rev., 96: , 21. Harper, C. L., J. Volkening, K. G. Heumann, C.-Y. Shih and H. Wiesmann Hf- 182 W: New cosmochronometric constraints on terrestrial accretion, core formation, the astrophysical site of the r- process, and the origina of the solar system. Lunar Planet Sci. Conf Absts. 22: Harper, C. L. and S. B. Jacobsen, Evidence from coupled 147 Sm- 143 Nd and 146 Sm- 142 Nd systematics for very early (4.5-Gyr) differentiation of the Earth's mantle, Nature, 36: , Jacobsen, S. B. and C. L. Harper Accretion and early differentiation history of the Earth based on extinct radionuclides. In Earth Processes: Reading the Isotope Code, Vol. 95, S. R. Hart and A. Basu. ed., pp Washington: AGU. Jacobsen, S. and G. J. Wasserburg, 198, Sm-Nd isotopic evolution of chondrites, Earth Planet. Sci. Lett., 5, Kleine, T., C. Münker, K. Mezger and H. Palme, 22, Rapid accretion and early core formation on asteroids and the terrestrial planets from Hf-W chronometry, Nature, 418: Lee, D. C. and A. N. Halliday Hafnium-tungsten chronometry and the timing of terrestrial core formation. Nature. 378: Lee, D. C. and A. N. Halliday Hf-W evidence for early differentiation of Mars and the Eucrite parent body. Lunar Planet. Sci. Conf. Absts. 28: 79. Lee, T., D. A. Papanastassiou and G. J. Wasserburg, Demonstration of 26 Mg excess in Allende and evidence for 26 Al, Geophys. Res. Lett., 3: MacPherson, G. J., A. Davis and E. Zinner The distribution of aluminum-26 in the early Solar System-A reappraisal. Meteoritics. 3: Papanastassiou, D. A., and G. J. Wasserburg, Initial strontium isotopic abundances and the resolution of small time differences in the formation of planetary objects. Earth Planet. Sci. Lett., 5: Podosek, F. A., 197. Dating of meteorites by high temperature release of iodine correlated 129 Xe, Geochim. Cosmochim. Acta, 34: Podosek, F. and T. D. Swindle Extinct Radionuclides. in Meteorites and the Early Solar System, ed Tuscon: Univ. of Arizona Press. Shuloyukov, A., and G. W. Lugmair, Fe in eucrites, Earth Planet. Sci. Lett., 119: Swindle, T. D. and F. Podosek Iodine-Xenon Dating. in Meteorites and the Early Solar System, ed Tuscon: Univ. of Arizona Press. Tatsumoto, M., R. J. Knight, and C. J. Allègre, Time differences in the formation of meteorites ad determined from the ratio of lead-27 to lead-26, Science, 18: Yin, Q., S. B. Jacobsen, Y. K., J. Blichert-Toft, P. Télouk and F. Albarède, 22. A short timescale for terrestrial planet formation from Hf-W chronometry of meteorites, Nature, 418: /23/7