isotopic constraints on the formation of earth-like planetsisotopic constraints on the formation of...

Isotopic Constraints on the Formation of Earth-Like Planets

Alex N. HallidayDepartment of Earth Sciences, ETH Zentrum, NO, Sonneggstrasse 5, CH-8092, Zurich, Switzerland

Abstract. The terrestrial “rocky” planets of our solar system are depleted in major volatiles aswell as moderately volatile elements like potassium. Isotopic chronometry provides strong evidencethat accretion of the terrestrial planets took time-scales on the order of 107–108 years, consistentwith some dynamic simulations. Mars appears to have formed within the first 15 Myrs of the solarsystem. Growth of the Earth was relatively protracted, the Moon most probably forming >45 Myrsafter the start of the solar system, although the exact age is still poorly constrained. There is evidencethat significant compositional change occurred during the growth of the Earth-Moon system. Theisotopic compositions of Sr and W in lunar samples and Xe in the Earth provide evidence that theproto-planets that contributed to the growth of the Earth were more volatile rich than the present Earthor Moon. Time-integrated radioactive parent / radiogenic daughter elemental ratios Rb/Sr and Hf/Wcan be deduced from Sr and W isotope data respectively. These show that the materials that eventuallyformed the Moon had, on average, an order of magnitude higher volatile / refractory element budgetand were significantly more oxidizing. Similarly, Xe isotope data for the Earth shows that it grewfrom material with heavy noble gas budgets that were at least two orders of magnitude higher thantoday. Therefore, mechanisms that result in such changes, presumably as a result of accretion needto be incorporated into future dynamic models for the growth of the terrestrial planets. A majorunsolved question remains the origin of Earth’s water.

1. Introduction

The rocky, inner or terrestrial planets are distinct from the gaseous and icy giant,outer or jovian planets, as well as the currently detectable extrasolar planets, in theirmuch smaller size and higher uncompressed density. Just as we had no convincingevidence for extrasolar planets until the mid-90s (Mayor and Queloz, 1995), wecurrently lack any evidence for extrasolar terrestrial planets. As of the time ofMichel Mayor’s 60th birthday we still do not know if our solar system is unusual inthis respect. Resolving this issue and finding habitable worlds is a prime objectivein astrophysics and observational astronomy (Seager, 2003). When we reconvenefor Michel Mayor’s 80th there may well be a host of new kinds of such planets todiscuss. All our current understanding is theoretical or based on observations ofour own solar system. From this we can say that the formation of the terrestrialplanets was stochastic and strongly influenced by late stage events that resulted ina huge spectrum of surface environments. A good example is the Earth itself, theorigin of whose water currently is unclear, but may have been added at a late stageby chance bolides (Owen and Bar-Nun, 2000). Water has had a huge effect on theEarth, not just in its surface environments, but also in reducing the viscosity of themantle thereby facilitating efficient convection. The water content of the Earth hasa direct effect on the products and styles of volcanism and has ultimately led to

PlanetarySystems.tex; 28/12/2006; 12:00; p.69

60 A. N. Halliday

the long-term growth of the low density masses of rock we call continents (Taylor,1992). Terrestrial planets with a geological history and recent surface environmentslike that of the Earth probably are quite distinctive but could also turn out to beparticularly hard to replicate.

Terrestrial planets are of course distinct because they are relatively depleted inmajor volatile elements like hydrogen and helium. However, they are also relativelydepleted in moderately volatile elements like potassium that are still incompletelycondensed at ∼ 1,000K (Gast, 1960; Wasserburg et al., 1964; Cassen, 1996). Theshortage of these volatile constituents has come to be viewed as a consequence ofgrowth from a portion of the inner solar system that was already swept clear ofnebular gas and was dominated by dust that had been heated by T-Tauri radiationclose in to the Sun (e.g. Hayashi et al., 1985) or had been heated during collapseof the solar nebula (Boss, 1990). Therefore, the most widely accepted dynamicmodels for the growth of the terrestrial planets assume accretion after the nebulawas largely removed. The absence of nebular gas has a significant effect on the cal-culated time-scales for the growth of the terrestrial planets. Dynamic simulationsindicate that the growth of the terrestrial planets would then take 107–108 yearsin the absence of Jupiter (Safronov, 1954; Wetherill, 1986). In the presence of aminimum mass solar nebula this would shorten to a few times 106 years (Hayashiet al., 1985).

There now exists good evidence of solar noble gases in the Earth and also Mars(Bogard et al., 2001). For the Earth this includes He, Ne and Xe (Caffee et al., 1999;Pepin and Porcelli, 2002). A solar Ar component is not yet well resolved and littledata exist for Kr. Although others have been proposed (Trieloff et al., 2000), thesimplest and most efficient mechanism for incorporating solar noble gases into theEarth would be via in-gassing (Porcelli et al., 2001) – dissolving a small fractionof the constituents of a hot thick blanketing nebular proto-atmosphere into thesilicate and metallic liquids of a magma ocean (Hayashi et al., 1979; Mizuno et al.,1980). This is consistent with new dynamic simulations that indicate that at leastsome small amount of nebular gas would appear necessary to dampen eccentricitiesduring planetary growth (Agnor and Ward, 2002). If the Earth and Mars trappedsignificant nebular gases they must have accreted a fraction of their mass quicklyi.e. before dispersal of the solar nebula.

These models can be tested with isotopic studies of early solar system objectssuch as meteorites and lunar samples. There now exist sufficient data to showthat different isotopic methods do not always yield the same time-scales (Halliday,2003). This is most probably caused by differences in behaviour between the ele-ments during the history of accretion (Halliday, 2004). Therefore, it is anticipatedthat in future we will be able to go far beyond determining time-scales. We willincreasingly be able to utilize these chronometers to determine the physical andchemical environment in which the terrestrial planets grew.

All isotopic approaches make certain assumptions and in this chapter some ofthese assumptions are briefly explained. More detailed discussion can be found in


Isotopic Constraints 61

Halliday (2003, 2004). For example, the calculated time-scale of planetary accre-tion depends on the ratio of the radioactive parent to radiogenic daughter element.If this ratio were different in the early Earth or its precursor components becauseof different volatile budgets, the calculated time-scales for Earth accretion wouldbe incorrect. Some lines of evidence for such changes in volatile budgets are dis-cussed here. Equilibration during accretion is also an issue. The most widely usedapproaches are tungsten and lead isotopes. These methods assume that, as theplanet grows, accreting metal and silicate isotopically equilibrates (or mixes) com-pletely with the silicate portion of the growing planet. There is evidence that this isunlikely. The isotopic time-scales deduced would then be too short.

2. Isotopic Chronometry of Accretion

Isotope geochemistry has expanded dramatically in the past few years thanks totechnological advances and is now a very broad subject encompassing virtually theentire periodic table. Much of this expansion relates to mass dependent and massindependent isotopic fractionations. However, the development of new methods hasalso facilitated the study of new radiogenic isotope systems. A variety of isotopicchronometers have been particularly useful for deducing early solar system time-scales (Table I). These range from very short-lived nuclides like 26Al (T1/2 = 0.7Myrs) to almost stable isotopes like 238U with a half-life similar to the age of theEarth. The age of the solar system is 4.57 × 109 years. Therefore, a number ofearly solar system radionuclides in Table I are, by now, effectively extinct. Theywere present in the early solar system because of production in other stars, manyof them massive, shortly before (within a few half-lives of) the formation of thesun. Being extinct, the initial abundance can only be inferred. This is achieved byusing differences in the atomic abundance of the daughter isotope in early objects(meteorites) of known absolute age.

To utilize any of these nuclides for deducing early solar system time-scales it isnecessary to precisely measure the present day isotopic composition of the daugh-ter element. Therefore, the development of these chronometers is strongly linkedto breakthroughs in mass spectrometry. In the simplest system the present day iso-topic composition of the daughter element is a function of the initial abundance ofthe parent and daughter nuclide at the start of the solar system, the parent / daughterelement ratio, and how much time elapsed. If you like, the isotopic composition ofthe daughter represents a time-integrated parent / daughter elemental ratio. If anobject forms without a change in parent / daughter ratio, its time of formation willbe undetectable with isotopic chronometry. It is the large change in parent / daugh-ter ratio that accompanies certain processes associated with planetary accretion thatrenders isotopic systems useful. In fact it is this change that is being dated. If thereis no change that accompanies accretion there is no way to distinguish when anobject formed relative to the start of the solar system.


62 A. N. Halliday

Table I. Main isotopic decay systems in use in studying the origin and early evolution of theterrestrial planets. * Spontaneous fissionogenic product

Parent Daughter(s) Half-life (years) Applications

Long-lived

40K 40Ca, 40Ar 1.25 × 109 volcanic rocks, uplift ages, lunar bombardment87Rb 87Sr 4.88 × 1010 mantle and crust evolution, marine carbonates,

early solar system138La 138Ba, 138Ce 1.05 × 1011 very limited applications147Sm 143Nd 1.06 × 1011 mantle and crust evolution, early solar system176Lu 176Hf 3.57 × 1010 mantle and crust evolution, early solar system187Re 187Os 4.23 × 1010 mantle and crust evolution, early solar system232Th 208Pb 1.40 × 1010 limited applications235U 207Pb 7.04 × 108 mantle and crust evolution, early solar system238U 206Pb 4.47 × 109 mantle and crust evolution, early solar system

Extinct

22Na 22Ne 2.605 early solar system26Al 26Mg 7.3 × 105 early solar system53Mn 53Cr 3.7 × 106 early solar system60Fe 60Ni 1.5 × 106 early solar system92Nb 92Zr 3.6 × 107 early solar system107Pd 107Ag 6.5 × 106 early solar system129I 129Xe 1.57 × 107 early solar system, terrestrial degassing135Cs 135Ba 2.3 × 106 early solar system146Sm 142Nd 1.03 × 108 early solar system, crustal evolution182Hf 182W 9 × 106 early solar system, core formation205Pb 205Tl 1.5 × 107 early solar system244Pu 136Xe* 8.2 × 107 early solar system, terrestrial degassing

The age of an early rock or mineral can therefore be dated if the parent /daughter ratio is fractionated during formation. On this basis a number of earlysolar system objects have been dated and provide a framework within which toassess the time-scales for overall growth of the terrestrial planets. The earliestprecisely dated objects that are thought to have formed within our solar systemare the calcium aluminum refractory inclusions (CAIs) found in many chondrites(Table II). How these formed is unclear but they are thought to be recrystallizedrefractory condensates. Chondrules are also found in chondrites and very commonbut enigmatic round globules of what is thought to represent dust balls that weremelted. It would appear that at least some chondrules formed about 2 million years



Table II. Recent estimates of the ages of early solar system objects

Object Sample(s) Method Reference Age (Ga)

Earliest Solar System Allende CAIs U-Pb Gopel et al., 1991 4.566 ±0.002

Earliest Solar System Efremovka CAIs U-Pb Amelin et al.,2002

4.5672±0.0006

Chondrule formation Acfer chondrules U-Pb Amelin et al.,2002

4.5647±0.0006

Angrites Angra dos Reis &LEW 86010

U-Pb Lugmair andGaler, 1992

4.5578±0.0005

Early eucrites Chervony Kut Mn-Cr Lugmair and Shu-kulyokov, 1998

4.563±0.001

Mars accretion Minimum age Sm-Nd Harper et al., 1995 ≥ 4.54

Mars accretion Mean age Hf-W Lee et al., 1997 4.560

Mars accretion Minimum age Hf-W Lee et al., 1997 ≥ 4.54

Mars accretion Minimum age Hf-W Halliday et al.,2001

≥ 4.55

Mars accretion Minimum age Hf-W Kleine et al., 2002 ≥ 4.55

Earth accretion Mean age U-Pb Halliday, 2000,2004

≤ 4.55

Earth accretion Mean age U-Pb Halliday, 2000 ≥ 4.49

Earth accretion Mean age Hf-W Yin et al., 2002 ≥ 4.55

Lunar highlands Ferroan anorthosite60025

U-Pb Hanan & Tilton,1987

4.50±0.01

Lunar highlands Ferroan anorthosite60025

Sm-Nd Carlson & Lug-mair, 1988

4.44±0.02

Lunar highlands Norite from breccia15445

Sm-Nd Shih et al., 1993 4.46±0.07

Lunar highlands Ferroan noritic an-orthosite in breccia67016

Sm-Nd Alibert et al.,1994

4.56±0.07

Moon Best estimate of age U-Pb Tera et al., 1973 4.47±0.02

Moon Best estimate of age U-Pb,Sm-Nd

Carlson & Lug-mair, 1988

4.44-4.51

Moon Best estimate of age Hf-W Halliday et al.,1996

4.47±0.04

Moon Best estimate of age Hf-W Lee et al., 1997 4.51±0.01

Moon Maximum age Hf-W Halliday, 2000 ≤ 4.52

Moon Maximum age Rb-Sr Halliday and Por-celli, 2001

≤ 4.55

Moon Best estimate of age Hf-W Lee et al., 2002 4.51±0.01

Moon Best estimate of age Hf-W Kleine et al., 2002 4.54±0.01

Moon Best estimate of age Hf-W Halliday, 2004 < 4.52


64 A. N. Halliday

Figure 1. The current best estimates for the time-scales over which very early inner solar systemobjects and the terrestrial planets formed. The approximated mean life of accretion (τ) is the timetaken to achieve 63% growth at exponentially decreasing rates of growth. The dashed lines indicatethe mean lives for accretion deduced for Mars based on W isotopes (Lee et al., 1997) and the Earthbased on Pb isotopes (Halliday, 2000, 2003, 2004). Estimates for the mean life of accretion of theEarth and the age of the Moon that are based solely on W isotopes (Kleine et al., 2002; Yin et al.,2002) are shorter than the values given here. See Table II for details of other sources.

after the start of the solar system. How chondrules formed is unclear. The mostwidely accepted models involve dust in the solar nebula. Therefore, the isotopicdating provides evidence that the disk from which the planets grew was still adusty environment with rapid melting events, 2 million years after the start of thesolar system (Fig. 1). Differentiated asteroids and their planetary cores and silicatereservoirs, as represented by some iron and basaltic achondrite meteorites appearto have formed in the first 10 Myrs of the solar system (Table II). Exactly how earlyis unclear. The ages of meteorites provide a maximum constraint on how slowly theparent bodies formed.

For many larger planetary objects, the timescaleas are still harder to ascertain.The parent / daughter fractionation is more complex because it takes place as theplanet grows and this may take some significant period of time (>106 years). Foran object like the Earth there is no such thing as a precise age when all of the con-stituents were amalgamated. It is more practical to instead assume a style of growthand use the isotopic data to define a rate of growth of a planet from the integratedhistory of the chemical fractionation of the parent / daughter ratio provided by theisotopic composition of the daughter.

Such a major parent / daughter fractionation clearly affected the noble gases.Xenon isotopes provide powerful evidence that the noble gases were vastly moreabundant in the early earth and that a major fraction has been lost, probably by



shock-induced blow off of the atmosphere. The Xe constraints are based on thedecay of formerly live 129I (T1/2 = 16 Myrs) and 244Pu (fission T1/2 = 82 Myrs)(Table I). Put simply, the small difference in composition between the isotopiccomposition of xenon in the Earth relative to its initial starting composition isconsistent with low parent / daughter ratios. The amount of xenon in the Earthtoday is thought to be at least two orders of magnitude lower than is necessary tosatisfy the Xe isotope data (Porcelli et al., 2001). That is, the ratio of elements like Iand Pu to Xe is huge; at least two orders of magnitude higher. Therefore, there musthave been a major fractionation in parent / daughter ratio at a relatively late stage.Because Xe is a gas that has a strong preference for residing in the atmosphereand because similar effects are found in both the I-Xe and Pu-Xe system, it is clearthat there was a significant amount of xenon loss from the Earth, as opposed toselective I or Pu gain. The best estimate for the age of this loss is 50 to 80 Myrs(Fig. 1) (Ozima and Podosek, 1999; Porcelli and Pepin, 2000). It has been proposedthat this may have been associated with the Moon-forming giant impact. The Xeconstraints represent just one piece of isotopic evidence that accretion took 107–108

years and that there were significant changes in composition during planetesimaland planet growth in the terrestrial planet forming region.

The most powerful chronometers for determining the growth rates of the terres-trial planets are 182Hf–182W and 235,238U–207,206Pb (Halliday, 2003). In both casesthe parent / daughter ratio is strongly fractionated during core formation. All ofthe terrestrial planets have a metallic iron core formed by segregation of denseiron-rich liquids. Core formation separates the metal-loving elements (includingthe daughter elements W, Pb) from the silicate-loving elements (in this case theradioactive parents Hf, U) (Fig. 2). As such, the W and Pb isotopic compositions inthe silicate portion of the planet reflect how fast it grew and internally fractionatedby metal segregation. Although it was once thought that accretion was rapid (Hanksand Anderson 1969) and formation of the core was protracted (Solomon, 1979),there exists strong evidence that core formation is in fact early and rapid (Sasakiand Nakazawa, 1986), such that the core grows proportionally in response to thegrowth of the planet. On this basis the isotopic data yield constraints on the ratesof growth of the planet, provided certain assumptions are valid.

To determine accretion rates one needs to know the parent / daughter ratio in thetotal planet as well as in the reservoir, the formation of which is being dated. Thisis sometimes complex because we do not know independently how much residesin the inaccessible core. Some elements like Hf and W are refractory and were notaffected by heating in the inner portions of the disk. Therefore, the Hf/W ratio ofthe Earth is well known from independent measurements of chondrites – meteoritesrepresenting average accumulations of early solar system debris (Newsom, 1995).However, Pb is moderately volatile and therefore variably depleted in the Earth,Moon, Vesta, Mars and almost certainly in Venus and Mercury. Therefore, twoprocesses have produced depletions of Pb in the silicate Earth – core formation andvolatile loss. Two approaches are mainly relied upon to determine the amount of Pb


66 A. N. Halliday

Figure 2. The Hf/W ratio of the total Earth is chondritic (average solar system) because Hf and Ware both refractory elements. The U/Pb ratio of the Earth is enhanced relative to average solar systembecause approximately 80% of the Pb was lost by volatilization or incomplete condensation mainlyat an early stage of the development of the circumstellar disk. The fractionation within the Earth forHf/W and U/Pb is similar. In both cases the parent (Hf or U) prefers to reside in the silicate portionof the Earth. In both cases the daughter (W or Pb) prefers to reside in the core.

in the total Earth. The first is a comparison between the amount of Pb depletion inchondrites and that of other volatile elements that are not partitioned into planetarycores (Allegre et al., 1995). The second is based on estimates of the amount ofvolatile depletion expected given the average condensation temperature (Galer andGoldstein, 1996). These two approaches are in broad agreement for Pb.

It is also necessary to have a well-defined value for the isotopic compositionof the silicate Earth relative to the rest of the solar system. The W isotopic com-position of the silicate Earth is indeed extremely well defined because after 4.5billions of convective mixing the Earth’s silicate reservoir has homogenized anyearly isotopic variability. The W isotopic age of the Earth could in principle bedetermined from the composition of a tungsten carbide drill bit! However, withW the problem has been to get the correct value for the average solar system.Earliest Hf-W papers assumed a certain value (Jacobsen and Harper, 1996). Earlymeasurements of chondrites were variable (Lee and Halliday, 1996, 2000). Mostimportantly it is now recognized that the early W isotopic measurements for car-bonaceous chondrites reported in Lee and Halliday (1996) are incorrect and thetime-scales have had to be reassessed in the light of this (Kleine et al., 2002;Schoenberg et al., 2002; Yin et al., 2002; Halliday, 2004).

The problem for U-Pb is almost the opposite. The present day value for thesilicate Earth is very hard to constrain. The reason is that 235U and 238U are stillalive in the Earth and capable of producing isotopic variations in 207Pb/204Pb and



Figure 3. Estimates of the lead isotopic composition of the bulk silicate Earth (BSE) plotted relativeto the Geochron defined as the slope corresponding to the start of the solar system. All estimates plotto the right of this line, indicating protracted accretion and / or core formation. Data from Doe andZartman (1979), Davies (1984), Zartman and Haines (1988), Allegre et al. (1988), Allegre and Lewin(1989), Kwon et al. (1989), Liew et al. (1991), Galer and Goldstein (1991), Kramers and Tolstikhin(1997), Kamber and Collerson (1999) and Murphy et al. (2003).

206Pb/204Pb. Therefore, the Earth is very heterogeneous in its Pb isotopic compo-sition (Galer and Goldstein, 1996). This is shown in Fig. 3 in which 11 estimatesof the average Pb isotopic composition of the silicate Earth are presented. Eachof these would imply a different time-scale for earth accretion, as discussed belowand shown in Fig. 5.

The simplest kind of age calculation just deduces a time of fractionation ormodel age assuming the fractionating process was instantaneous. More complexmodels calculate an accretion rate (Jacobsen and Harper, 1996; Halliday, 2000,2004) by making assumptions about the style of growth. Generally speaking thisis an exponentially decreasing rate of growth to emulate what has arisen fromdynamic simulations. To quantify the accretion time-scale with an exponentiallydecreasing rate of growth requires the use of a time constant for accretion. Theinverse of this is the mean life, which corresponds to the time taken to achieve63% growth. Some of the most recent models (Halliday, 2004) also try and simulatethe effect of a more stochastic style of accretion with growth from sporadic large(Moon to Mars-sized) impactors (Fig. 4). The staggered growth model shown inFig. 4 assumes that the Earth grew from 90 to 99% of its current mass as a resultof a collision with an impacting Mars-sized planet called Theia that produced theMoon. The changes in the rate of accretion with time are calculated from the timingof this event. Therefore, the model time of the Giant Impact is directly related tothe mean life of accretion in these models (Fig. 5).


68 A. N. Halliday

Figure 4. Change in mass fraction of the Earth as a function of time in the model used in Halliday(2004) illustrated with an accretion scenario calculated from a giant impact at 50Myrs after the startof the solar system. The approximated mean life of accretion (τ) is the time taken to achieve 63%growth. Both this and the timing of each increment are calculated from the timing of the giant impact(tG I ) to achieve an overall exponentially decreasing rate of growth for the Earth broadly consistentwith dynamic simulations. The smooth curves show the corresponding exponentially decreasing ratesof growth. The step function curves define the growth used in the isotopic calculations. Growth ofthe Earth is modelled as a series of collisions between differentiated objects. The overall rate ofaccretionary growth of the Earth may have decreased in some predictable fashion with time, butthe growth events would have become more widely interspersed and larger. Therefore, the modelsimulates further growth by successive additions of 1% Earth mass objects from 1 to 10% of thecurrent mass, then by 2% objects to 30% and then by 4% objects to 90%. The Moon-forming giantimpact is modelled to take place when the Earth was ∼ 90% of its current mass and involved animpactor planet Theia that was ∼10% of the (then) mass of the Earth. Therefore, in the model thegiant impact contributes a further 9% of the current Earth mass. There is evidence against largeamounts of accretion after the giant impact.

3. Tungsten and Lead Isotope Differences and Disequilibrium Accretion

The rates of accretion of the Earth based on the model in Fig. 4 and the variousestimates of the bulk silicate Earth Pb isotopic composition (Fig. 3) are shown inFig. 5. Also shown are the results for W. It is clear that the kinds of time-scalesinvolved are on the order of 107 to 108 years, providing strong vindication ofthe Safronov-Wetherill style of protracted accretion via major impacts (Halliday,2003).

It is also immediately apparent that a discrepancy exists between the W resultsand every estimate based on Pb isotope data (Halliday, 2004). Either all of thesePb isotope estimates are incorrect or there is some basic difference in the waythese clocks operate. There appears to be better agreement between the Pb and Xeisotope constraints on the time-scales. There are a number of parameters that playinto these models that have a significant level of uncertainty associated.



Figure 5. Calculated values for the Earth’s mean life of accretion (τ) and time of the giant impact(TG I ) given in Myrs as deduced from the different estimates of the Pb isotopic composition of thebulk silicate earth (BSE) shown in Fig. 4 using the type of accretion model shown in Fig. 3 (seeHalliday, 2004). All calculations assume continuous core formation and total equilibration betweenaccreted material and the BSE. The 238U/204Pb values assumed for the Earth = 0.7 (Allegre et al.,1995). The Hf-W timescales using the same style of model are significantly shorter.

Currently, the most likely explanation for the apparent discrepancy is that thereis a difference in the level of equilibration of refractory W as opposed to moderatelyvolatile Pb as metal in the impactor mixes with the silicate Earth during accretion.In all these models it is assumed that growth takes place via collisions. The incom-ing planetesimals and planets are likely to be differentiated into silicate and metallike Vesta, Mars and the Moon-forming planet Theia (Taylor and Norman, 1990).Many isotopic models assume that all of the metal and silicate in the impactormixes and equilibrates with the silicate portion of the Earth before further growthof the Earth’s core (Kleine et al., 2002; Schoenberg et al., 2002; Yin et al., 2002).In reality the situation could have been very different however (Yoshino et al.,2003). The likelihood of metal from the impactor mixing and equilibrating withthe silicate portions of the Earth depends on whether the metal in the incomingplanetary core becomes fragmented into small droplets (Rubie et al., 2003). Thefact that the silicate Earth has a W isotopic composition that is so similar to averagesolar system or chondrites means that in general terms equilibration must indeedbe the norm. However, the small difference that is now known to exist could re-flect an event like the giant impact that involved particularly rapid amalgamationof large fragments of impactor metal with the Earth’s core with little chance forequilibration with the silicate portion of the Earth (Halliday, 2004).


70 A. N. Halliday

4. Loss of Moderately Volatile Elements During Accretion

The depletion of the terrestrial planets in moderately volatile elements has beenrecognized for about 40 years. This was long considered a result of incompletecondensation because it was believed that the inner solar system formed from a hotsolar nebular gas. Now it is understood that accretion of most of the material in theplanets came from dust inherited from a presolar molecular cloud.

Oxygen isotopes provide a monitor of the degree of mixing of this presolarmaterial (Clayton and Mayeda, 1975; Wiechert et al., 2001). The oxygen isotopiccomposition varies among different classes of meteorites and is therefore assumedto be highly heterogeneous in general among inner solar system objects (Clayton etal., 1973; Clayton, 1986, 1993). This is not a normal mass dependent fractionationso much as a mass independent effect, the origin of which is poorly understoodor is at least controversial at the present time. Regardless of the reason behind theheterogeneity it is thought that the oxygen isotopic composition provides a monitorof the origins of the material in a planetary object. The provenance of material in anobject the size of the Earth is expected to have been quite broad (Wetherill, 1994).As such, the oxygen isotopes merely provide some average. The Earth is distinctfrom all classes of chondrites except enstatite chondrites, which may have formedin the more reducing inner regions of the solar system. The provenance should bemuch broader than this. Similarly, the provenance appears to be distinct from thatof the material forming the Asteroid 4 Vesta, or Mars, as judged from studies ofmeteorites thought to be derived from these objects (Clayton, 1986, 1993). TheMoon however, has an oxygen isotopic composition that is identical to that of theEarth (Clayton and Mayeda, 1975), recently demonstrated to persist to extremelyhigh precision (Wiechert et al., 2001). This is a striking finding that provides goodevidence that the Earth and Moon were formed from material of similar provenanceand presumably similar composition (Wiechert et al., 2001).

The Moon is widely considered to be the product of a collision between theproto-Earth and a Mars-sized planet sometimes referred to as “Theia” (Cameronand Benz, 1991; Canup and Asphaug, 2001). (Theia was the mother of Selene,the goddess of the Moon.) Roughly 80% of the Moon forms from the debris de-rived from Theia (Canup and Asphaug, 2001). If this is correct, the oxygen iso-topic evidence provides evidence that Theia and the Earth had similar averageprovenance.

Yet the Moon is far more depleted in volatile elements than the Earth (Tay-lor and Norman, 1990). A likely explanation for this is that there were majorreductions in volatile constituents during the Moon-forming giant impact (O’Neill,1991). Support for this is found in the initial 87Sr/86Sr of early lunar rocks fromthe Highlands (Halliday and Porcelli, 2001). These yield an initial 87Sr/86Sr for theMoon that is clearly resolvable from the initial composition of the solar system,consistent with radiogenic growth in an environment with relatively high Rb/Sr.This time-integrated Rb/Sr can be determined to be ∼0.07, similar to the Rb/Sr



Figure 6. Hf/W appears to be negatively correlated with Rb/Sr in the primitive mantles of innersolar system planetesimals and planets. A possible explanation for this is that the loss of moderatelyvolatile elements was linked to loss of other volatiles during planetary collisions such that the mantleschanged from more oxidizing to more reducing. The Moon is an extreme example of this. The factthat the mixture of material from the proto-Earth and Theia that is calculated to have formed theMoon is so like Mars provides evidence that such volatile-rich objects may have been common in theinner solar system during the early stages of planetary accretion. See Halliday and Porcelli (2001)and Halliday (2004) for further details.

of Mars but an order of magnitude less depleted in moderately volatile Rb thanthe Moon today (Rb/Sr∼0.006). Therefore, it appears as though Mars-like objectsonce existed in the vicinity of the Earth.

Tungsten isotopes paint a strikingly similar picture (Halliday, 2004). The Hf/Wratio of silicate reservoirs in the terrestrial planets is variable and is thought todepend on the level of oxidation of the mantle because this controls how “metalloving” W is during core formation. If the mantle is relatively reducing the vastmajority of the W is segregated into the core and vice versa. There is a reasonablyclear relationship between the degree of depletion in moderately volatile elements(like Rb) and the degree of depletion of W in the silicate portions or primitivemantles of different planets and planetesimals (Halliday, 2004). Just as the time-integrated Rb/Sr can be deduced from Sr isotopes, the time integrated Hf/W can bededuced from W isotopes. Tungsten isotope data for the Moon provide evidencethat the precursor objects that provided the raw materials for the Moon had a time-


72 A. N. Halliday

averaged Rb/Sr and Hf/W that was within error identical to the actual Rb/Sr andHf/W of Mars (Fig. 6).

Therefore, the lunar data provide evidence of planetesimals and planets in thevicinity of the Earth that were more like Mars in terms of chemical composition.The fact that the oxygen isotopic composition of the Earth and Moon are identicalprovides evidence that such proto-planetary compositions may well have been anormal feature of inner solar system or terrestrial proto-planetary compositions.This however, assumes that oxygen isotopes provide a comparative monitor of themix of raw ingredients in the inner solar system. The isotopic evidence for impact-driven chemical changes in the terrestrial planets during growth is however wellsupported by evidence that the planet Mercury lost a major fraction of its silicateby impact erosion (Benz et al., 1987).

Acknowledgements

This research is supported by ETH and SNF. The paper benefited from very fruitfuldiscussions with Willy Benz and Don Porcelli. I’d like to thank Ruedi von Steigerfor his patience and Michel Mayor for letting geochemists into Switzerland todabble in planets.

References

Agnor, C. B. and Ward W. R.: 2002, ‘Damping of terrestrial-planet eccentricities by density-waveinteractions with a remnant gas disk’, Astrophys. J. 567, 579–586.

Alibert, C., Norman M. D. and McCulloch M. T.: 1994, ‘An ancient Sm-Nd age for a ferroan noriticanorthosite clast from lunar breccia 67016’, Geochim. Cosmochim. Acta 58, 2921–2926.

Allegre, C. J., Manhes G. and Gopel C.: 1995, ‘The age of the Earth’, Geochim. Cosmochim. Acta59, 1445–1456.

Allegre, C. J. & Lewin, E.: 1989, ‘Chemical structure and history of the Earth: Evidence from globalnon-linear inversion of isotopic data in a three box model’, Earth Planet. Sci. Lett. 96, 61–88.

Allegre, C. J., Lewin, E. & Dupre, B.: 1988, ‘A coherent crust-mantle model for the uranium-thorium-lead isotopic system’, Chem. Geol. 70, 211–234.

Amelin, Y., Krot A. N., Hutcheon I. D. and Ulyanov A. A.: 2002, ‘Lead isotopic ages of chondrulesand calcium-aluminum-rich inclusions’, Science 297, 1678–1683.

Benz, W., Cameron A. G. W. and Slattery W. L.: 1987, ‘Collisional stripping of Mercury’s mantle’,Icarus 74, 516–528.

Bogard, DD, Clayton RN, Marti K, Owen T, Turner G: 2001, ‘Martian volatiles: Isotopic composi-tion, origin, and evolution’, Space Sci. Rev. 96, 425–458.

Boss, A. P.: 1990, ‘3D Solar nebula models: implications for Earth origin’. In: Origin of the Earth,H. E. Newsom and J. H. Jones, eds., pp. 3–15, Oxford University Press, Oxford.

Caffee, M. W., Hudson G. B., Velsko C., Huss G. R., Alexander E. C. Jr. and Chivas A. R.: 1999,‘Primordial noble gases from Earth’s mantle: Identification of a primitive volatile component’,Science 248, 2115–2118.

Cameron, A. G. W. and Benz W.: 1991, ‘Origin of the Moon and the single impact hypothesis IV’,Icarus 92, 204–216.



Canup, R. M. and Asphaug E.: 2001, ‘Origin of the moon in a giant impact near the end of the Earth’sformation’, Nature 412, 708–712.

Carlson, R.W. and Lugmair G. W.: 1988, ‘The age of ferroan anorthosite 60025: oldest crust on ayoung Moon?’ Earth Planet. Sci. Lett. 90, 119–130.

Cassen, P.: 1996, ‘Models for the fractionation of moderately volatile elements in the solar nebula’,Meteoritics Planet. Sci. 31, 793–806.

Clayton, R. N.: 1986, ‘High temperature isotope effects in the early solar system’. In Stable Isotopesin High Temperature Geological Processes, eds. J. W. Valley, H. P. Taylor and J. R. O’Neil(Washington D.C.: Mineral. Soc. Am.), pp. 129–140.

Clayton, R. N.: 1993, ‘Oxygen isotopes in meteorites’, Ann. Rev. Earth Planet. Sci. 21, 115-149.Clayton, R. N. and Mayeda T. K.: 1975, ‘Genetic relations between the moon and meteorites’, Proc

Lunar Sci Conf XI, 1761–1769.Clayton, R. N., Grossman, L. and Mayeda T. K.: 1973, ‘A component of primitive nuclear

composition in carbonaceous meteorites’, Science 182, 485–487.Galer, S. J. G. and Goldstein S. L.: 1996, ‘Influence of accretion on lead in the Earth’. In: Isotopic

studies of crust-mantle evolution, A.R. Basu and S.R. Hart, eds., 75-98, AGU.Galer, S. J. G. and Goldstein, S. L.: 1991, ‘Depleted mantle Pb isotopic evolution using conformable

ore leads’, Terra Abstr. 3, 485–486.Gast, P.W.: 1960, ‘Limitations on the composition of the upper mantle’, J. Geophys. Res. 65 1287–

1297.Gopel, C., Manhes G. and Allegre C. J.: 1991, ‘Constraints on the time of accretion and thermal

evolution of chondrite parent bodies by precise U-Pb dating of phosphates’, Meteoritics 26, 73.Halliday, A. N.: 2000, ‘Terrestrial accretion rates and the origin of the Moon’, Earth Planet. Sci. Lett.

176, 17–30.Halliday, A. N. and Porcelli D.: 2001, ‘In search of lost planets – the paleocosmochemistry of the

inner solar system’, Earth Planet. Sci. Lett. 192, 545–559.Halliday, A. N., Rehkamper M., Lee D-C. and Yi W.: 1996, ‘Early evolution of the Earth and Moon:

new constraints from Hf-W isotope geochemistry’, Earth Planet. Sci. Lett. 142, 75–89.Halliday, A. N., Wanke H., Birck J-L. and Clayton R. N.: 2001, ‘The accretion, bulk composition

and early differentiation of Mars’, Space Sci. Rev. 96, 197–230.Halliday, A.N.: 2003, ‘The origin and earliest history of the Earth’, pp 509–557. In: Meteorites,

comets and planets (ed. A.M. Davis) Vol. 1 Treatise on Geochemistry (eds. H.D. Holland andK.K. Turekian), Elsevier-Pergamon, Oxford.

Halliday, A.N.: 2004, ‘Mixing, volatile loss and compositional change during impact-drivenaccretion of the Earth’, Nature in press

Hanan, B.B. and Tilton G.R.: 1987, ‘60025: Relict of primitive lunar crust?’ Earth Planet. Sci. Lett.84, 15–21.

Hanks, T. C. and Anderson D. L.: 1969, ‘The early thermal history of the earth’, Phys. Earth Planet.Int. 2, 19–29.

Harper, C. L., Nyquist L. E., Bansal B., Wiesmann H. and Shih C.-Y.: 1995, ‘Rapid accretion andearly differentiation of Mars indicated by 142Nd/144Nd in SNC meteorites’, Science 267, 213–217.

Hayashi, C., Nakazawa K and Nakagawa Y.: 1985, ‘Formation of the solar system’. In Protostarsand Planets II, ed. D. C. Black and M. S. Matthews, 1100–1153 (Tucson: University of ArizonaPress).

Hayashi, C., Nakazawa K. and Mizuno H.: 1979, ‘Earth’s melting due to the blanketing effect of theprimordial dense atmosphere’, Earth Planet. Sci. Lett. 43, 22–28.

Jacobsen, S. B. and Harper C. L. Jr.: 1996, ‘Accretion and early differentiation history of the Earthbased on extinct radionuclides’. In Earth Processes: Reading the Isotope Code, eds. A. Basu andS. Hart. Washington D. C.: AGU, pp. 47–74.


74 A. N. Halliday

Kamber, B. S. & Collerson, K. D.: 1999, ‘Origin of ocean-island basalts: a new model based on leadand helium isotope systematics’, J. Geophys. Res. 104, 25479–25491.

Kleine, T., Munker, C., Mezger, K., Palme, H.: 2002, ‘Rapid accretion and early core formation onasteroids and the terrestrial planets from Hf-W chronometry’, Nature 418, 952–955

Kramers, J. D. and Tolstikhin I. N.: 1997, ‘Two terrestrial lead isotope paradoxes, forward transportmodeling, core formation and the history of the continental crust’, Chem. Geol. 139, 75–110.

Kwon, S.-T., Tilton, G. R. & Grunenfelder, M. H.: 1989, in:Carbonatites—Genesis and Evolution,ed. Bell, K.. Unwin-Hyman, London, 360–387.

Lee, D.-C. and Halliday A. N.: 1996, Hf-W isotopic evidence for rapid accretion and differentiationin the early solar system’, Science 274, 1876–1879.

Lee, D.-C. and Halliday A. N.: 1997, ‘Core formation on Mars and differentiated asteroids’, Nature388, 854–857.

Lee, D-C., Halliday A. N., Leya I., Wieler R. and Wiechert U.: 2002, ‘ Cosmogenic tungsten and theorigin and earliest differentiation of the Moon’, Earth Plan. Sci. Lett. 198, 267–274.

Lee, D-C. and Halliday, A.N.: 2000, ‘Accretion of primitive planetesimals: Hf-W isotopic evidencefrom enstatite chondrites’, Science 288, 1629-1631.

Liew, T. C., Milisenda, C. C. & Hofmann, A. W.: 1991, ‘Isotopic constrasts, chronology of elementtransfers and high-grade metamorphism: the Sri Lanka Highland granulites, and the Lewisian(Scotland) and Nuk (S.W. Greenland) gneisses’, Geol. Rundsch. 80, 279–288.

Lugmair, G. W. and Galer S. J. G.: 1992, ‘Age and isotopic relationships between the angrites LewisCliff 86010 and Angra dos Reis’, Geochim. Cosmochim. Acta 56, 1673–1694.

Lugmair, G. W. and Shukolyukov A.: 1998, ‘Early solar system timescales according to 53Mn-53Crsystematics’, Geochim. Cosmochim. Acta 62, 2863–2886.

Mayor, M. and Queloz, D.: 1995, ‘A jupiter-mass companion to a Solar-Type star’, Nature 378,355–359.

Mizuno, H., Nakazawa K. and Hayashi C.: 1980, ‘Dissolution of the primordial rare gases into themolten Earth’s material’, Earth Planet. Sci. Lett. 50, 202–210.

Murphy, D. T., Kamber, B. S. & Collerson, K. D.: 2003, ‘A refined solution to the first terrestrialPb-isotope paradox’, J. Petrol. 44, 39–53.

Newsom, H. E.: 1995, ‘Composition of the Solar System, Planets, Meteorites, and Major TerrestrialReservoirs’. In Global Earth Physics, A Handbook of Physical Constants, AGU Reference Shelf1, American Geophysical Union.

O’Neill, H. St. C.: 1991, ‘The origin of the Moon and the early history of the Earth - a chemicalmodel; Part I: The Moon’, Geochim. Cosmochim. Acta 55, 1135–1158.

Owen, T. and Bar-Nun A.: 2000, ‘Volatile contributions from icy planetesimals’. In: Origin of theEarth and Moon. R. M. Canup and K. Righter, eds. Univ. of Arizona Press, Tucson, 459–471.

Ozima, M. and Podosek, F. A.: 1999, ‘Formation age of Earth from 129I/127I and 244Pu/238Usystematics and the missing Xe’, J. Geophys. Res. 104, B11, 25493–25499.

Pepin, R. O. and Porcelli, D.: 2002, ‘Origin of noble gases in the terrestrial planets’, In: Noble gasesin geochemistry and cosmochemistry (eds. D. Porcelli, C.J. Ballentine and R. Wieler) Reviews inMineralogy and Geochemistry 47, 191–246.

Porcelli, D. and Pepin R. O.: 2000, ‘Rare gas constraints on early earth history’. In: Origin of theEarth and Moon. R.M.Canup and K.Righter, eds., 435–458. Univ.of Arizona Press, Tucson.

Porcelli, D., Cassen P. and Woolum D.: 2001, ‘Deep Earth rare gases: Initial inventories, capturefrom the solar nebula and losses during Moon formation’, Earth Planet. Sci. Lett 193, 237–251.

Rubie, D. C., Melosh H. J., Reid J. E., Liebske C. and Righter K.: 2003, ‘Mechanisms of metal-silicate equilibration in the terrestrial magma ocean’, Earth Planet. Sci Lett., 205, 239–255.

Safronov, V.S.: 1954, ‘On the growth of planets in the protoplanetary cloud’, Astron. Zh. 31, 499–510.Sasaki, S. and Nakazawa K.: 1986, ‘Metal-silicate fractionation in the growing Earth: energy source

for the terrestrial magma ocean’, J. Geophys. Res. 91, B9231–9238.



Schonberg, R., Kamber B. S., Collerson K. D. and Eugster O.: 2002, ‘New W isotope evidencefor rapid terrestrial accretion and very early core formation’, Geochim. Cosmochim. Acta 66,3151–3160.

Seager, S.: 2003, ‘The search for Earth-like extrasolar planets’, Earth Planet. Sci. Lett. 208, 113–124.Shih, C.-Y., Nyquist L. E., Dasch E. J., Bogard D. D., Bansal B. M. and Wiesmann H.: 1993, ‘Age

of pristine noritic clasts from lunar breccias 15445 and 15455’, Geochim. Cosmochim. Acta 57,915–931.

Solomon, S. C.: 1979, ‘Formation, history and energetics of cores in the terrestrial planets’, EarthPlanet. Sci. Lett. 19, 168–182.

Taylor, S. R.: 1992, Solar System Evolution A New Perspective, ed. S. R. Taylor. New York,Cambridge Univ. Press.

Taylor, S. R. and Norman M. D.: 1990, ‘Accretion of differentiated planetesimals to the Earth’, in:Origin of the Earth, H. E. Newsom and J. H. Jones, eds. Oxford University Press, Oxford, pp.29–43.

Tera, F., Papanastassiou D.A. and Wasserburg G.J.: 1973, ‘A lunar cataclysm at ∼3.95 AE and thestructure of the lunar crust’, Lun. Planet. Sci. IV, 723–725.

Trieloff, M., Kunz J., Clague D. A., Harrison D. and Allegre C.J.: 2000, ‘The nature of pristine noblegases in mantle plumes’, Science 288, 1036–1038.

Wasserburg, G.J., MacDonald F., Hoyle F. and Fowler W. A.: 1964, ‘Relative contributions ofuranium, thorium, and potassium to heat production in the Earth’, Science 143 465–467.

Wetherill, G. W.: 1986: ‘Accmulation of the terrestrial planets and implications concerning lunarorigin’, in Origin of the Moon, eds W. K. Hartmann, R. J. Phillips and G. J. Taylor. Houston,Lunar Planetary Institute, pp. 519–550.

Wetherill, G. W.: 1994, ‘Provenance of the terrestrial planets’, Geochim. Cosmochim. Acta 58, 4513–4520.

Wiechert, U., Halliday A. N., Lee D.-C., Snyder G. A., Taylor L. A. and Rumble D. A.: 2001,‘Oxygen isotopes and the Moon-forming giant impact’, Science 294, 345–348.

Yin, Q. Z., Jacobsen S. B., Yamashita K., Blicher-Toft J., Telouk P. and Albarede F.: 2002, ‘A shorttime scale for terrestrial planet formation from Hf-W chronometry of meteorites’, Nature 418,949–952.

Yoshino, T., Walter M. J. and Katsura T.: 2003, ‘Core formation in planetesimals triggered bypermeable flow’, Nature 422, 154–157.

Zartman, R. E. and Haines, S. M.: 1988, ‘The plumbotectonic model for Pb isotopic systematicsamong major terrestrial reservoirs—a case for bi-directional transport’, Geochim. Cosmochim.Acta 52, 1327–1339.


isotopic constraints on the formation of earth-like planetsisotopic constraints on the formation of...

Documents