1 
 

In Search of Late-Stage Planetary  
Building Blocks 

 
 
 
 
 
 
 

Richard J. Walker, Katherine Bermingham, Jingao Liu*, Igor S. Puchtel,  
Mathieu Touboul** and Emily A. Worsham 

 
Corresponding Author – Richard J. Walker (rjwalker@umd.edu) 

 
 

Isotope Geochemistry Laboratory 
Department of Geology 
University of Maryland 

College Park, MD 20742 USA 
 

Present Address:  
 

*1-26 Earth Sciences Building, Department of Earth and Atmospheric Sciences 
University of Alberta, Edmonton, Alberta, Canada T6G 2E3 

 
**Laboratoire de Geologie de Lyon: Terre, Planetes et environnement (LGLTPE) 

UMR CNRS 5276 (CNRS, ENS, Université Lyon1) 
Ecole Normale Supérieure de Lyon, 69364 Lyon Cedex 07, France 

 
 
 
 

11,075 Words (Abstract + Text) 
 
 
 
 

Submitted to: Chemical Geology 
February 13, 2015  



2 
 

 
Abstract 1 

Genetic contributions to the final stages of planetary growth, including materials associated 2 

with the giant Moon-forming impact, late accretion, and late heavy bombardment are examined 3 

using siderophile elements. Isotopic similarities between the Earth, Moon and enstatite 4 

chondrites for both lithophile and siderophile elements collectively lead to the suggestion that the 5 

genetics of the building blocks for Earth and the impactor involved in the Moon-forming event 6 

were broadly similar. The bulk genetic fingerprint of materials added to Earth by late accretion, 7 

defined as the addition of ~0.5 wt. % of mass to the silicate Earth following cessation of core 8 

formation, is characterized by 187Os/188Os and Pd/Ir that are similar to those in some enstatite 9 

chondrites. However, the integrated fingerprint of late accretion differs from enstatite chondrites 10 

in terms of the relative abundances of certain other HSE, most notably Ru/Ir. A minor, final 11 

~0.05% addition of material to the Earth and Moon, believed by some to be part of a late heavy 12 

bombardment, included a genetically distinct component with much more fractionated relative 13 

HSE abundances than evidenced in the average late accretionary component.   14 

Heterogeneous 182W isotopic data for ancient terrestrial rocks suggest that some very early-15 

formed terrestrial mantle domains remained chemically distinct for long periods of time 16 

following primary planetary accretion. This evidence for sluggish mixing of the early mantle 17 

suggests that if late accretionary contributions to the mantle were genetically diverse, it may be 18 

possible to identify the disparate primordial components in the terrestrial rock record using 19 

siderophile element isotopic tracers, such as Ru and Mo. 20 

 21 

Keywords: building blocks, giant impact, late accretion, late heavy bombardment, siderophile 22 

elements  23 



3 
 

1. Introduction 24 

The origins of the rocky planets, especially with regard to assembly processes and the 25 

chemical nature of their building blocks, have been the topic of intense interest and debate for 26 

decades. It is now generally agreed that the terrestrial planets were dominantly built through a 27 

series of energetic collisions of bodies of increasingly greater mass, termed oligarchic growth 28 

(e.g., Kokuba and Ida, 1998; Raymond et al., 2006). In the case of Earth, this process may have 29 

culminated in a final giant impact, involving an impactor comprising 5% or more of the mass of 30 

the present Earth, and leading to the formation of the Moon (Hartmann and Davis, 1975; Canup 31 

and Asphaug, 2001; Cuk and Stewart, 2012; Canup, 2012).  32 

In addition to constraining the dynamical processes involved in the construction of the 33 

rocky planets, it is equally important to assess the origins and nature of the materials from which 34 

they were built. Comparisons of the general compositions of the terrestrial planets have 35 

commonly been made on the basis of chemical models developed for these planetary bodies, 36 

most notably the Earth, Moon, and Mars; that is, bodies from which we are reasonably confident 37 

we have samples. Such models often require the assumption that the bodies were constructed 38 

from a combination of materials that were compositionally similar to primitive meteorites 39 

present in our collections (Wänke and Dreibus, 1988; 1994; McDonough and Sun, 1995; Wänke, 40 

2001; Taylor et al., 2006). The bulk planetary concentrations of a number of poorly-constrained 41 

elements are then estimated by applying bootstrapping methods that assume that the ratios of 42 

these elements to relatively well-constrained, geochemically-comparable major elements, are 43 

similar to the ratios observed in primitive meteorites (e.g., McDonough and Sun, 1995). Models 44 

of the chemical composition of inaccessible planetary reservoirs, such as Earth’s core, 45 

necessarily require these types of general assumptions (McDonough, 2003). 46 



4 
 

The chemical and genetic makeup of the planets can potentially be further constrained by 47 

isotopic comparisons to one another and to primitive meteorites. For example, planetary 48 

materials exhibit a large range in mass independent variations in 17O (the per mil deviation in 49 

17O/16O from the terrestrial fractionation line), which can be used as genetic fingerprints of 50 

precursor materials. It has been hypothesized that the heterogeneities in 17O originated as a 51 

result of self-shielding effects in the photo-dissociation of CO by exposure to ultraviolet light 52 

within the solar nebula (e.g., Thiemens and Heidenreich, 1983; Clayton, 2002; Lyons and 53 

Young, 2005). Variations in 17O among differentiated bodies have, therefore, commonly been 54 

interpreted to reflect the formation of precursor materials at greater or lesser distances from the 55 

Sun, possibly coupled with time of formation (e.g., Yurimoto and Kuramoto, 2004). As an 56 

example of the application of O isotopes to issues of genetics, the similarity in the 17O of the 57 

Earth and enstatite chondrites has commonly been interpreted to mean that the major building 58 

blocks of the Earth formed in a region of the protoplanetary disk similar to where enstatite 59 

chondrites formed (Clayton et al., 1984; Javoy et al., 2010). Conversely, differences in the 17O 60 

compositions of the Earth and Mars have been cited as evidence that the building blocks of these 61 

two bodies differed substantially (Franchi et al., 1999). Although there is no perfect fit of all 62 

physical parameters between any types (or likely any combination of types) of primitive 63 

meteorites and the Earth, Moon or Mars, constraining the general categories of accretionary 64 

materials, nevertheless, remains an important objective of cosmochemistry. 65 

Here, we focus mainly on the final ~10 to ~0.05% of Earth’s accretion. Late stages of 66 

major terrestrial planetary accretion may have included the participation of materials that formed 67 

in different portions of the protoplanetary disk, including water- and organic-rich materials 68 

(Weidenschilling et al., 1997; Chambers, 2001). Thus, although limited in mass, late stage 69 



5 
 

planetary growth may have had a disproportionate effect on the volatile contents of the rocky 70 

planets (e.g., Albarede et al., 2013). Further, late stage additions may have carried genetically 71 

distinct elemental and isotopic fingerprints. Because of the comparatively limited mass 72 

contributed by these processes, elemental and isotopic tracers comprising major elements, such 73 

as O, are of limited value in constraining the nature of these final building blocks. Thus, we will 74 

instead explore the possibility of tracing the late-stages of planetary growth using insights gained 75 

from elemental and isotopic variability of so-called siderophile, or Fe-loving, elements.  76 

In this overview, the elemental and isotopic fingerprints of late stage building blocks that 77 

may be recorded in mantle rocks from the Earth, as well as mantle-derived and impact generated 78 

rocks from Mars and the Moon, respectively, will be examined. In addition to considering the 79 

average elemental and isotopic characteristics of siderophile elements contained in the silicate 80 

portions of these bodies, we will also explore the possibility that the signals of individual 81 

building blocks might be identified through small differences in the isotopic compositions of the 82 

siderophile elements Ru and Mo, which varied among early solar system materials as a result of 83 

their incorporating differing proportions of diverse nucleosynthetic components. The basis for 84 

this optimism is the discovery that primordial mantle heterogeneities, recorded by lithophile, 85 

atmophile and siderophile short-lived radiogenic isotope systems, survived long enough to be 86 

preserved in the terrestrial rock record (Caro et al., 2003; Willbold et al., 2011; Mukhopadhyay 87 

et al., 2012; Touboul et al., 2012; 2014). If the interpretations of long-lived chemical/isotopic 88 

heterogeneity in the mantle presented by these studies are correct, isotopically distinct domains 89 

within the mantle, imparted during late stage accretion of genetically distinct materials, might 90 

also be preserved in the rock record. 91 

 92 



6 
 

2.  Overview of Siderophile Elements 93 

Siderophile elements are those elements that strongly partition into metallic Fe relative to 94 

silicate melt, and are consequently concentrated, to greater or lesser extents, in the cores of the 95 

rocky planets (Goldschmidt, 1937). Because of this, their concentrations in silicate mantles and 96 

crusts are low compared to primitive meteorites, the compositions of which are presumed to be 97 

representative of the majority of the planetesimals involved in the final stages of rocky planet 98 

accretion (Anders and Grevesse, 1989). Siderophile trace elements are commonly divided into 99 

sub-groups based on the intensity of their siderophilic tendencies under the typical 1 atmosphere 100 

experimental conditions initially employed to characterize the nature of metal-silicate 101 

partitioning of these elements (e.g., Kimura et al., 1974; Borisov et al., 1994). The moderately 102 

siderophile elements (MSE), including Co, W, Ni, Ge, and Mo, are characterized by metal-103 

silicate D values (concentration ratio of an element in liquid metal to liquid silicate) ranging 104 

from about 10 to 1000. The highly siderophile elements (HSE), including Re, Os, Ir, Ru, Pt, Rh, 105 

Pd, and Au, are characterized by D values of greater than 10,000. 106 

One important characteristic of siderophile elements is that the intensity of their 107 

siderophilic behavior can shift considerably at increasingly higher temperatures and pressures 108 

(e.g., Ringwood, 1966; Murthy, 1991; Li and Agee, 1996; Holzheid et al., 2000). The general 109 

tendency of most, but not all, siderophile elements is towards lower D values, as pressure and 110 

temperature conditions increase (Righter and Drake, 1997; Mann et al., 2012). Their partitioning 111 

characteristics are also affected by other intensive parameters of a given planetary body, such as 112 

oxygen fugacity (Cottrell and Walker, 2006; Wade and Wood, 2005). These shifts in partitioning 113 

behavior are important to recognize when considering issues of planetary growth and core 114 

formation. Depending upon the conditions where metal last equilibrated with silicates during 115 



7 
 

progressive core formation in a growing body, appropriate D values likely changed considerably 116 

during growth of sizable planetary bodies, thus, affecting the final absolute and relative 117 

concentrations of the siderophile elements in these mantles (Wade and Wood, 2005).  118 

For the purposes of genetic tracing of late stage building blocks, it is also important to 119 

recognize that the processes leading to the present abundances of the MSE contained within the 120 

silicate portions of the rocky planets may not have been the same as for the HSE. The chondrite-121 

normalized abundances of the MSE estimated for the bulk silicate Earth (BSE) vary considerably 122 

(Fig. 1). Experimental studies have shown that the abundances of the MSE can be accounted for 123 

if metal-silicate equilibration occurred at elevated temperatures and pressures (e.g., Hillgren et 124 

al., 1994; Righter et al., 1997; Li and Agee, 2001). Thus, numerous studies have concluded that 125 

the abundances of these elements are consistent with metal-silicate equilibration at an average 126 

depth equivalent to pressures of 20-60 GPa. A major shift in the O fugacity of the terrestrial 127 

mantle, resulting from the disproportionation of ferrous iron into ferric iron plus metal, as occurs 128 

in Bridgmanite at high pressure (Frost and McCammon, 2008), has also been proposed as having 129 

had a significant influence on the final concentrations of the MSE in the Earth’s mantle (e.g., 130 

Wade and Wood, 2005).  131 

By contrast, the BSE is characterized by relative abundances of HSE that are in 132 

approximately chondritic proportions (Fig. 1) (Chou, 1978; Morgan, 1986; Meisel et al, 2001; 133 

Becker et al., 2006; Fischer-Gödde et al., 2011), and absolute abundances that are only ~200 134 

times lower than bulk CI chondrite abundances (Morgan, 1986). Further, the 187Os/188Os and 135 

186Os/188Os estimated for BSE, reflecting the long-term decay of 187Re and 190Pt (where t½ for 136 

187Re and 190Pt are ~42 and 450 Gyr, respectively), are also within the range of chondritic 137 

meteorites, and provide a robust record for long-term, chondritic Re/Os and Pt/Os (Morgan, 138 



8 
 

1985; Walker et al., 1997; Meisel et al., 2001; Brandon et al., 2006). These characteristics of the 139 

HSE in the BSE do not appear to be the consequence of the high pressure and temperature metal-140 

silicate equilibration. Experimental studies have shown that large differences in the metal-silicate 141 

distribution coefficients of siderophile elements at relevant temperatures and pressures, would 142 

have led to non-chondritic relative abundances in the mantle (Holzheid et al., 2000; Brenan and 143 

McDonough, 2009; Mann et al., 2012). This is observed for MSE but not for HSE (Fig. 1). 144 

Instead, the HSE may owe their presence in the BSE to continued accretion of planetesimals with 145 

bulk chondritic compositions, following cessation of core formation (Kimura et al., 1974; Chou, 146 

1978). This process has been termed the late meteorite influx by some, and late accretion by 147 

others (henceforth, we will use the term late accretion).  148 

The addition of bodies with chondritic bulk compositions to silicate mantles would lead to 149 

the establishment of comparatively high absolute, and chondritic relative abundances of HSE in 150 

the affected mantles. In the case of Earth, mass balance constraints suggest that it would be 151 

necessary to add 0.3 to 0.8 wt. % of the mass of the total Earth (~2 x 1022 kg) to the mantle by 152 

late accretion in order to account for present mantle HSE abundances (Walker, 2009). Implicit in 153 

such models are the assumptions that abundances of all HSE in the mantle were very low prior to 154 

late accretion, and that metal present in late stage impactors was ultimately oxidized so that the 155 

siderophile elements added to the mantle after core formation remained in the mantle (Kimura et 156 

al., 1974). Although there are some problems associated with the late accretion hypothesis it 157 

currently is the only viable process to explain the HSE characteristics of the BSE (Walker, 158 

2009).  159 

The two mechanisms for the establishment of siderophile elements into the terrestrial 160 

mantle are not necessarily in conflict. Models such as those proposed by Wade and Wood (2005) 161 



9 
 

and Rubie et al. (2011; 2015) lead to the establishment of typically >90% of the MSE 162 

abundances in the mantle by high pressure and temperature metal-silicate partitioning. Repeated 163 

processing of metal through the mantle, as a result of oligarchic growth and the resulting 164 

multiple stages of magma ocean formation and evolution, would lead to a biasing of the MSE 165 

present in the mantle today towards the mass added by the later major stages of accretion (Azbel 166 

et al., 1993; Kramers, 1998; Rubie et al., 2011, 2015). Subsequent late accretion of ~0.5 wt.% 167 

mass addition would add >95% of the HSE, but a maximum of only about 10% of an MSE, such 168 

as W, to the mantle total. Thus, as noted by Dauphas et al. (2004), the genetics of the MSE 169 

present in the mantle are not required to have been the same as the genetics of the HSE. 170 

 171 

3.  Isotopic Tracers 172 

Two types of isotopic tracers of building blocks are considered here; radiogenic and 173 

nucleosynthetic. Long-lived radiogenic tracers have long proved to be useful for constraining the 174 

vigor and nature of mixing in the silicate portions of planets. For example, long-lived systems 175 

including 87Rb-87Sr, 147Sm-143Nd, 176Lu-176Hf, and 238,235U-232Th-206,207,208Pb have been 176 

extensively used to trace mantle mixing over Earth history, particularly with respect to recycled 177 

crustal components (Zindler and Hart, 1986; Hofmann, 2003). Of the long-lived systems 178 

consisting of HSE, however, only the 187Re-187Os isotopic system, and to a lesser extent the 190Pt-179 

186Os system, have been shown to be useful for characterizing late stages of planetary accretion 180 

to Earth, Moon and Mars (e.g., Morgan, 1985; Walker et al., 2002a; Meisel et al., 1996; Day et 181 

al., 2007; Brandon et al., 2012).  182 

Short-lived radiogenic isotopic systems have much greater utility for examining the timing 183 

of early planetary accretion and differentiation processes with high resolution (e.g., Caro et al., 184 



10 
 

2003; Foley et al., 2005; Boyet and Carlson, 2005; Debaille et al., 2009; McLeod et al., 2014). 185 

Although short-lived isotopic systems, most notably the lithophile-atmophile 129I-129Xe system 186 

(t½ = 15.7 Myr), the lithophile 146Sm-142Nd system (t½ = 103 Myr), and the lithophile-siderophile 187 

182Hf-182W system (t½ = 8.9 Myr), were initially pursued primarily to date cosmochemical 188 

materials and processes, radiogenic decay of these systems has also led to the production of long-189 

term isotopic heterogeneity in an array of sizable cosmochemical reservoirs, including the lunar 190 

and martian mantles (Nyquist et al., 1995; Foley et al., 2005; Debaille et al., 2009). 191 

Consequently, variations in the abundances of the daughter isotopes can also serve as tracers of 192 

planetary mantle mixing. Because the parent isotopes of these systems were extant for relatively 193 

short periods of time, ranging only from about 60 Myr for 182Hf, to as long as 600 Myr for 146Sm, 194 

the heterogeneities they record can only have formed early in solar system history.  For example, 195 

the large differences in 182W/184W and 142Nd/144Nd isotopic ratios between nakhlite and 196 

shergottite meteorites, with both groups presumably derived from the martian mantle, can only 197 

be explained as a result of a major, possibly global, differentiation event within the first 15 to100 198 

million years of solar system history (Foley et al., 2005; Debaille et al., 2009). Little subsequent 199 

mixing between the mantle domains, from which the precursor melts to these rocks originated, 200 

has evidently occurred.    201 

Of major significance to the possibility of identifying late-stage building blocks of Earth is 202 

the detection of small isotopic anomalies in all three radiogenic isotope systems, present in 203 

materials derived from the terrestrial mantle (e.g., Caro et al., 2003, 2006; Willbold et al., 2011; 204 

Mukhopadhyay, 2012). Here, the term anomaly refers to an isotopic composition that differs 205 

from that of the dominant composition measured for modern terrestrial samples. The existence of 206 

these anomalies highlights the long-term survival of isotopically-distinct mantle domains, which 207 



11 
 

likely formed within the first 50 million years of solar system history. With regard to identifying 208 

building blocks, the reasoning applied here is that if endogenous radiogenic isotope anomalies 209 

that formed very early in Earth history survived hundreds of millions to billions of years, then 210 

chemical and potentially nucleosynthetic isotopic heterogeneities, introduced to the mantle 211 

through the incorporation of genetically diverse building blocks during the period of late 212 

accretion, might also have survived and been incorporated into the rock record. 213 

The application of nucleosynthetic anomalies as tracers of building blocks is based on the 214 

observation that the isotopic compositions of numerous elements present in presolar grains 215 

reflect diverse formational pathways in high energy, stellar environments. Nucleosynthetic 216 

processes include the slow accumulation of neutrons, or the s-process, such as by He burning in 217 

the outer shell of an asymptotic giant branch star; rapid accumulation of neutrons, or the r-218 

process, such as during a Type II supernovae; and by proton enrichment, or the p-process, such 219 

as by photo-disintegration of heavy elements or neutrino interactions (Burbidge et al., 1957). 220 

Evidence for each of these nucleosynthetic processes has been confirmed through the 221 

observation of predicted, large-scale isotopic enrichments and depletions in diverse presolar 222 

grains extracted from low metamorphic-grade primitive meteorites (e.g., Zinner, 1998; Nittler, 223 

2003). In addition to direct measurement of nucleosynthetic variations in presolar grains, 224 

information regarding the identity and origin of presolar materials has also been obtained 225 

through chemical leaching of bulk primitive meteorites, followed by separate analysis of 226 

leachates and residues (Dauphas et al., 2002a; Hidaka et al., 2003; Schönbächler et al., 2005; 227 

Yokoyama et al., 2010). 228 

Although the isotopic variability observed among presolar components was greatly 229 

attenuated by nebular mixing processes, some elements continued to remain isotopically 230 



12 
 

heterogeneous during the formation of moderate-sized bodies. For example, Yin et al. (2002) and 231 

Dauphas et al. (2002b) reported isotopic variations on the order of several parts per ten thousand 232 

for the MSE Mo, in bulk samples of chondrites and iron meteorites. Some of these iron 233 

meteorites have been interpreted to sample the cores of diverse asteroidal- to embryonic-size 234 

bodies, in certain cases with diameters that may have exceeded 300 km (Yang et al., 2008). More 235 

recently, Burkhardt et al. (2011) reported a similar range of mass independent Mo isotopic 236 

composition variability among iron meteorites, chondrites and pallasites (Fig. 2a-b). Similarly, 237 

Chen et al. (2010) reported comparably large 100Ru isotopic heterogeneities among iron 238 

meteorite and chondrite groups (Fig. 3a-b). The causes for such large-scale isotopic 239 

heterogeneity remain debated, but can include the preferential concentration of some types of 240 

presolar components into parent bodies, resulting from thermal or physical sorting mechanisms 241 

of host phases in the solar nebula (Trinquier et al., 2007; Regelous et al., 2008). Alternately, 242 

these large-scale nucleosynthetic anomalies could reflect injections of diverse nucleosynthetic 243 

materials from nearby supernovae at different times during the evolution of the nebula (Bizzarro 244 

et al., 2007), followed by incomplete homogenization of these materials throughout the nebula 245 

(Carlson et al., 2007; Andreasen and Sharma, 2007). Regardless of the true cause, the evidence 246 

for isotopic heterogeneity in nucleosynthetic componentry among diverse planetesimals is 247 

strong. Like mass independent isotopic differences in O isotopes, the isotopic variations in these 248 

elements can potentially serve as fingerprints of their unique origins in the nascent solar system. 249 

 250 

4.  The Final ~10 wt. % of Accretion 251 

 The canonical model for the formation of the Moon involves a giant impact of an 252 

impactor comprising approximately 10% of the mass of the Earth (Canup and Asphaug, 2001). 253 



13 
 

Although recent models for the putative giant impact have expanded the possible range for the 254 

mass of the impactor from as little as 5% (Cuk and Stewart, 2012) to as much as 45% of the 255 

current mass of the Earth (Canup, 2012), there is little doubt that it was the final major 256 

accretionary addition to Earth. The event led to the transit of appreciable metal from the impactor 257 

core through the Earth’s mantle. As a consequence, it could also have led to a substantial 258 

modification of siderophile element abundances and isotopic compositions retained in the silicate 259 

portion of the Earth, following the impact (e.g., Rubie et al., 2011). As evidenced by the mass 260 

balance constraints discussed above, subsequent late accretion could not have modified the 261 

abundances of the MSE in the mantle by more than ~10%. Thus, the giant impact would have 262 

been the last major event involved in the establishment of the modern characteristics of MSE in 263 

the terrestrial mantle. Consequently, some characteristics of MSE in the mantle today may 264 

provide insights to the nature of the final massive impactor involved in the construction of Earth 265 

and Moon. 266 

Of all the MSE, the modification of the mantle by giant impact has been most frequently 267 

studied using W isotopes, because of the radiogenic tracer capability inherent in the system 268 

(Halliday, 2004; Dwyer et al., 2014). Unfortunately, the W isotopic composition of the mantle 269 

today can provide only modest constraints regarding the nature of the giant impactor. This is 270 

because the outcomes of giant impact models involving the Hf-W isotopic system are strongly 271 

dependent on assumptions about the timing of the impact, and of core formation on both the 272 

proto-Earth and impactor. Of particular importance is the degree of equilibration that occurred 273 

between the core of the impactor and the silicate Earth, as metal from the impactor transited the 274 

mantle on its way to merge with the Earth’s core (e.g., Halliday, 2004; 2008). In contrast to the 275 

W present in metal derived from the core of the impactor, it is assumed that nearly all of the W 276 



14 
 

present in the silicate portion of the impactor ended up mixing into the terrestrial mantle, or 277 

being lofted away from Earth to form the silicate portion of the Moon.  278 

With regard to metal-silicate equilibration, one endmember possibility is that the core of 279 

the impactor quickly merged with the core of the Earth, with negligible exchange of W between 280 

the merging core and the silicate Earth (Fig. 4a). In this case, the 182W/184W of the Earth’s 281 

mantle would likely have increased as a result of the addition of W from only the silicate portion 282 

of the impactor (Halliday, 2004). The 182W/184W of the silicate portion of the impactor was 283 

probably more radiogenic that that of the proto-Earth’s mantle, because the silicate portion of the 284 

embryonic-size impactor is usually presumed to have accreted and differentiated a core more 285 

rapidly than Earth, creating a high Hf/W mantle, while 182Hf was still plentiful (Dauphas and 286 

Pourmand, 2011). Thus, for this scenario, the 182W isotopic composition of the silicate Earth 287 

could have risen 100 ppm or more as a result of the impact. 288 

The other endmember possibility is that the core of the impactor broke into small droplets, 289 

resulting from Rayleigh-Taylor instabilities, and the droplets equilibrated with surrounding 290 

silicate liquid as they sank through the mantle to merge with the terrestrial core (Dahl and 291 

Stevenson, 2010). In this case, the W isotopic composition of the silicate Earth would likely have 292 

decreased significantly (Fig. 4b), as W with low 182W/184W from the impactor core (182W≤ -293 

200) equilibrated with the presumably more radiogenic silicate Earth (Halliday, 2008).  In this 294 

case, the mantle of the Earth just prior to the impact may have been more than 200 ppm more 295 

radiogenic than the present mantle. 296 

Because of these uncertainties in the extent of equilibration between metal and silicate, it is 297 

currently impossible to estimate the 182W value of the impactor mantle, as well as the age of 298 

average core formation for either the impactor or Earth. Nevertheless, W isotopes may have 299 



15 
 

some utility in characterizing the differentiation history of this major building block of Earth. 300 

One possible explanation for the W isotopic similarity between the Earth and Moon (Touboul et 301 

al., in review) is that the impactor and Earth formed from genetically similar materials and had 302 

roughly similar average core segregation ages (e.g., Dauphas et al., 2014). Information that is 303 

currently being gleaned about the drawdown of mantle HSE abundances at the time of giant 304 

impact (e.g., Touboul et al., in review), coupled with an improved understanding of the rate of 305 

isotopic versus elemental equilibration between metal and silicate, may one day enable stronger 306 

constraints to be placed on the formation and differentiation history of the impactor. 307 

More germane to the current genetic characterization of this stage of terrestrial growth may 308 

be the isotopic composition of the MSE Mo. The Mo isotopic composition of the silicate Earth 309 

differs from all known chondrite groups and most iron meteorite groups (Dauphas et al., 2002b; 310 

Burkhardt et al., 2012)(Fig. 2a-b). The only major meteorite group measured to high precision 311 

that matches the Mo isotopic composition of the silicate Earth is the group IAB iron meteorites 312 

(Burkhardt et al., 2011). It has been proposed that this group of iron meteorites formed as a result 313 

of metal pooling at the base of an impact crater following an impact onto a primitive body 314 

(Wasson and Kallemyn, 2002). This iron group, however, shares little other elemental or isotopic 315 

commonality with the Earth. For example, the 17O composition range of silicates present in 316 

IAB irons differs substantially from the Earth (Wasson and Kallemeyn, 2002). Therefore, even if 317 

their Mo isotopic compositions overlap, the group is probably not genetically closely related to 318 

Earth.  319 

Enstatite chondrites have commonly been touted as good genetic, if not compositional, 320 

matches to the Earth (e.g., Javoy et al, 2010). This conclusion has been based on the fact that the 321 

isotopic compositions of several lithophile elements in enstatite chondrites, including O, Ti, Cr 322 



16 
 

and Ba, as well as the siderophile element Ni, are excellent (Cr, Ti, Ni, Ba) or near (O) matches 323 

to their isotopic compositions in the silicate Earth (Clayton, 2004; Carlson et al., 2007; Trinquier 324 

et al., 2007; 2009; Regelous et al., 2008; Herwartz et al., 2014). Not all isotopic compositions 325 

match, however. Fitoussi and Bourdon (2012) reported a ~30 ppm difference in the Si isotopic 326 

composition between the silicate Earth and enstatite chondrites. They argued that this difference 327 

cannot be accounted for by incorporation of Si in the terrestrial core. With the possible exception 328 

of the siderophile Ni, however, the previously mentioned elements should provide little 329 

constraint on late stages of terrestrial accretion. Instead, their isotopic compositions are 330 

fingerprints of the dominant contributors to the bulk planet. 331 

The Mo isotopic compositions of enstatite chondrites overlap with the BSE, within the 332 

level of analytical precision (Burkhardt et al., 2011) (Fig. 2b). Burkhardt et al. (2011), however, 333 

concluded that the characteristic w-shaped pattern of isotopic enrichments and depletions for 334 

enstatite chondrites, as with ordinary and carbonaceous chondrites, is consistent with minor s-335 

process depletion.  Even if there is a small difference in the Mo isotopic composition of enstatite 336 

chondrites and the BSE, this does not necessarily mean that the Mo isotopic composition of the 337 

bulk Earth differs from that of enstatite chondrites. Most of Earth’s Mo is sited in the core, and 338 

most of that Mo was added to the Earth through initial stages of planetary accretion. It is, 339 

therefore, possible that the Mo isotopic composition of the core is characterized by a slight 340 

enrichment in s-process isotopes compared to enstatite chondrites. In this event, the final ~10% 341 

growth of Earth may have involved materials with an average Mo isotopic composition that was 342 

slightly depleted in r-process isotopes relative to the silicate Earth today, such that the core and 343 

silicate Earth combined has a composition that is similar to enstatite chondrites. As with W, the 344 

fraction of Mo present in the mantle derived from the impactor is not known, nor is the degree of 345 



17 
 

equilibration between the core of the impactor and the Earth’s mantle. Consequently, the true Mo 346 

isotopic composition of the impactor cannot be well established at this time. Nevertheless, the 347 

present results suggest that the final major addition of Mo to Earth in the form of a giant impact 348 

involved an impactor with a Mo isotopic composition broadly similar to enstatite chondrites (see 349 

additional discussion below). 350 

Application of the MSE to study the nature of late stage building blocks is not limited to 351 

Earth. The Mo isotopic compositions of the two shergottites analyzed by Burkhardt et al. (2011) 352 

overlap with the composition of the silicate Earth, so no genetic difference between these two 353 

bodies can be detected at the current level of analytical resolution for this element. However, the 354 

uncertainties on the measured Mo isotopic composition for these samples are relatively large. 355 

Future refinement of the Mo isotopic composition of Mars, and determination of whether or not 356 

it has a uniform isotopic composition, will be critical to assessing whether late stage additions to 357 

the Earth and Mars came from genetically similar materials, and whether late stage building 358 

blocks to Mars were well homogenized before the large-scale differentiation events that led to 359 

isotopic heterogeneity in 182W and 142Nd. 360 

  361 

5.  The Final ~0.5 wt.% of Accretion 362 

5.1. Highly Siderophile Elements in the Bulk Silicate Earth 363 

The absolute and relative abundances of HSE in primitive meteorites vary among the major 364 

chondrite groups (Walker et al., 2002a; Horan et al., 2003; Brandon et al., 2005; Tagle and 365 

Berlin, 2008; Fischer-Gödde et al., 2010)(Fig. 5). For example, Pd/Ir in some enstatite chondrites 366 

tend to be higher than for ordinary, carbonaceous, or R-type chondrites. Given that Re/Os also 367 

vary among the chondritic groups, the 187Os/188Os of chondrites can serve as an important, 368 



18 
 

complementary parameter to discriminate among chondrite groups. Most notably, carbonaceous 369 

chondrites are characterized by generally lower 187Os/188Os, and therefore lower long-term 370 

Re/Os, than ordinary and enstatite chondrites (Fig. 6). Modest, long-term differences in Pt/Os 371 

have also resulted in small differences in 186Os/188Os among some chondrite groups (Brandon et 372 

al., 2006).  373 

Although variations in the relative abundances of HSE are becoming increasingly better 374 

constrained for chondritic components (e.g., Horan et al., 2009; Archer et al., 2014), except for 375 

calcium-aluminum rich inclusions, there has been only limited success in determining the causes 376 

of the variations among the HSE characteristics of bulk chondrites (Mason and Taylor, 1982; 377 

Sylvester et al., 1990; Fischer-Gödde et al., 2010). Thus, no nebular or parent body processes can 378 

currently be firmly linked to the HSE characteristics observed in bulk chondrites. This means the 379 

process that led to lower, long-term Re/Os in carbonaceous chondrites, compared to other 380 

chondrite groups, may not be related to the accompanying, generally more volatile-rich nature of 381 

carbonaceous chondrites. Despite these current limitations to fingerprinting late accretionary 382 

additions, if most of the mass of HSE present in the BSE today was added as a result of late 383 

accretion, the relative abundances of these elements in the BSE should provide an averaged 384 

compositional snapshot of the final ~0.5% of mass addition to Earth.  385 

Establishing HSE abundances in the BSE with sufficient precision to make comparisons to 386 

possible cosmochemical precursors is problematic. Because of the sensitivity of 187Os/188Os 387 

ratios to discriminate among small, long-term differences in Re/Os, Os isotopes have been 388 

especially heavily used among the HSE to characterize the nature of late accretion to Earth (Hirt 389 

et al., 1963; Morgan, 1985; Meisel et al., 1996; 2001). However, the application is not so 390 

straightforward. Although Os is highly compatible within the mantle, the Os isotopic 391 



19 
 

composition of the BSE cannot be measured directly via analysis of the dominant silicate 392 

reservoir in Earth, the oceanic mantle. This is because Re behaves incompatibly during mantle 393 

melting (Barnes et al., 1985; Rehkämper et al, 1999; Pearson et al., 2004). Consequently, oceanic 394 

and continental crustal extraction has modified the Re/Os of the residual oceanic mantle over 395 

Earth history. Abundances of Re estimated for the continental crust are relatively low (Peucker-396 

Ehrenbrink and Jahn, 2001), and given its limited mass, its formation is unlikely to have led to 397 

significant modification of the Re/Os of the residual mantle. By contrast, the formation of 398 

oceanic crust, with its comparatively high Re concentrations, may have significantly modified 399 

the Re/Os of the residual mantle. The magnitude of this modification is open for debate, as some 400 

of the Re extracted into oceanic and continental crust has been recycled back into the mantle. 401 

How much of the recycled Re and Pt has been re-mixed back into the oceanic mantle is 402 

problematic to assess (Walker et al., 2002b).  403 

To circumvent this problem, Meisel et al. (1996; 2001) applied a projection method 404 

utilizing the compositions of variably melt depleted mantle peridotite xenoliths to estimate the 405 

187Os/188Os of the BSE. They plotted 187Os/188Os (record of long-term Re/Os) versus Al2O3 or 406 

Lu, indicators of mantle melt depletion that are not as easily modified by secondary processes as 407 

Re, and projected the resulting linear trends to points of intersection with an assumed Al2O3 or 408 

Lu composition for the BSE. Using this method, Meisel et al. (2001) reported a ratio of 0.1296 ± 409 

0.0008 (2). This ratio must be considered a minimum because the projections were made mainly 410 

from samples derived from sub-continental lithospheric mantle. Such peridotites must have been 411 

physically separated from the oceanic mantle at some time prior to isolation as sub-continental 412 

lithospheric mantle. The oceanic mantle itself was most likely variably depleted in Re prior to 413 

these reservoirs transitioning from oceanic to sub-continental lithospheric mantle, so the melt 414 



20 
 

depletion events recorded in these rocks likely include at least one stage of prior melt depletion. 415 

The 187Os/188Os estimated for the BSE is at the upper end of the range of compositions recorded 416 

in bulk ordinary and enstatite chondrites (Fig. 6). Given this, the 187Os/188Os of the BSE appears 417 

to be most similar to ordinary and enstatite chondrites, or even slightly more radiogenic, and this 418 

likely means that carbonaceous chondrites, or similar materials, were not major players in late 419 

accretion (Walker et al., 2002a). Given the volatile-rich nature of some carbonaceous chondrite 420 

groups, this observation in turn has been taken as evidence that late accretion provided little 421 

water to Earth (Drake and Righter, 2002), although as noted, there is currently no known process 422 

that relates the incorporation of volatiles and low Re/Os into the parent bodies of carbonaceous 423 

chondrites.  424 

Absolute and relative abundances of other HSE have been estimated for the BSE using a 425 

similar approach as for 187Os/188Os (Becker et al., 2006; Fischer-Gödde et al., 2011). When 426 

collectively considering the HSE characteristics estimated for the BSE as compared with 427 

chondrites (Fig. 7), it appears to be most similar to enstatite chondrites for Re/Os, Os/Ir, Pt/Ir and 428 

Pd/Ir (Becker et al., 2006). However, the BSE appears to be modestly suprachondritic with 429 

respect to Ru/Ir and possibly Pd/Ir (Becker et al., 2006). Thus, although the current estimate of 430 

the HSE composition of the BSE is most like enstatite chondrites, it is not a perfect match to any 431 

known chondrite group, or individual chondrites present in our collections. This could mean that 432 

late accretion to Earth was dominated by one or more components that were somehow processed 433 

in the nebula, or on a parent body, such that HSE were fractionated relative to presently sampled 434 

chondrites. This in turn could occur as a result of the location of formation within a chemically 435 

heterogeneous protoplanetary disk, or formation at a different time in a chemically evolving 436 

nebula. Given that there are bulk chondrites with individual HSE ratios that extend beyond those 437 



21 
 

estimated for the BSE (although no one chondrite has all of the HSE characteristics of the BSE), 438 

nebular or parent body processes can evidently cause such fractionations. It is also possible that 439 

the fractionated HSE abundances present in the dominant late accretionary component, could 440 

result from another process, such as crystal-liquid fractionation of a metal component (e.g., 441 

Fischer-Gödde et al., 2011).  442 

It is important to recognize that there are some limitations to the projection approach 443 

towards characterizing the BSE and fingerprinting late accretionary additions, using HSE. Of 444 

greatest concern is the precision and accuracy of HSE estimates for the BSE. The absolute and 445 

relative concentration estimates in the BSE are based on measurements of mantle peridotites that 446 

may represent the end stage of multiple processes, including mantle melting, metasomatism and 447 

crustal recycling (Alard et al., 2000; le Roux et al., 2007). Thus, although the broadly chondritic 448 

nature of the BSE is of little doubt, the relatively small differences in elemental ratios that are 449 

key to discriminating among chondritic groups, or fingerprinting a heretofore unidentified 450 

chondrite-like contributor to the mantle, have been called into question (e.g., Lorand et al., 451 

2009). The primary question is whether any mantle peridotites can provide sufficient fidelity in 452 

recording the BSE composition. The typically strongly compatible natures of Os, Ir and Ru, 453 

make them less susceptible to modification by partial melting and metasomatic processes, 454 

compared to Pt, and especially the incompatible Pd, Au and Re. Consequently, they provide the 455 

strongest constraints on the HSE composition of the BSE, and the suprachondritic nature of Ru/Ir 456 

in the BSE appears in little doubt. Further, the similarity of HSE characteristics of large numbers 457 

of variably “fertile” mantle peridotites, ranging from abyssal peridotites, to peridotites from the 458 

mantle sections of ophiolites, to oceanic mantle xenoliths, supports the contention that the BSE is 459 

characterized by suprachondritic Ru/Ir and possibly Pd/Ir (Becker et al., 2006; Fischer-Gödde et 460 



22 
 

al., 2011). However, compilation of an even larger number of data from all types of mantle 461 

lithologies, combined with an improved understanding of how HSE from oceanic crust have 462 

been recycled back into the oceanic mantle, may ultimately be required to assemble a high-463 

confidence understanding of HSE in the BSE. 464 

 465 

5.2.  Constraining the Physical Nature of Late Accretion 466 

One key aspect of utilizing siderophile elements as genetic tracers of planetary building 467 

blocks requires knowledge of the specific physical processes involved in their incorporation into 468 

planetary mantles. This is especially true for the HSE. For example, late accretion of HSE to 469 

Earth’s mantle may have occurred as a consequence of a relatively gentle rain of smaller bodies 470 

onto the surface of the planet, as envisioned by some workers to form a late veneer (Anders, 471 

1968; Turekian and Clark, 1969). In this case, a dog’s breakfast of HSE-rich materials of diverse 472 

genetics may have been slowly mixed downward into the mantle as a result of crustal recycling, 473 

coupled with mantle convection. Maier et al. (2009) noted that such a process can potentially 474 

account for the low abundances of HSE in some early Archean komatiites. Komatiites are high 475 

MgO lavas, which are generally presumed to be derived from high degrees of partial melting in 476 

deep mantle upwellings, or mantle plumes (Campbell et al., 1989). Possible identification of a 477 

deep mantle source that was initially depleted in HSE, for komatiites >2.9 Ga, is consistent with 478 

the slow downward mixing of a late veneer into the deep mantle. Maier et al. (2009) argued that 479 

the “normal” HSE abundances found in komatiites formed later than ~2.9 Ga indicates that the 480 

putative veneer had become well-mixed into the deep mantle sources of komatiites by that time. 481 

The mantle source abundances of HSE have also been estimated for komatiites for which 187Re-482 

187Os isotopic systematics confirm post-crystallization, closed-system behavior of the rocks with 483 



23 
 

respect to HSE abundances (Fig. 8). The isotopic evidence for closed system behavior provides 484 

some additional confidence that the projected mantles source abundances are accurate.  These 485 

komatiites systems show significant variations in the absolute HSE abundances between the 486 

sources of late Archean komatiite systems, which are generally similar to those in the sources of 487 

early Archean komatiite systems, although there are notable exceptions, with the oldest early 488 

Archean komatiite system (3.55 Ga Schapenburg) having substantially lower HSE abundances 489 

compared to all other komatiite systems studied to-date.    490 

Another way to advance understanding of the physical nature of late accretion is by 491 

combining observations of the chemical characteristics of planetary materials with dynamical 492 

models for the first ~500 million years of solar system history. Here, comparing the 493 

characteristics of the HSE present in the terrestrial mantle with abundances present in the lunar 494 

and martian mantles may be particularly important. The abundances and isotopic compositions 495 

of the HSE in at least the upper portion of the terrestrial mantle are generally well defined and 496 

limited in variation. As noted, the 187Os/188Os and 186Os/188Os estimated for the BSE are within 497 

the range of chondritic meteorites (Meisel et al., 2001; Walker et al., 1997; Brandon et al., 2006). 498 

Further, oceanic and subcontinental lithospheric mantle peridotites the world over typically have 499 

relatively uniform abundances of Os and Ir, two HSE that are highly compatible during partial 500 

melting of the mantle (e.g., Rehkämper et al, 1999; Morgan et al., 2001). While there has long 501 

been a question as to whether the upper mantle is more enriched in HSE than the lower mantle, 502 

seismic tomography over the past 20 years has documented the exchange of matter between 503 

upper and lower mantle, providing evidence for mantle plumes rising from the lower mantle into 504 

the upper mantle and subducting slabs transiting from the upper mantle into the lower mantle 505 

(e.g., Goes et al., 1999; Nolet et al., 2006). It is, therefore, likely that there are no global-scale 506 



24 
 

HSE concentration variations within the mantle. Consequently, uncertainty in the mass of late 507 

accreted materials to the mantle, necessary to account for the observed abundances of HSE, 508 

primarily reflects the factor of two variation in HSE abundances between different types of 509 

chondritic meteorites (e.g., Horan et al., 2003), rather than uncertainties in the HSE content of 510 

the mantle.  511 

The concentrations of HSE in the lunar and martian mantles are much more difficult to 512 

constrain than for the terrestrial mantle. No mantle samples have, as yet, been collected from 513 

either body. Hence, mantle abundances of the HSE must be deduced from mantle-derived 514 

volcanic rocks. Based on HSE abundances present in leached samples of picritic glass spherules 515 

and lunar basalts, Walker et al. (2004) and Day et al. (2007), respectively, estimated that HSE 516 

concentrations in the lunar mantle are a factor of 20 or more lower than in the terrestrial mantle 517 

(Fig. 9a). Such a large difference in concentration, if correct, cannot be explained by 518 

gravitational focusing or inefficiency of impactor retention. The lower concentrations may 519 

instead reflect proportionally much less mass added to the lunar mantle by late accretion, 520 

compared to the Earth. This in turn could result from a longer period of late accretion for Earth, 521 

compared to the Moon, i.e., one that began well before formation of the Moon. However, recent 522 

W isotopic data for the Moon suggest that the Moon-forming giant impact efficiently removed 523 

HSE from the mantle to Earth’s core and reset the late accretionary clock for the two bodies 524 

(Touboul et al., in review). If so, this means that the dominant late accretionary periods of the 525 

Earth and Moon began after formation of the Moon, and that they were contemporaneous. Unless 526 

substantial quantities of late accreted materials to the Moon are sequestered from our view in its 527 

lower crust (Schlichting et al., 2012), an interpretation for which there is currently no physical 528 



25 
 

evidence, there remains a sizable mismatch in the proportions of late accreted materials added to 529 

the mantles of the Earth and Moon. 530 

In contrast to the lunar mantle, the martian mantle appears to have HSE abundances that 531 

are surprisingly similar to those present in the terrestrial mantle. Basaltic and ultramafic 532 

shergottite meteorites, commonly believed to come from Mars, are characterized by HSE 533 

abundances that scale with the MgO content of the rocks in a way that is similar to terrestrial 534 

volcanic rocks (Brandon et al., 2012)(Fig. 9b). In addition, shergottites exhibit a range of initial 535 

187Os/188Os that is very similar to the range of compositions present in terrestrial mantle-derived 536 

volcanic rocks (Brandon et al., 2012). Given the fact that Mars is commonly presumed to have 537 

formed prior to the Earth (Dauphas and Pourmand, 2011), it might be expected to have a larger 538 

proportion of late accreted material mixed throughout its mantle. Instead, all existing data 539 

suggest that the HSE concentrations in the martian mantle are similar to those in the terrestrial 540 

mantle, indicating that a roughly similar proportion of late accreted materials was added to the 541 

mantle of Mars (Brandon et al., 2012).  542 

Bottke et al. (2009) reported that one way to account for the similarity of HSE abundances 543 

in the terrestrial and martian mantles, but much lower HSE abundances in the lunar mantle is by 544 

a process they termed stochastic late accretion. The principle of stochastic late accretion is that 545 

most late accretionary mass was added to the Earth and Mars by a very limited number of 546 

impacts of approximately Pluto mass bodies (~1 x 1022 kg). By chance, the Moon was not struck 547 

by any bodies of this size, and so retained relatively low abundances of HSE. Subsequent 548 

dynamical models have highlighted the probability that bodies of similar mass may have 549 

survived beyond the formation age of the Moon, and thus been available to drive late accretion 550 

(Marchi et al., 2014). 551 



26 
 

If stochastic late accretion correctly accounts for the apparent disparity in HSE abundances 552 

between the lunar and terrestrial mantles, it has a major implication for tracing late stage building 553 

blocks of the Earth, and possibly Mars. It requires that mass was added to the Earth by a limited 554 

number of impact events that likely would have generated discrete magma seas or lakes, rather 555 

than as a chemically and isotopically well-mixed veneer of small bodies. Consequently, if late 556 

stage impactors were added to the mantle in such a way that global melting did not occur, then 557 

the impactors may have imparted isotopically distinct HSE signatures to different portions of the 558 

mantle, assuming the impactors were genetically different from the average BSE, and from one 559 

another.  560 

But what is the likelihood that moderately-sized, early-formed mantle heterogeneities 561 

remained isotopically distinct for hundreds of millions of years, until they melt to produce rocks 562 

that become incorporated into the rock record? This is where it is important to consider data for 563 

the short-lived radiogenic isotope systems. Perhaps of greatest importance for consideration here 564 

are 182W isotopic data for rocks that were ultimately derived from the terrestrial mantle. 565 

Anomalous 182W/184W ratios have been identified in a number of ancient rocks, including ≥3.8 566 

Ga supracrustal rocks from Nuvvuagittuq, Quebec (Touboul et al., 2014), ~3.7 Ga supracrustal 567 

rocks from Isua, Greenland (Willbold et al., 2011), and 2.8 Ga komatiites from Kostomuksha, 568 

Fennoscandia (Touboul et al., 2012). All 182W anomalies for terrestrial rocks, reported to date, 569 

range between +5 and +15 ppm (Fig. 10).  570 

Terrestrial enrichments in 182W have been interpreted in two different ways. Willbold et al. 571 

(2011) reported 182W enrichments averaging ~13 ppm for 3.7 Ga supracrustal rocks from Isua, 572 

Greenland. These authors proposed that the enriched compositions reflect derivation of precursor 573 

rocks from a mantle domain that formed prior to a final major stage of late accretion, and that 574 



27 
 

this mantle domain remained mostly free of late accreted materials until it melted to form the 575 

Isua rocks. Thus, the mantle precursor materials to the Isua rocks formed by normal planetary 576 

accretion, were stripped of HSE by metal segregation during core formation, but remained 577 

isolated from HSE and W added by subsequent late accretion. This is a process that could lead to 578 

isotopic heterogeneity in the mantle long after 182Hf became extinct. It can be viewed as an 579 

exogenous process because it would have been controlled by the growth of 182W on bodies other 580 

than Earth. More importantly, this mantle domain was not contaminated with late accreted HSE 581 

as a result of mantle mixing until after the early Archean melting event that produced the Isua 582 

precursor rocks, presumably well after completion of the dominant phase of late accretion. This 583 

suggests inefficient mixing of the mantle during the Hadean through early Archean. If this 584 

interpretation is correct for the Isua rocks, then the mantle domain sampled by them should have 585 

been relatively devoid of HSE. Willbold et al. (2011), however, did not report complementary 586 

HSE for these rocks, although it should be recognized that constraining the concentrations of 587 

HSE in the mantle source(s) of such highly altered supracrustal rocks is challenging.  588 

A second means to account for anomalous 182W in mantle-derived rocks is by solid-liquid 589 

fractionation processes that may have occurred in the mantle while 182Hf was still extant. 590 

Because the absolute HSE abundances estimated for the mantle source of the Kostomuksha 591 

komatiites are identical, within uncertainties (Puchtel and Humayun, 2005), to those in the BSE 592 

estimates of Becker et al. (2006), Touboul et al. (2012) excluded the exogenous model of 593 

Willbold et al. (2011) for these rocks. They instead concluded that either metal-silicate 594 

fractionation in a basal magma ocean, or silicate crystal-liquid fractionation in a more 595 

conventional, whole mantle magma ocean led to the creation of a mantle domain characterized 596 

by high Hf/W. This in turn led to the formation of a high 182W/184W domain, as 182Hf decayed. 597 



28 
 

Because of the short lifetime of 182Hf, it was concluded that the fractionation events occurred 598 

within the first 30 Myr of solar system history. This process is considered endogenous because 599 

all of the necessary steps occurred within the Earth. Similar processes may also have led to the 600 

creation of some 142Nd anomalies (e.g., Brown et al., 2015). 601 

Regardless of the true mechanisms involved in the generation of terrestrial 182W anomalies, 602 

it is clear that their presence in the rock record requires the long term survival of chemical 603 

heterogeneity in the terrestrial mantle despite major melting events, such as the Moon-forming 604 

giant impact. Much larger 182W and 142Nd isotopic anomalies have been determined for some, 605 

but not all martian meteorites, so the martian mantle also likely escaped a final large-scale 606 

mixing event during the final stages of its growth. Thus, if these bodies experienced late stages 607 

of accretion from genetically diverse materials, it might be expected that attenuated signals from 608 

the various materials might be summoned from the rock record. 609 

The most promising element to examine for recording nucleosynthetic heterogeneities 610 

created by genetically diverse late accretion to Earth is Ru. As a HSE, Ru was strongly 611 

concentrated into metal by core formation processes. As noted above, the large range of 612 

nucleosynthetic isotopic compositions recorded in meteorites (e.g., Chen et al., 2010) provides 613 

this element with the utility for discriminating among diverse, late accretionary contributions to 614 

the mantle. Limited high precision analyses of terrestrial materials have, as yet, not identified 615 

any isotopic heterogeneity within the mantle (Bermingham et al., 2015), but the search has just 616 

begun. 617 

 618 

 619 

 620 



29 
 

5.3. The Mo-Ru Connection 621 

In the discussion above, it is noted that the Mo and Ru present in the mantle today were 622 

most likely emplaced by different late-stage accretionary processes. The Mo abundance of the 623 

mantle was dominantly established by the final stages of oligarchic growth, including the Moon-624 

forming giant impact. It probably represents a mixture of Mo from the silicate portion of the 625 

proto-Earth, as well as Mo from both the core and mantle of the giant impactor. By contrast, 626 

most Ru was likely added to the mantle by late accretion. Dauphas et al. (2004; 2014) made the 627 

important observation that, when plotting the range of nucleosynthetic heterogeneities for Mo 628 

and Ru in bulk planetary materials, most of these materials plot along a generally linear trend 629 

with the Earth, IAB irons, and enstatite chondrites plotting at one end, and some carbonaceous 630 

chondrites plotting at the other end of the trend (Fig. 11). Based on this correlation, these authors 631 

surmised that because each element recorded the genetics of different, late-stage accretionary 632 

events, it is logical to conclude that the materials involved in both stages of late accretion were 633 

genetically related and may have been derived from roughly the same portion of the 634 

protoplanetary disk. Thus, there may have been no major change in the genetic make-up of 635 

materials involved in the Moon-forming giant impact as compared to the final ~0.5% of late 636 

accreted materials (Dauphas et al., 2004; 2014). Differing genetics of these additions would 637 

otherwise have led to Earth plotting off of this correlation.  638 

However, Mo and Ru isotope data underpinning this important observation remain 639 

somewhat limited. The correlation is presently constructed using group averages determined for 640 

a limited number of meteorites from which the Mo and Ru isotope compositions are not obtained 641 

from the same pieces of meteorite. Consequently, the precise relationship between Mo-Ru in 642 



30 
 

different meteorites needs to be clarified using high precision Mo and Ru isotope composition 643 

data obtained from the same meteorite pieces.  644 

 645 

 646 

6.  Late Heavy Bombardment: The final ~0.05% of Late Accretion? 647 

It has long been hypothesized that the Earth-Moon system, and likely the entire inner solar 648 

system, underwent a phase of late accretion, termed late heavy bombardment (LHB), within the 649 

interval from ~4.1 to ~3.8 Ga. The evidence for this putative event primarily comes from 650 

geochronologic information obtained from a variety of shocked and/or melted lunar rocks (e.g., 651 

Turner et al., 1973; Tera et al., 1974; Kring and Cohen, 2002). For example, Tera et al. (1974) 652 

recognized that most rocks collected by the Apollo missions formed a linear trend on a plot of 653 

207Pb/206Pb versus 238U/204Pb that intersects concordia at about 3.9 Ga. Because of the ubiquity of 654 

this age, they inferred that the Moon underwent what they referred to as a terminal cataclysm. 655 

They envisioned the cataclysm to have been a relatively brief period of heavy bombardment 656 

(<300 million years) during which the surface of the Moon, and presumably the Earth, was 657 

peppered with impactors as large as ~200 km in diameter (e.g., Hurwitz and Kring, 2014), 658 

leading to the creation of at least some of the lunar basins. Subsequent studies of lunar impact 659 

melt rocks have provided strong support for a major disturbance in ages at about 3.9 Ga (e.g. 660 

Dalrymple and Ryder, 1993; Cohen et al., 2000), as few impact-modified lunar rocks yield ages 661 

older than ~3.9 Ga. Although the LHB had a major effect on shaping the surface of the Moon, it 662 

likely involved much less mass than is envisioned for late accretion as a whole. Even generous 663 

estimates for the mass of the LHB place the mass of materials involved as no more than about 664 

10% of estimates for the overall mass of late accretionary additions (Morgan et al., 2001).   665 



31 
 

Dynamical models for the evolution of the solar system have suggested some possible 666 

causes for a period of LHB (Morbidelli et al., 2001; Strom et al., 2005; Gomes et al., 2005). For 667 

example, Gomes et al. (2005) suggested that migration of Uranus and Neptune, resulting from 668 

Jupiter and Saturn entering a 2 to 1 orbital resonance, may have led to the perturbation in the 669 

orbits of small bodies from both the asteroid belt and the Kuiper belt. Despite these observations 670 

and models, it is also possible that the ~3.9 Ga age represents a re-set age for the samples from a 671 

limited areal extent on the near-side of the Moon, from which all Apollo samples were collected, 672 

or a sampling of ejecta from only the youngest of the major basins (Spudis et al., 2011). 673 

In addition to seeking to understand the timing of the LHB, it is also imperative to 674 

constrain the chemical characteristics of the materials involved because of the possibility that 675 

they delivered substantial water, and other volatile species, such as water, to the Earth and Moon. 676 

The primary means to examine the chemical characteristics of materials from the putative LHB 677 

has been to analyze lunar impact melt rocks that were created as a result of the basin-forming 678 

impacts. Such studies have been pursued since the Apollo missions, with chemical 679 

characterizations focused upon siderophile elements (e.g., Morgan et al., 1972; 1974; Korotev, 680 

1994). Siderophile element data for nearly all studies prior to ca. 2000 were obtained by neutron 681 

activation analysis, so elements such as Ir, Au, Ni and Ge, that can be well-measured by this 682 

method, were most commonly considered. For example, Morgan et al. (1974) concluded that 683 

rocks from the Apollo 17 site included meteoritic components from at least six impactors, none 684 

of which had siderophile element characteristics that perfectly matched known meteorites.  685 

Such studies as Morgan et al. (1974) assumed that endogenous lunar highlands or basaltic 686 

rocks formed with very low siderophile element abundances, so that the siderophile elements 687 

present in the impact melt rocks were nearly entirely derived from one or more large impactors. 688 



32 
 

A substantial number of studies reporting siderophile element data for so-called pristine lunar 689 

rocks have generally borne out this assumption (e.g., Warren and Wasson, 1977; Ryder et al., 690 

1980; Warren et al., 1991; Day et al., 2007; 2010).  691 

Although the early studies provided highly valuable insights into the chemical nature of 692 

impactors, some of the chief elements used to discriminate among possible impactors, such as 693 

Au and Ge, are somewhat volatile and could potentially have been modified by high temperature 694 

impact processes. To circumvent this problem, Norman et al. (2002) first applied the isotope 695 

dilution technique, teamed with inductively-coupled plasma mass spectrometry, to measure a 696 

larger suite of HSE in Apollo 17 impact melt rocks. That study measured and reported data for 697 

Re, Ir, Ru, Pt, and Pd, and identified at least three sources of HSE to the Apollo 17 suite. One 698 

source had HSE ratios similar to ordinary chondrites. A second component was characterized by 699 

HSE similar to EH chondrites. In order to account for supra-chondritic Re/Ir, Ru/Ir, and Pd/Ir in 700 

most of the rocks, Norman et al. (2002) appealed to the possibility of a third component, either 701 

an endogenous component enriched in Re, Ru and Pd, or an older component in the target crust 702 

that was incorporated into the crust by an earlier impactor with non-chondritic relative 703 

abundances of HSE.  704 

Four subsequent studies have utilized similar isotope dilution techniques to measure the 705 

abundances of Re, Os, Ir, Ru, Pt, and Pd in lunar impact melt rocks, as well as measure 706 

187Os/188Os, which serves as a sensitive proxy for long-term Re/Os (Puchtel et al., 2008; Fischer-707 

Gödde and Becker, 2012; Sharp et al., 2014; Liu et al., in press). A major difference between 708 

these studies and the study of Norman et al. (2002) is that they examined multiple pieces of each 709 

rock studied. In approximately half of the rocks examined, the resulting plots of Ir versus each of 710 

the other HSE measured yielded linear trends with intercepts indistinguishable from 0, within 711 



33 
 

regression uncertainties (Fig 12).  In such cases, the trends can be assumed to represent mixing 712 

between the exogenous impactor and the HSE poor lunar target rocks, similar to interpretations 713 

for terrestrial impact melt rocks (e.g., McDonald et al., 2001). The slopes of the linear trends can, 714 

therefore, be assumed to record the HSE ratios of the basin forming impactors.  715 

Puchtel et al. (2008) and Sharp et al. (2014) reported and interpreted data mainly for Apollo 716 

17 impact melt rocks. Both studies reported a “dominant” component for the site, most notably 717 

characterized by suprachondritic Re/Os (as measured by 187Os/188Os), as well as Ru/Ir and Pd/Ir 718 

comparable to the results from Norman et al. (2002). They interpreted the results to suggest that 719 

the dominant source of HSE to the site, most likely the spatially associated Serenitatis basin 720 

impactor, shared broad similarities to some chondritic meteorites (enstatite chondrites), but 721 

sampling a composition not presently found in our meteorite collections. By contrast, a feldspar 722 

rich, or granulitic component present as clasts in some of these rocks, was determined to be 723 

characterized by relative abundances of HSE more similar to known chondrites. 724 

Fischer-Gödde and Becker (2012) focused mainly on impact melt rocks from the Apollo 16 725 

site. Here, they found Re/Os, Ru/Ir, Pt/Ir, and Pd/Ir ratios extending much higher than in known 726 

chondrites, and even well beyond the range of Apollo 17 rocks. They also analyzed some 727 

granulitic rocks and reported that, like prior studies, this component is most like ordinary 728 

chondrites. Of note, this study recognized that virtually all of the HSE data for Apollo samples 729 

plot along linear trends of HSE/Ir versus 187Os/188Os. They interpreted this to mean that all of the 730 

Apollo impact melt rocks incorporated at least two HSE-rich components at the time of their 731 

formation. One was very similar to carbonaceous chondrites and is the major component in 732 

granulitic rocks. The other component resembles a chemically evolved group IVA iron 733 

meteorite. Consequently, they proposed that both components became variably mixed during 734 



34 
 

basin-forming impacts, but were not substantially modified by the inclusion of HSE derived from 735 

the basin-forming impactors. 736 

Most recently, data from Liu et al. (in press) for Apollo 15 and 16 melt rocks filled in the 737 

gaps in the apparent linear trend recognized by Fischer-Gödde and Becker (2012), thus, 738 

strengthening their observation. In the compilation of data reported by Liu et al. (in press), nearly 739 

all data for lunar impact melt rocks plot along a continuous linear trend ranging from a HSE 740 

composition that is broadly chondritic, to an endmember with 187Os/188Os, Ru/Ir and Pd/Ir ratios 741 

far above those of known chondrites (Fig. 13). Two possible scenarios to explain the observed 742 

trends are: 1) Variable mixing between an earlier granulitic contaminant and a series of later-743 

stage impactors that happened to form co-linear, suprachondritic Re/Os, Ru/Ir, Pt/Ir and Pd/Ir. 2) 744 

Variable mixing between two components present in the lunar crust prior to the late-stage basin 745 

forming impacts. For this scenario, one component was chondritic in composition and the other 746 

component had fractionated HSE, and could have been a core fragment, as suggested by Fischer-747 

Gödde and Becker (2012). Although the latter scenario requires the involvement of a portion of 748 

an evolved core, it currently appears to be the most simplistic explanation for the trend. It also 749 

requires that the later-stage basin forming impacts (e.g., Imbrium) added only very limited HSE 750 

to the sampled impact melt rocks from multiple sites. These models await genetic testing using 751 

the nucleosynthetic anomalies characteristic of siderophile elements Mo and Ru.  752 

  753 

7.  Putting It All Together 754 

At the present time, combined lithophile and siderophile element data suggest that the 755 

primary building blocks of Earth were broadly isotopically similar to enstatite chondrites, and 756 

that there was not a major change in the provenance of building blocks when comparing the pre-757 



35 
 

giant impact Earth, to the Moon-forming giant impactor, and to the materials involved in the 758 

subsequent late accretion of ~0.5 wt. % of mass to the silicate Earth. This conclusion is based on 759 

the reasoning that the isotopic similarities between the Earth and enstatite chondrites, for 760 

lithophile elements such as O and Cr indicate genetic similarity to enstatite chondrites prior to 761 

the giant impact. Yet, the siderophile Mo isotopic composition of Earth’s mantle is also very 762 

similar to that of enstatite chondrites. The isotopic composition of Mo present in the mantle was 763 

likely strongly affected by additions from the Moon-forming giant impactor, whereas the 764 

isotopic compositions of lithophile elements such as O and Cr were not. Thus, the collective 765 

enstatite chondrite-like isotopic compositions of lithophile and siderophile elements suggest that 766 

both the Earth and the giant impactor formed in the same region of the protoplanetary disk as 767 

enstatite chondrites (e.g., Dauphas et al., 2014).   768 

The genetic heritage of late accreted materials during the final 0.5 wt. % of terrestrial 769 

accretion is best monitored via Ru isotopes, combined with the relative abundances of the HSE 770 

in the BSE. The Ru isotopic composition of the mantle, as for Mo isotopes, is similar to enstatite 771 

chondrites, meaning that the Earth plots near enstatite chondrites at the end of the cosmic Ru-Mo 772 

correlation trend. Some aspects of the projected relative abundances of the HSE in the BSE also 773 

match certain enstatite chondrites (Pd/Ir and 187Os/188Os).  However, other aspects of the HSE 774 

signature of the BSE, such as Ru/Ir, do not to match any known chondrite groups. Thus, the late 775 

accreted materials must include at least one component with more fractionated HSE than is 776 

known to occur in chondrites. The origin of this chemical signature remains poorly constrained, 777 

but is suggestive of a not yet sampled primitive meteorite component. Very limited Os isotopic 778 

data for Mars suggest a similar late accretionary component was added to its mantle. 779 



36 
 

Finally, the Earth and Moon were bombarded by an additional flux of planetesimals 780 

hundreds of millions of years after primary accretion. The accretionary additions associated with 781 

this period could have totaled as much as 0.05 wt. % of the mass of the Earth. The materials 782 

involved in this final, minor accretionary period also involved the addition of HSE with some 783 

fractionated ratios. The chemical and isotopic natures of these materials are best monitored 784 

through the analysis of lunar impact melt rocks that were created by the late-stage basin-forming 785 

events. The fingerprints of these impactors are complex and encompass a range of HSE 786 

compositions. Some components evident in this bombardment cohort appear to be similar in 787 

HSE characteristics to carbonaceous and ordinary chondrites. An additional component was 788 

characterized by substantially higher Re/Os, Ru/Ir and Pd/Ir, compared to any known chondrites, 789 

including enstatite chondrites. The dominant signature of at least some materials involved in the 790 

LHB, therefore, appear distinct from the prior late-stage building blocks. The Ru and Mo 791 

isotopic compositions of lunar impact melt rocks have not yet been determined, so it remains 792 

unknown whether or not the LHB can be genetically linked to a type of primitive meteorite.   793 

Although the siderophile element data for the Earth suggest that there was no major change 794 

in the provenance of its building blocks through to the end of late accretion (but before LHB), it 795 

remains unknown whether or not the building blocks consisted of a genetically homogeneous 796 

flux, or included diverse materials that ultimately mixed to form what now appears to be a 797 

uniform fingerprint for the BSE. As outlined above, the likelihood of isotopic variability of Ru 798 

and Mo among possible building blocks, combined with the apparent sluggishness of early 799 

mantle mixing of primordial 182W isotopic heterogeneities, suggests that isotopic evidence for 800 

diverse late stage impactors might be found in Earth’s early rock record, and possibly in younger 801 

rocks. Conversely, given the high level of precision that is now available to search for such 802 



37 
 

isotopic heterogeneities, the future lack of discovery of isotopic anomalies may signal either that 803 

the materials involved in the final stages of terrestrial accretion were genetically similar, or that 804 

early mixing processes attenuated early Earth heterogeneities before evidence for them could be 805 

incorporated in the surviving rock record.  806 

 807 

Acknowledgements  808 

This work has been supported by NASA grants NNX13AF83G and NNA14AB07A, and 809 

NSF-CSEDI grants EAR1160728 and EAR1265169. These sources of funding are gratefully 810 

acknowledged. RJW also thanks the Carnegie Institution of Science’s Department of Terrestrial 811 

Magnetism, and the Lunar and Planetary Institute for hosting him during the writing of this 812 

paper.   813 



38 
 

References 814 

 815 

Alard, O., Griffin W.L., Lorand, J.-P., Jackson, S.E., O’Reilly, S.Y., 2000. Non-chondritic 816 

distribution of the highly siderophile elements in mantle sulphides.  Nature 407, 891-894. 817 

Albarede, F., Ballhaus, C., Blichert-Toft, J., Lee, C.-T., Marty, B., Moynier, F., Yin, Q.-Z., 2013. 818 

Asteroidal impacts and the origin of terrestrial and lunar volatiles. Icarus 222, 44–52. 819 

Anders, E., 1968. Chemical processes in the early Solar System, as inferred from meteorites. 820 

Acc. Chem. Res. 1, 289-298. 821 

Anders, E., Grevesse, N., 1989. Abundances of the elements: meteoritic and solar. Geochim. 822 

Cosmochim. Acta 53, 197-214. 823 

Andreasen, R., Sharma, M., 2007. Mixing and homogenization in the early Solar System: clues 824 

from Sr, Ba, Nd, and Sm isotopes in meteorites. Astrophys. Journ. 665, 874-883. 825 

Archer, G.J., Bullock, E.S., Ash, R.D., Walker, R.J., 2014. 187Re-187Os isotopic and highly 826 

siderophile element systematics of the Allende meteorite: evidence for primary nebular 827 

processes and late-stage alteration. Geochim. Cosmochim. Acta 131, 402-414. 828 

Azbel, I.Y., Tolstikhin, I.N., Kramers, J.D., Pechernikova, G.V., Vityazev, A.V., 1993. Core 829 

growth and siderophile element depletion of the mantle during homogeneous Earth accretion. 830 

Geochim. Cosmochim. Acta 57, 2889-2898. 831 

Barnes, S.-J., Naldrett, A.J., Gorton, M.P., 1985. The origin of the fractionation of plantinum-832 

group elements in terrestrial magmas. Chem. Geol. 53, 303-323. 833 

Becker, H., Horan, M.F., Walker, R.J., Gao, S., Lorand, J.-P., Rudnick, R.L., 2006. Highly 834 

siderophile element composition of the Earth’s primitive upper mantle: Constraints from 835 

new data on peridotite massifs and xenoliths. Geochim. Cosmochim. Acta 70, 4528-4550.  836 



39 
 

Bermingham, K.R., Walker, R.J., Worsham, E.A., 2015. Refinement of the Mo-Ru isotope 837 

cosmic correlation using high precision Mo and Ru isotope data. Lunar and Planet. Sci. 838 

Conf. 46, abstract # 1588. 839 

Bizzarro, M., Ulfbeck, D., Trinquier, A., Thrane, K., Connelly, J.N., Meyer, B.S., 2007. 840 

Evidence for a late supernova injection of 60Fe into the protoplanetary disk. Science 316, 841 

1178–1181. 842 

Borisov, A., Palme, H., Spettel B., 1994. Solubility of palladium in silicate melts: implications 843 

for core formation in the Earth. Geochim. Cosmochim. Acta 58, 705-716. 844 

Bottke, W.F., Walker, R.J., Day, J.M.D., Nesvorny, D., Elkins-Tanton, L., 2010. Stochastic late 845 

accretion to Earth, the Moon and Mars. Science 330, 1527-1530. 846 

Boyet, M., Carlson, R.W., 2005. 142Nd evidence for early (>4.53 Ga) global differentiation of the 847 

silicate Earth. Science 309, 576-581. 848 

Brandon, A.D., Humayun, M., Puchtel, I.S., Leya, I., Zolensky, M., 2005. Osmium isotope 849 

evidence for an s-process carrier in primitive chondrites. Science 309, 1233-1236. 850 

Brandon, A.D., Walker, R.J., Puchtel, I.S., 2006. Platinum-Os isotope evolution of the Earth’s 851 

mantle: constraints from chondrites and Os-rich alloys. Geochim. Cosmochim. Acta 70, 852 

2093-2103. 853 

Brandon, A.D., Puchtel, I.S., Walker, R.J., Day, J.M.D., Irving, A.J., Taylor, L.A., 2012. 854 

Evolution of the martian mantle inferred from the 187Re-187Os isotope and highly 855 

siderophile element systematics of shergottites meteorites. Geochim. Cosmochim. Acta76, 856 

206-235. 857 

Brenan, J.M., McDonough, W.F., 2009. Core formation and metal–silicate fractionation of 858 

osmium and iridium from gold. Nature Geoscience 2, 798-801. 859 



40 
 

Brown, S., Elkins-Tanton, L., Walker, R.J., 2014. Effects of magma ocean crystallization and 860 

overturn on the development of 142Nd and 182W isotopic heterogeneities in the primordial 861 

mantle.  Earth Planet. Sci. Lett. 408, 319-330. 862 

Burbidge, E.M., Burbidge, G.R., Fowler, W.A., Hoyle, F., 1957. Synthesis of the elements in 863 

stars. Rev. Mod. Phys. 59, 547-650.  864 

Burkhardt,C., Kleine, T., Oberli, F., Pack, A., Bourdon, B., Wieler, R., 2011. Molybdenum 865 

isotope anomalies in meteorites: Constraints on solar nebula evolution and origin of the 866 

Earth. Earth Planet. Sci. Lett. 312, 390-400. 867 

Campbell. I.H., Griffiths, R.W., Hill, R.I., 1989. Melting in an Archaean mantle plume: head it's 868 

basalts, tails it's komatiites. Nature 339, 697-699. 869 

Canup, R.M., Asphaug, E., 2001. Origin of the Moon in a giant impact near the end of the 870 

Earth's formation. Nature 412, 708-712. 871 

Canup, R.M., 2012. Forming a Moon with an Earth-like composition via a giant impact. Science 872 

338, 1052-1055. 873 

Carlson, R.W., Boyet, M., Horan, M., 2007. Chondrite barium, neodymium, and samarium 874 

isotopic heterogeneity and early earth differentiation. Science 316, 1175–1178. 875 

Caro, G., Bourdon, B., Birck, J.-L., Moorbath, S., 2003. 146Sm–142Nd evidence from Isua 876 

metamorphosed sediments for early differentiation of the Earth’s mantle. Nature 423, 428–877 

432. 878 

Caro, G., Bourdon, B., Birck, J-L., Moorbath, S., 2006. High-precision 142Nd/144Nd 879 

measurements in terrestrial rocks: constraints on the early differentiation of the Earth’s 880 

mantle. Geochim. Cosmochim. Acta 70, 164–191. 881 



41 
 

Chambers, J.E., 2001. Planetary accretion in the inner Solar System. Earth Planet. Sci. Lett. 223, 882 

241–252. 883 

Chen, J.H., Papanastassiou, D.A.,Wasserburg, G.J., 2010. Ruthenium endemic isotope effects in 884 

chondrites and differentiated meteorites. Geochim. Cosmochim. Acta 74, 3851–3862. 885 

Chou, C.-L., 1978. Fractionation of siderophile elements in the Earth’s upper mantle. Proc. 886 

Lunar Planet. Sci. Conf. 9th, 219-230. 887 

Clayton, R.N., Mayeda T.K., Rubin A.E., 1984. Oxygen isotopic compositions of enstatite 888 

chondrites and aubrites. Proc. Fifteenth Lunar Planet. Sci Conf., Part 1, Journ. Geophys. 889 

Res. 89, C245-249.  890 

Clayton, R.N., 2002. Self-shielding in the solar nebula. Nature 415, 860–861. 891 

Cohen, B.A., Swindle, T.D., Kring, D.A., 2000. Support for the lunar cataclysm hypothesis from 892 

lunar meteorite impact melt ages. Science 290, 1754-1756. 893 

Cottrell, E., Walker, D., 2006. Constraints on core formation from Pt partitioning in mafic 894 

silicate liquids at high temperatures. Geochim. Cosmochim. Acta 70, 1565-1580. 895 

Cuk, M., Stewart S.T., 2012. Making the Moon from a fast-spinning Earth: a giant impact 896 

followed by resonant despinning. Science 338, 1047-1052. 897 

Dahl, T.W., Stevenson, D.J., 2010. Turbulent mixing of metal and silicate during planet 898 

accretion - and interpretation of the Hf–W chronometer. Earth Planet. Sci. Lett. 295, 177-899 

186. 900 

Dalrymple, G.B. and Ryder, G., 1993. 40Ar/39Ar age spectra of Apollo 15 impact melt rocks by 901 

laser step‐heating and their bearing on the history of lunar basin formation. Journ. 902 

Geophys.  Res. 98, 13,085-13,095. 903 



42 
 

Dauphas, N. Marty, B., Reisberg, L., 2002a. Molybdenum nucleosynthetic dichotomy revealed 904 

in primitive meteorites, Astrophys. Journ. 569 (2002) L139– L142.  905 

Dauphas, N. Marty, B., Reisberg, L., 2002b. Molybdenum evidence for inherited planetary scale 906 

isotope heterogeneity of the protosolar nebula. Astrophys. Journ. 565, 640-644. 907 

Dauphas, N. Davis, A.M., Marty, B., Reisberg, L., 2004. The cosmic molybdenum–ruthenium 908 

isotope correlation. Earth Planet. Science Lett. 226, 465– 475. 909 

Dauphas, N., Pourmand, A., 2011. Hf–W–Th evidence for rapid growth of Mars and its status as 910 

a planetary embryo. Nature 473, 489–492. http://dx.doi.org/10.1038/nature10077. 911 

Dauphas, N., Burkhardt, C., Warren, P.H., Teng, F-Z., 2014. Geochemical arguments for an 912 

Earth-like Moon-forming impactor. Phil. Trans. Roy. Soc. A 372, 20130244.  913 

Day, J.M.D., Pearson, D.G., Taylor, L.A., 2007. Highly siderophile element constraints on 914 

accretion and differentiation of the Earth-Moon system. Science 315, 217-219.  915 

Day, J.M.D., Walker, R.J., James, O.B., Puchtel, I.S., 2010. Osmium isotope and highly 916 

siderophile element systematics of the lunar crust. Earth Planet. Sci. Lett. 289, 595-605. 917 

Debaille, V., Brandon, A.D., O’Neill, C., Yin, Q.-Z., Jacobsen, B., 2009. Early martian mantle 918 

overturn inferred from isotopic composition of nakhlite meteorites. Nature Geoscience 2, 919 

548-552. 920 

Drake, M.J., Righter, K., 2002. Determining the composition of the Earth. Nature 416, 39-44. 921 

Dwyer, C.A., Nimmo, F., Chambers, J.E., Bulk chemical and Hf–W isotopic consequences of 922 

incomplete accretion during planet formation. Icarus 245, 145-152. 923 

Fischer-Gödde, M., Becker, H., Wombacher, F., 2010. Rhodium, gold and other highly 924 

siderophile element abundances in chondritic meteorites. Geochim. Cosmochim. Acta 74, 925 

356–379. 926 



43 
 

Fischer-Gödde, M., Becker, H., Wombacher, F., 2011. Rhodium, gold and other highly 927 

siderophile elements in orogenic peridotites and peridotite xenoliths. Chem. Geol. 280, 928 

365–383 929 

Fischer-Gӧdde, M., Becker, H., 2012. Osmium isotope and highly siderophile element 930 

constraints on ages and nature of meteoritic components in ancient lunar impact rocks. 931 

Geochim. Cosmochim. Acta 77, 135–156. 932 

Fitoussi, C., Bourdon, B., 2012. Silicon isotope evidence against an enstatite chondrite Earth. 933 

Science 335, 1477-1480. 934 

Foley, C.N., Wadhwa, M., Borg, L.E., Janney, P.E., Hines, R., Grove, T.L., 2005. The early 935 

differentiation history of Mars from 182W-142Nd isotope systematics in the SNC meteorites. 936 

Geochim. Cosmochim. Acta 69, 4557-4571. 937 

Franchi, I.A., Wright, I.P., Sexton, A.S., Pillinger, C.T., 1999. The oxygen-isotopic composition 938 

of Earth and Mars. Meteorit. Planet. Sci. 34, 657-661.  939 

Frost, D.J., McCammon, C.A., 2008. The redox state of Earth’s mantle. Ann. Rev. Earth Planet. 940 

Sci. 36, 389–420. 941 

Goes, S., Spakman, W., Bijwaard, H., 1999. A lower mantle source for central European 942 

volcanism. Science 286, 1928–1931. 943 

Goldschmidt, V.J., 1937. The principles of distribution of chemical elements in minerals and 944 

rocks.  Journ. Chem. Soc., 655-673 945 

Gomes, R., Levinson, H. F., Tsiganis, K., Morbidelli, A., 2005. Origin of the cataclysmic late 946 

heavy bombardment period of the terrestrial planets. Nature 435, 466-470. 947 

Halliday, A.N., 2004. Mixing, volatile loss and compositional change during impact-driven 948 

accretion of the Earth. Nature 427, 505-509. 949 



44 
 

Halliday, A.N., 2008. A young Moon-forming giant impact at 70-110 million years accompanied 950 

by late-stage mixing, core formation and degassing of the Earth. Phil. Trans. Roy. Soc. 951 

London, Series A, 366, 4163-4181. 952 

Herwartz, D., Pack, A., Friedrichs, B., Bischoff, A., 2014. Identification of the giant impactor 953 

Theia in lunar rocks. Science 344, 1146-1150.  954 

Hidaka, H., Ohta, Y., Yoneda, S., 2003, Nucleosynthetic components of the early solar system 955 

inferred from Ba isotopic compositions in carbonaceous chondrites. Earth Planet. Sci. Lett. 956 

214, 455–466.  957 

Hillgren, V.J., Drake, M.J., Rubie, D.C., 1994. High pressure and temperature experiments on 958 

core-mantle segregation in the accreting Earth. Science 264, 1442–1445. 959 

Hirt, B., Herr, W., Hoffmeister W., 1963. Age determinations by the rhenium-osmium method. 960 

In: Radioactive Dating. International Atomic Energy Agency, Vienna, 35-43. 961 

Hofmann, A.W., 2003, Sampling mantle heterogeneity through oceanic basalts: isotopes and 962 

trace elements. 61-101. In: Carlson R.W. (ed.) Treatise on Geochemistry, vol. 2: The 963 

Mantle and Core, pp. 547–568. Elsevier, Oxford. 964 

Holzheid, A., Sylvester, P., O’Neill, H. St.C., Rubie, D.C., Palme, H., 2000. Evidence for a late 965 

chondritic veneer in the Earth’s mantle from high-pressure partitioning of palladium and 966 

platinum. Nature 406, 396-399.  967 

Horan, M.F., Walker, R.J., Morgan, J.W., Grossman, J.N., Rubin, A., 2003. Highly siderophile 968 

elements in chondrites. Chem. Geol. 196, 5-20. 969 

Horan, M.F., Alexander, C.M.O’D.,Walker, R.J., 2009. Highly siderophile element evidence for 970 

early solar system processes in components from ordinary chondrites. Geochim. 971 

Cosmochim. Acta 73, 6984-6997. 972 



45 
 

Hurwitz, D.M., Kring, D.A., 2014. Differentiation of the South Pole-Aitken basin impact melt 973 

sheet: Implications for lunar exploration. Journ. Geophys. Res.-Planets 119, 1110-1133. 974 

Javoy, M., et al. (2010) The Chemical composition of the Earth: enstatite chondrite model, Earth 975 

Planetary Sci. Lett., 293, 259–268. 976 

Kimura, K., Lewis, R.S., Anders, E., 1974. Distribution of gold and rhenium between nickel-iron 977 

and silicate melts. Geochim. Cosmochim. Acta 38, 683-781. 978 

Kokuba, E., Uda, S. 1998. Oligarchic growth of protoplanets. Icarus 131, 171-178. 979 

Kortev, R., 1994. Compositional variation in Apollo 16 impact-melt breccias and inferences for 980 

the geology and bombardment history of the Central Highlands of the Moon. Geochim. 981 

Cosmochim. Acta 58, 3931-3969. 982 

Kramers, J.D., 1998. Reconciling siderophile element data in the Earth and Moon, W isotopes 983 

and the upper lunar age limit in a simple model of homogeneous accretion. Chem. Geol. 145, 984 

461-478. 985 

Kring, D.A., Cohen, B.A., 2002. Cataclysmic bombardment throughout the inner solar system 986 

3.9–4.0 Ga. Journ. Geophys. Res. 107, 10.1029/2001JE001529. 987 

Le Roux, V., Bodinier, J.-L., Tommasi, A., Alard, O., Dautria, J.-M., Vauchez, A., Riches, 988 

A.J.V., 2007. The Lherz spinel lherzolite: Refertilized rather than pristine mantle. Earth 989 

Planet. Sci. Lett. 259, 599-612. 990 

Li, J., Agee, C.B., 1996. Geochemistry of mantle-core differentiation at high pressure. Nature 991 

381, 686-689. 992 

Li, J., Agee, C.B., 1996. The effect of pressure, temperature, oxygen fugacity and composition 993 

on partitioning of nickel and cobalt between liquid Fe-Ni-S alloy and liquid silicate: 994 

implications for the Earth’s core formation. Geochim. Cosmochim. Acta 65, 1821–1832. 995 



46 
 

Liu, J., Sharp, M., Ash R.D., Kring, D.A., Walker, R.J. Diverse impactors in Apollo 15 and 16 996 

impact melt rocks: evidence from osmium isotopes and highly siderophile elements. 997 

Geochim. Cosmochim. Acta, in press.  998 

Lorand, J.-P., Alard, O., Godard, M., 2009. Platinum-group element signature of the primitive 999 

mantle rejuvenated by melt-rock reactions: evidence from Sumail peridotites (Oman 1000 

Ophiolite). Terra Nova 21, 35–40. 1001 

Lyons, J.R., Young, E.D., 2005. CO self-shielding as the origin of oxygen isotope anomalies in 1002 

the early solar nebula. Nature 435, 317-320. 1003 

Maier, W.D., Barnes, S.J., Campbell, I.H., Fiorentini, M.L., Peltonen, P., Barnes, S.-J., Smithies, 1004 

R.H., 2009. Progressive mixing of meteoritic veneer into the early Earth’s deep mantle. 1005 

Nature 460, 620-623. 1006 

Mann, U., Frost, D.J., Rubie, D.C., Becker, H., Audétat, A., 2012. Partitioning of Ru, Rh, Pd, Re, 1007 

Ir and Pt between liquid metal and silicate at high pressures and high temperatures - 1008 

implications for the origin of highly siderophile element concentrations in the Earth’s 1009 

mantle. Geochim. Cosmochim. Acta 85, 593-615. 1010 

Marchi, S., Bottke, W.F., Elkins-Tanton, L.T., Bierhaus, M., Wuennemnn, K., Morbidelli, A., 1011 

Kring, D.A., 2014. Widespread mixing and burial of Earth’s Hadean crust by asteroid 1012 

impacts. Nature 511, 578-582. 1013 

Mason, B., Taylor, S.R., 1982. Inclusions in the Allende meteorite. Smithson. Contrib. Earth Sci. 1014 

25, 1–30. 1015 

McDonald, I., Andreoli, M.A.G., Hart, R.J., Tredoux, M., 2001. Platinum-group elements in the 1016 

Morokweng impact structure, South Africa: evidence for the impact of a large ordinary 1017 



47 
 

chondrite projectile at the Jurassic–Cretaceous boundary. Geochim. Cosmochim. Acta 65, 1018 

299–309. 1019 

McDonough, W.F., Sun, S-s., 1995. The composition of the Earth. Chem. Geol. 120, 223-253. 1020 

McDonough, W.F., 2003. Compositional Model for the Earth’s Core. In: Carlson R.W. (ed.) 1021 

Treatise on Geochemistry, vol. 2: The Mantle and Core, pp. 547–568. Elsevier, Oxford. 1022 

McLeod, C.L., Brandon, A.D., Armytage, R.M.G., 2014. Constraints on the formation age and 1023 

evolution of the Moon from 142Nd–143Nd systematics of Apollo 12 basalts. Earth Planet. 1024 

Sci. Lett. 396, 179-189. 1025 

Meisel, T., Walker, R.J., Morgan, J.W., 1996. The osmium isotopic composition of the primitive 1026 

upper mantle, Nature 383, 517-520. 1027 

Meisel, T., Walker, R.J., Irving, A.J., Lorand, J.-P., 2001. Osmium isotopic compositions of 1028 

mantle xenoliths: a global perspective. Geochim. Cosmochim Acta. 65, 1311-1323. 1029 

Morbidelli, A., Petit, J.-M., Gladman, B., Chambers, J., 2001. A plausible cause of the late heavy 1030 

bombardment. Met. Planet. Sci. 36, 371-380. 1031 

Morgan, J.W., Laul, J.C., Kränenbühl, U., Ganapathy, R., Anders, E., 1972. Major impacts on 1032 

the moon: characterization from trace elements in Apollo 12 and 14 samples. Proc. Third 1033 

Lunar Sci. Conf., Supplement 3, Geochim. Cosmochim. Acta 2, 1377-1395. 1034 

Morgan, J.W., Ganapathy, R., Higuchi, H., Kränenbühl, U., Anders, E., 1974. Lunar basins: 1035 

tentative characterization of projectiles, from meteoritic elements in Apollo 17 boulders. 1036 

Proc. Fifth Lunar Sci. Conf., Supplement 5, Geochim. Cosmochim. Acta 2, 1703-1736. 1037 

Morgan, J.W., 1985. Osmium isotope constraints on Earth’s late accretionary history. Nature 1038 

317, 703-705. 1039 



48 
 

Morgan, J.W., 1986. Ultramafic xenoliths: clues to Earth's late accretionary history.  J. Geophys. 1040 

Res. 91, 12,375-12,387.  1041 

Morgan, J.W., Walker, R.J., Brandon, A.D., Horan, M.F., 2001. Siderophile elements in Earth's 1042 

upper mantle and lunar breccias: Data synthesis suggests manifestations of the same late 1043 

influx. Met. Planet. Sci. 36, 1257-1275. 1044 

Mukhopadhyay, S., 2012. Early differentiation and volatile accretion recorded in deep-mantle 1045 

neon and xenon. Nature 486, 101-106. 1046 

Murthy, V., 1991. Early differentiation of the Earth and the problem of mantle siderophile 1047 

elements; a new approach. Science 253, 303-306.  1048 

Nittler, L.R., 2003. Presolar stardust in meteorites:recent advances and scientic frontiers. Earth 1049 

Planet. Sci. Lett. 209, 259-273. 1050 

Nolet, G., Karato, S.-I., Montelli, R., 2006. Plume fluxes from seismic tomography. Earth Planet. 1051 

Sci. Lett. 248, 685-699. 1052 

Norman, M. D., Bennett, V. C., Ryder, G., 2002. Targeting the impactors: siderophile element 1053 

signatures of lunar impact melts from Serenitatis. Earth Planet. Sci. Lett. 202, 217–228. 1054 

Nyquist, L.E., Wiesmann, H., Bansai, B.,Shih, C.-Y., Keith, J.E., Harper, C.L.,1995. 146Sm–1055 

142Nd formation interval for the lunar mantle. Geochim. Cosmochim. Acta 59,2 817–2837. 1056 

Pearson, D.G., Irvine, G.J., Ionov, D.A.., Boyd, F.R., Dreibus, G.E., 2004. Re–Os isotope 1057 

systematics and platinum group element fractionation during mantle melt extraction: a 1058 

study of massif and xenolith peridotite suites. Chem. Geol. 208, 29– 59 1059 

Peucker-Ehrenbrink, B., Jahn, B.-m., 2001. Rhenium-osmium isotope systematics and platinum 1060 

group element concentrations: loess and the upper continental crust. Geochem. Geophys. 1061 

Geosyst. 2, 2001GC000172. 1062 



49 
 

Puchtel, I.S., Brandon, A.D., Humayun, M., 2004a. Precise Pt-Re-Os isotope systematics of the 1063 

mantle from 2.7-Ga komatiites. Earth and Planetary Science Letters 224, 157-174. 1064 

Puchtel, I.S., Humayun, M., Campbell, A., Sproule, R., Lesher, C.M., 2004b. Platinum group 1065 

element geochemistry of komatiites from the Alexo and Pyke Hill areas, Ontario, Canada. 1066 

Geochimica et Cosmochimica Acta 68, 1361-1383. 1067 

Puchtel, I.S., Humayun, M., 2005. Highly siderophile element geochemistry of 187Os-enriched 1068 

2.8 Ga Kostomuksha komatiites, Baltic Shield. Geochim. Cosmochim. Acta 69, 1607-1618. 1069 

Puchtel, I.S., Humayun, M., Walker, R.J., 2007. Os-Pb-Nd isotope and highly siderophile and 1070 

lithophile trace element systematics of komatiitic rocks from the Volotsk suite, SE Baltic 1071 

Shield. Precamb. Res. 158, 119-137. 1072 

Puchtel, I.S., Walker, R.J., James, O.B., Kring, D.A., 2008. Osmium isotope and highly 1073 

siderophile element systematics of lunar impact melt rocks: implications for the late 1074 

accretion history of the Moon and Earth. Geochim. Cosmochim. Acta 72, 3022-3042. 1075 

Puchtel, I.S., Walker, R.J., Anhaeusser, C.R., Gruau, G., 2009a. Re-Os isotope systematics and 1076 

HSE abundances of the 3.5 Ga Schapenburg komatiites, South Africa: Hydrous melting or 1077 

prolonged survival of primordial heterogeneities in the mantle? Chem. Geol. 262, 355-369. 1078 

Puchtel, I.S., Walker, R.J., Brandon, A.D., Nisbet, E.G., 2009b. Pt-Re-Os and Sm-Nd isotope 1079 

and HSE and REE systematics of the 2.7 Ga Belingwe and Abitibi komatiites. Geochim. 1080 

Cosmochim. Acta 73, 6367-6389. 1081 

Puchtel, I.S., Walker, R.J., Touboul, M., Nisbet, E.G., Byerly, G.R., 2014. Insights into early 1082 

Earth from the Pt-Re-Os isotope and highly siderophile element abundance systematics of 1083 

Barberton komatiites. Geochim. Cosmochim. Acta 125, 394-413. 1084 

Raymond, S.N., Quinn, T., Lunine, J.I., 2006. Icarus 183, 265–282. 1085 



50 
 

Regelous, M., Elliott, T., Coath, C.D., 2008. Nickel isotope heterogeneity in the early Solar 1086 

System.  Earth Planet. Sci. Lett. 272, 330–338. 1087 

Rehkämper, M., Halliday, A.N., Fitton, J.G., Lee, D.-C., Wieneke, M., Arndt, N.T., 1999. Ir, Ru, 1088 

Pt, and Pd in basalts and komatiites: New constraints for the geochemical behavior of the 1089 

platinum-group elements in the mantle. Geochim. Cosmochim. Acta 63, 3915–3934