Proc. Nati. Acad. Sci. USA Vol. 77, No. 12, pp. 6973-6977, December 1980 Geophysics Chemical composition of Earth, Venus, and Mercury (planets/solar nebula/element abundances/mantle/core) JOHN W. MORGAN* AND EDWARD ANDERSt *U.S. Geological Survey, National Center, Reston, Virginia 22092; and tEnrico Fermi Institute, University of Chicago, Chicago, Illinois 60637 Contributed by Edward Anders, September 15,1980 ABSTRACT Model compositions of Earth, Venus, and Frct'ion Condensed 0.2 0.4 0.6 0.8 1.0 Mercury are calculated from the premise that planets and K chondrites underwent four identical fractionation processes in -V,Zr, Re the solar nebula. Because elements of similar properties stay -Al 0 1Pt rnefa!5, REEU,Th -AL together in these processes, five constraints suffice to define the 1600 CoTi03 CONDENSATE composition of a planet: mass of the core, abundance of U, and the ratios K/U, Tl/U, and FeO/(FeO + MgO). Complete abun- 140C.ACo2Al~~~~~~srO7MgAI204 _ 1e~lo~ ~ ~ q2iO.MS0 MTA dance tables, and normative mineralogies, are given for all three ~ 2°S'4 Mg S IO3m SILICTE planets. Review of available data shows only a few gross trends ifa.K)AISi3O for the inner planets: FeO decreases with heliocentric distance, Me0tal Silicates MnS whereas volatiles are depleted and refractories are enriched in Cu, Ag,Zn Ga the smaller planets. Ge, Sn, Sb 800 Fe H 0 F, \ FeO+H2 C1,Br, VOLTILES _ I1i It has been known for over a century that the inner planets 'FeHSFeS.H22 SSeTe 600 14GC~~~~~~~~~~~~~~~~MgAFeS differ in density and, hence, in composition (1, 2). But a single 4nG- (r}F1Fe~~zbrBIBJ1ril~~nJ characteristic, such as density, cannot constrain the abundance 4~~~~~0 Iii~~~~~~~~~~~~~ Pb,siI 1 of 83 stable elements without additional assumptions. One oC03 Fe304 Clay minerals MgSQ4 S Fe3t4, Fe2O3 Organic Compont fruitful assumption (3, 4) is that the planets formed by exactly FIG. 1. Condensation of solar gas at 10-4 atm (10.1 Pa). Three the same processes as the chondrites, both being condensates types ofdust condense from a solar gas: an early condensate consisting from the solar nebula that experienced the same few fraction- of refractory minerals, metallic nickel-iron, and magnesium silicates ation processes (5). In chondrites, elements of similar cosmo- (21, 22). On cooling, iron reacts with H20 and H2S to give FeO and FeS. At this point, the mineral assemblage resembles that of the chemical properties are fractionated by similar factors, and they ordinary chondrites and inner planets. Major changes occur below divide into five groups on this basis. If this is also true for planets, 400 K, yielding a carbonaceous chondrite mineral assemblage. then we only require the abundances of five "index ele- ments"-one for each group-in the planet to calculate the metal reacts with H2S and H20 to form FeS (a fourth compo- abundances of all 83 elements. nent) and FeO, which somehow entered the Mg silicates. Compositions based on this approach have been calculated Judging from the chondrites, these primary condensates did for Earth (4), Moon (4, 6), Mars (7, 8), and the eucrite parent not stay together in their original, cosmic proportions but be- body-presumably the asteroid 4 Vesta (refs. 9, 10, see also refs. came mutually fractionated by physical processes in the nebula 11 and 12). All except those for the Earth have gone through (arrows in Fig. 2). The nature of these processes is not known, two iterations. The present paper corrects this oversight and also but likely possibilities are preferential settling of larger or gives tentative model compositions for Venus and Mercury, earlier-condensed grains and preferential sticking of metal although index data for these two planets are still incomplete. grains by ferromagnetic attraction (23, 24). Some part of the Values for noble gases are tabulated but not discussed; they will material also suffered brief remelting (chondrule formation) be taken up in a future paper. shortly before accretion, causing volatile loss and reduction of We must stress that the compositions calculated by this model K EaorlyCiondensote Metal are working hypotheses to be tested and refined, not definitive 1800 - 1sIo,1 Sil iCate truths to be enshrined in handbooks. For the Earth and Moon, Refroctories MgAI2O4 for which data are ample enough to permit numerous tests, it 400 -,rnrtinn nonn * _ appears that the first-generation models (4) are reasonable first Metal + FeNiCo | jmgS103 m approximations although still subject to further improvement ,-U200- SIncote (13-20). The rules for such improvements are quite restrictive; Volatiles (Ml .FeNiCo MgSrO3 if one element is changed, all other elements from that cos- CuGaGeSn NaKMnF mochemical group must be changed by the same factor. 800 FraoctionOtiron * FeNiCo eM Ls ~e~ 1NoKMnFgeS 700 THE MODEL tReactiso with H2s nH20 : Si04_ The 5 Volatiles (H basic assumption is that planetary matter condensed from 500 a cooling solar gas (2, 21, 22). Under conditions of thermody- riRerel/Ing + * FeNiCo m (Mg.Fe 2 SiO4 ;Accretion I-!FeNiC' A d MMQapgFe) SiO_ namic equilibrium, three types of material condense: (i) an A ...E~TSe,i FIf early condensate rich in Ca, Al, and other refractories, (ii) nickel-iron, and (iii) magnesium silicates (Fig. 1). Below --700 FIG. 2. Evolution of planetary matter in solar nebula. Three K, metal and silicates take up volatile elements. In addition, fractionation processes (arrows), not predicted by the condensation sequence, take place on cooling. They effectively double the number of components, from three to six. Presence of moderately (M) and The publication costs of this article were defrayed in part by page highly (H) volatile elements is indicated by light and dark shading. charge payment. This article must therefore be hereby marked "ad- Partially condensed elements are enclosed in parentheses; for con- tveisement" in accordance with 18 U. S. C. §1734 solely to indicate venience, they are assigned to a seventh component that is volatile- this fact. rich and resembles carbonaceous chondrites. 6973 FeNECo-{~~Mg,Fe)2 6974 Geophysics: Morgan and Anders Proc. Natl. Acad. sci. USA 77 (1980) L IBe F LL 1 1 1 H No Mg A I SK t Ar ,1 CHONDRITES K Co Sc Ti V Cr Mn FelCo Ni Cu Zn Go GeJA Sej`B- r Rb Sr Y Zr Nb Mo Tc Ru Rh Pd A Sn Sb Te X Cs,.Bo La Hf Ta W Re Os Ir Pt ETl94 o C3Vo m VENUS 0 PpjB/i- , 5 RuU C /04 /05 /06 /07 /08 /09 1/0 ///11//4/jj/f<j. //6 _ SHERGOTTITES * EARTH :.....Ir_ :. ~~~~Ave iL12°121 Ce Pr Nd Pm Srn Eu Gd Tb Dy Ho Er Tm Yb Lu .- _...................................................................... EUCRITES MARS Th Po U N/o Pu Am Cmn 8k Cf Es fmn MWdA/o Lw / U 3.)m:c.... Vclak.er J Hlighlands Ave Surface - MOON FIG. 3. Elements condensing from a solar gas divide into five major groups on the basis of volatility. These groups stay together, 2 IC3 more or less, in gas/dust fractionations. K ppro FeS to metal. Two additional components are thus made: FIG. 4. K/U ratios of meteorites and planets. Several chondrite classes (30) have K/U ratios below the C1 chondrite value (star), re- remelted metal and remelted silicate (numbers 5 and 6 in Fig. flecting depletion of K during chondrule formation (remelting). Ratios 2). in planetary surface rocks (4,31-33, t) and differentiated meteorites The 19 highly volatile elements condensing below 600 K (30) are even lower. Because K and U do not fractionate readily in present a special problem because they may be only partly igneous processes, these ratios may be representative of the bulk condensed, to an extent that cannot be predicted a priori but planets. 0, K/U; 0, K/Th X 3.6. may be determined empirically from observed abundances. For convenience, these elements may be assigned to a seventh, The first two, in principle, can be estimated directly from volatile-rich component (no. 7 in Fig. 2) resembling some type geophysical data (heat flow, density), with some iteration via of carbonaceous chondrite (4). [Actually, this "group" consists thermal and seismic models. Fe can thus be determined to of three subgroups-volatile metals, major volatiles such as H, better than +10% in all three planets, but U is known to only C, and N, and noble gases-that differ in condensation behavior about +30% in the Earth (28,29) and not at all for Venus and and should really be represented by three different index ele- Mercury; the (Earth-like and Moon-like) values in Table 1 are ments (7, 8).] educated guesses, based on the apparent trend of U content with Sulfur also is hard to constrain. Its cosmic abundance is size (8) and a few scraps of evidence suggesting Earth-like and comparable to that of Fe [5.0 X 105 vs. 8.9 X 105 atoms/106 Si Moon-like composition for these two planets. atoms (4,25)] and so it may not fully condense if less than 5 X The next two index elements, K and TI, can be estimated 105 atoms of Fe' remain after metal loss and FeO formation. from their abundance ratios relative to U in surface rocks. These Hence, we normally base S on the cosmic S/K ratio or on the ratios tend to be nearly constant in a given planet (Fig. 4), re- amount of available Fe, whichever is smaller (4, 6). However, flecting the coherence of these large ions in igneous processes these values may be too low; part of the S volatilized during (3,31).