Humayun M. and Clayton RN

Pergamon

Geochimica et Cosmochimica Acta, Vol. 59, No. 10, pp. 2131-2148, 1995 Copyright 0 1995 Elsevier Science Ltd Printed in the USA. All rights reserved

0016-7037/95 $9.50 + .oo

0016-7037( 95)00132-8

Potassium isotope cosmochemistry: Genetic implications of volatile element depletion

MUNIR HUMAYUN * J and ROBERT N. CLAYTON”*

‘Department of the Geophysical Sciences, University of Chicago, Chicago, IL 60637, USA *Department of Chemistry and the Enrico Fermi Institute, University of Chicago, Chicago, IL 60637, USA

(Received August 4, 1994; accepted in revised form February IO, 1995)

Abstract-We report high precision ( ?0.5%0) potassium isotopic determinations on bulk chondrites, achondrites, and lunar samples, on a separated chondrule, and two CAIs. We find that potassium shows a remarkable isotopic homogeneity in various solar system bodies, even though there are chemical depletions of a factor of about 30, between Cl chondrites and eucrites and lunar samples. Theories that propose the evaporation of volatile elements from initially condensed (Cl chondrite) material to account for such chemical depletions, necessarily imply the existence of large isotopic mass fractionations, e.g., about +40%0 for Earth, +90%0 for eucrites and lunar rocks. Volatile loss of potassium (and by implication Na, Rb, Cs, and other elements of similar volatility) during chondrule formation is also ruled out. The high precision of the data place stringent limits of 52% on the quantity of potassium that could have been lost by partial volatilization. This is not detectable by standard chemical techniques, which can resolve 5-20% changes in the K/La and K/U ratios. The two-component models proposed by Larimer and Anders ( 1967) and by Wtike et al. ( 1984) invoke vaporization of alkalis which is not supported by the potassium isotope results. The chemical depletion of alkalis and other volatiles must have preceded the processes of chondrule, chondrite, and planetary formation, and occurred during the condensation of precursor dust, probably from a hot stage in the solar nebula.

INTRODUCTION

Volatile element depletion is one of the fundamental chemical processes affecting meteoritic and planetary materials. It is a characteristic of the bulk chemical composition of a planet or chondritic body. All planets appear to be depleted to some extent in volatile elements relative to Cl and solar compositions. Some volatile fractionation is also seen between various classes of chondrites, but is smaller than that witnessed by planets (sensu lato), which will be taken to include the Moon and differentiated objects such as the parent bodies of the Howardite-Eucrite-Diogenite (HED) , the Shergottite-Nakh- lite-Chassigny (SNC), and Angra dos Reis (ADOR) meteorites, etc. The extent of the process can be judged from Fig. 1. The processes of the origin of planets and volatile depletion must be intimately associated, but the mechanism of volatile element depletion is not well understood, and this is a fundamental obstacle in understanding the chemical processing of planetary materials.

The depletion of alkali elements in the Earth relative to chondrites was noted by Gast ( 1960) and by Wasserburg et al. ( 1964). The depletion of volatile elements in all planetary materials is now well established (Wolf and Anders, 1980; Wgnke and Dreibus, 1988; Taylor, 1979, 1992; Palme and Boynton, 1993). The determination of the extent of volatile element depletion in planets requires a good knowledge of the bulk composition, which is obscured in individual specimens by internal chemical differentiation. For elements that are highly incompatible during magmatic differentiation, the ratio of a volatile large ion lithophile (LIL) element to a refractory LIL element, e.g., K/La, K/U, or Rb/Sr, provides a measure

* Present address: Department of Terrestrial Magnetism, Carnegie Institution of Washington, 5241 Broad Branch Road, Washington, DC 20015, USA.

of the extent of volatile depletion. This technique was utilized by Wasserburg et al. ( 1964) and W&&e et al. (1972), and the depletions were reviewed by W&&e and Dreibus ( 1988) and Palme and Boynton ( 1993). An accurate assessment of depletion for volatile elements other than incompatible lith- ophiles is difficult. Measurements of “primitive” peridotite nodules by Jagoutz et al. ( 1979) have been used to constrain bulk mantle abundances for a large part of the periodic table. These results show a depletion in all volatile elements, including alkalis, P, Ga, Ge, halogens, and chalcophiles.

Processes of Volatile Element Depletion

Two kinds of mechanism can be proposed to account for the chemical fractionations observed in chondrites and planets: those involving incomplete condensation of volatiles, e.g., from a hot nebula (Larimer and Anders, 1967; Grossman and Larimer, 1974) and those involving the evaporation of Cl- like matter accreted from a cold solar nebula (Ringwood, 1966). The mechanism of volatile element depletion holds the key to whether planetary and chondritic material formed from a hot solar nebula or a cold nebula, and is, therefore, an independent constraint on models of the thermal structure of the nebula derived from calculations of protostellar origin (Ruden and Lin, 1986; Cameron, 1988; Boss, 1993). This also has important consequences for the preservation of isotopic anomalies in meteorites and the interpretation of Rb/Sr and U/Pb chronologies. It was to answer questions concem- ing these that this experiment was undertaken.

Partial condensation of nebular gas

It was recognized by Anders ( 1964) in an early attempt to explain the chemical features of primitive meteorites that these could be understood in terms of the mixing of two com-

2131

2132 M. Humayun and R. N. Clayton

0.01

Zhondrites Planets

m Cl o EH HIL/LL ??EL

I c2

0 SNClMars

0 Earth

Eucrites (EW

d Moon

Angra dos

Reis

FIG. 1. The depletion of potassium as determined from the K/La ratio (planets) or the K/Si ratio (chondrites) normalized to Cl chon- d&es. The K/Si for chondrites was calculated from data of Wasson and Kallemeyn (1988). The K/La ratios for planets and achondrites were taken from Wtike and Dreibus (1988, and references therein), except for the achondrite, Angra dos Reis (Wasserburg et al., 1977 and references therein).

ponents: a Cl-type material with its cosmic complement of volatiles and a material depleted in the volatiles by incomplete condensation. This was developed rigorously by the condensation calculations of Larimer ( 1967) and Grossman ( 1972). To explain the variable depletion of moderately and highly volatile elements in ordinary chondrites, Wai and Wasson ( 1977) proposed that gas-dust separation took place during condensation of the volatiles.

Evaporation of chondritic material

There is a plethora of models for the evaporation of volatiles from planets, many proposing extreme loss of volatiles, and even significant fractionation of the principal constituents of planets, Mg, Fe, and Si (e.g., Cameron and Benz, 1991; Ringwood, 1966, 1989, 1992). Removal of volatiles by evaporation from chondrules and melt droplets has been proposed by Larimer and Anders ( 1967), Wolf and Anders ( 1980), and W&&e et al. ( 1981, 1984). Experimental investigations of the chemical changes taking place during evaporation of Murchison and Allende chondrites by Notsu et al. ( 1978), Hashimoto et al. ( 1979), and Hashimoto (1983) have shown that evaporation can produce chemical fractionations similar to those produced by condensation. Yet, Wulf et al. ( 1995) found that evaporated Allende and Murchison chunks yielded residues which do not chemically resemble in detail the observed fractionations in meteorites under the experimental conditions explored. Chemical effects alone cannot cogently distinguish between the processes of evaporation and condensation. It was essential to develop a method that could separate the effects of evaporative loss from incomplete condensation.

Kinetic vs. Equilibrium Isotope Fractionation

The process of partial evaporation can be distinguished from incomplete condensation by the production in the former of large mass-dependent isotopic fractionation (Rayleigh distillation) in elements that otherwise lack isotopic variations in nature. During evaporation, kinetic isotope effects arise from the faster rate of loss of the lighter isotopes from the surface of a solid grain or liquid droplet. The magnitude of a kinetic isotope effect is proportional to the inverse square root of the masses of the evaporating species, which for potassium gives (Y = 0.9753 or =25%0. As long as the vapor produced is continually removed from the surface, an heavy isotopic enrichment is recorded in the residue (see Discussion, below). Such effects are not produced during condensation, as long as the gas is well-mixed (i.e., no diffusive or gravitational separation of isotopes has taken place), since the rate of uptake of the isotopes at the surface of a condensed phase is controlled by thermodynamic factors, and not by the rate at which gas is supplied to the surface.

Thus, during condensation, isotope separation takes place only if there are significant equilibrium isotope effects for the surface-vapor partitioning, a natural example being the isotope effects in hydrogen and oxygen produced during the condensation of water to form rain and snow. Only equilibrium isotope effects are to be expected for equilibrium condensation or equilibrium evaporation, but neither process can explain the moderately volatile element depletion patterns observed, i.e., nonchondritic K/Rb, Rb/Cs, etc. A limit for the partitioning of potassium isotopes between a condensed phase and a vapor phase was calculated using an extrapolation of the vapor pressure isotope effect of argon. At 83 K, ‘6Ar, and 40Ar in argon vapor are fractionated by 7%0 (Boato et al., 1960) which, extrapolated as l/T* to 1000 K, is only 0.01%0/ a.m.u. This gives an estimate of how small an isotope effect is to be expected for condensation ((w = 0.99998), taking into account the van der Waals bonding in the liquid. Thus, for potassium, the isotope fractionation factor for condensation (=0.02%0) is about three orders of magnitude smaller than the kinetic isotope effect ( x25%0), and is unlikely to produce measurable isotope effects (~0.5%0). Any observed isotopic effects can then be attributed to partial vaporization.

Element selection

The distinction between the processes of partial evaporation and incomplete condensation can be made by determin- ing isotopic mass fractionation in moderately volatile elements. Suitable elements with at least two isotopes must be:

1) relatively light in mass, 2) not subject to isotopic effects due to geological processes,

and 3) preferably lithophile, so that the chemical extent of vola-

tile depletion can be determined unambiguously from volatile/refractory elemental correlations.

These are very restrictive criteria and exclude:

1) the monoisotopic elements F, Na, P, Mn, As, Cs, and Au; 2) the geologically fractionated elements Li, B, Cl, and S;

and

K isotope cosmochemistry 2133

3) the chalcophile elements S, Cu, Zn, Ga, Ge, Se, Ag, Sn, Table 1. Potassium isotopic composition of chondrites.

Sb, and Te. sample N K(ppm) K@g) @‘Kf20m

The remaining moderately volatile elements are K and Rb, with K being chosen as the lighter and more abundant element. Hu- mayun and Clayton ( 1995) show that K is not fractionated by terrestrial geochemical processes, consistent with an equilibrium fractionation factor well below detection limits.

A method was developed to precisely and accurately measure isotopic differences of potassium between samples and a standard (Humayun and Clayton, 1995). Potassium isotopes were measured in various planetary materials to determine whether such materials have distinct isotopic compositions due to mass-dependent isotopic fractionation induced by thermal processing. Such isotopic effects must be correlated with the extent of volatile element depletion determined independently from chemical data. A preliminary account has been given by Humayun and Clayton ( 1993, 1994) and additional details may be found in Humayun ( 1994). The high precisions attained for potassium isotopic composition in the present study place the strictest limit yet on the evaporative loss of volatile elements from solar system bodies. It is difficult to see how this could be better accomplished using any other element, since sulfur is not limited by analytical precision, but by natural variability (Thode et al., 1961; Thode and Rees, 1979; Des Marais, 1983; Sakai et al., 1984; Gao and Thiemens, 1991, 1993a,b). Moreover S, Zn, and Se are present at the low condensation temperature end of the moderately volatile elements, so that a significant fraction of these elements may be contributed by late accreted “veneer,” making S, Zn, and Se poor tracers of the high-temperature history of planetary materials.

Orgueil C 1

Murcbison CM2

Murchison fusion crust

Allende CV3

AL6-S4, All. chondrule

Vigarano CV3 ‘93

Vigarano CV3 ‘92

Vigarano CAI

Leoville CV3 (a)

Leoville CV3 (b)

Leovilk CV3 ‘92

Leoville CAI, L6

Leoville CAL, L6 (corr)

mdarch (E4) Allegan (H5)

Mexo Madams LL3

Klymka LL3

Bishunpur LL3

Semarkona LL3

Kriihenberg (host) LL5

Kriihenberg (vein) LL5

44

14

9

25

26

26

64

21

22

25

30

18

395

563

285

488

216

258

248

104

90

11.7

414

302

189

261

31

160

655

62

51.7

62

457

5.1

20 856 406

20 754 420

29 756 1309

32 822 1306

32 789 981

56 838 1882

23 3605 990

23 11100 1043

+ 0.0 f 0.5 %0

+O.Sf 1.2%0

-0.2f 1.7460

- 0.5 f 0.7 %o

-0.1 f0.7Xo

- 0.2 f 0.6 %o

+ 0.4 f 0.5 %o

+ 0.2 It 0.7 %0

+ 1.5f0.5960

+ 0.8 f 0.6 lo

+ 1.7f0.6960

+ 7.1 f0.8 zo

- 0.3 f 2.8 Wo

+ 0.3 f 0.8 Xo

+ 0.2 f 0.5 %o

- 0.2 f 0.5 %o

- 0.5 + 0.4 lo

+ 0.3 f 0.4 460

+0.1*0.3%o

- 0.4 f 0.5 %o

- 0.5 f 0.5 %c

Weighted mean 532 + 0.20 f 0.14 %o

Average of samples 18 + 0.20 f 0.30 x0

EXPERIMENTAL PROCEDURES AND SAMPLES

Details of the chemical separation and mass spectrometry have been presented elsewhere (Humayun, 1994; Humayun and Clayton, 1995). The analysis of low-K, high-Mg samples poses additional problems for complete recovery and purity of the separated K. These will be discussed here. It was noted that Cr and Al were present in several separated K samples, with concentrations sufficient to produce distinctly different glass compositions. To determine the matrix effect of such impurities on the isotopic composition, standard glasses were prepared doped with 2% Cr and with several percent Al. The Cr standard gave apparently heavy h4’K = +0.9 ? 0.6%0, but most sample glasses had lower concentrations of Cr, and were corrected by linear interpolation of Cr content between doped and undoped standards. These corrections have been noted in Tables 1, 2, and 3, where applied. Two samples, Leoville ‘92 and Kenna ABC, had significantly higher concentrations of Cr, so no corrections were applied. These samples have been italicized in Tables 1 and 2, since the isotopic effects seen in these samples are partly due to matrix effects.

Here, and in other data tables, N= number of ion probe points measured, K @pm)= K concentration in sample, K (pg)= amount of K recovered. Murchison fusion crust and Leoville L-6 have been excluded from the mean. Corrections applied for matrix interferences from Cr: Allende CV3= -0.SztO.5960, Vigarano CV3 ‘92= -0.61tO.4%0, Vigarano CV3 ‘93= -0.4dzO.3960, and L-6 (corr.)= -0.s. No correction was applied to Leoville CV3 ‘92 which had much more Cr than the standard. An additional correction of -6.5% was applied to L-6 for matrix interferences from Al.

Samples

Chondrires

The preparation of doped Al standards was hampered by the low solubility of AlgO in Ba borate glass (giving a mixture of glass and crystalline solids), and so exact concentrations are not known. The measured isotopic effects are related to the Al+/Ba+ signal by the relation:

Potassium was obtained from seven chondrites (Orgueil, Vigarano ‘92, Leoville ‘92, Mezo Madaras, Krymka, Bishunpur, Semarkona) dissolved for separation of interstellar diamond and Sic by Huss ( 1989). with high-purity NIST acids and H,BO, substituted for com- mercial reagents to minimize blanks. Chemical analyses are in rea- sonable agreement with literature values. The low K value for Orgueil is due to dilution by a CaSO., vein, and Ca contents were 40% higher than the abundances of Anders and Grevesse ( 1989). Other chondrites (Murchison, Allende, Vigarano ‘93, Leoville ‘93, Indarch, Al- legan, Krahenberg) were dissolved in Teflon bombs. The two Kra- henberg fractions were obtained from the study of Wlotzka et al. ( 1983) to investigate the effects of veining. Chips of fusion crust were obtained from a =4 cm stone of Murchison.

a = 5.6 + 2.6 x 10m7 (z7Al+/‘~6Ba’)ramplc %0.

Application of this to various glasses that show larger than blank Al+/Ba+ signals gave a correction of less than 0.4%~ except for two impact melt coatings on Apollo 16 boulders (Humayun and Clayton, 1995), lbitira, and L-6, the latter being out of the range of the standards.

An aliquot of powder from a single large Allende chondrule, AL6- S4, was obtained from the oxygen isotopic study of Clayton et al. ( 1977). Vigarano CA1 consisted of a0.4 CA1 and 0.6 adhering matrix. A compact type A inclusion weighing = 1 g (L-6) was separated from Leoville, and a chip of L-6 was further cleaned with a dental burr to remove adhering matrix and surface alteration. A thin-section made from L-6 indicated no visible alkali alteration, and electron microprobe measurements of Na show no “hot-spots” (S. B. Simon, pers. commun., 1993). This sample has K = 11.7 ? 0.2 ppm, on the


low end of K contents of CAIs reported by Sylvester et al. ( 1993). and similar to a compact type A from Efremovka (Ef 1)

Achondrites

Two SNC meteorites were analyzed, Shergotty and Zagami. Both these samples were irradiated in 1977 with a neutron fluence of 3 X 10” n/cm’, along with BCR-I (mad.). No deleterious effects of the neutron exposure are expected, given the low fluence and the capture cross-sections of potassium and calcium isotopes. A 2.2 g sample of Juvinas (eucrite) was processed carefully through the large column and this was followed by a second cleanup step, also on the large column. A very small amount of powdered Ibitira (57 mg) was available and this was processed through the small column and had significant impurities, including Cr and Al. Ibitira is a vesicular basalt with the lowest K content of any of the eucrites; it was run for comparison with lunar vesicular basalts. The result for Ibitira is considered to be an upper limit only, given the complications. The samples of Kenna (ureilite) were from the oxygen isotope study of Clayton et al. ( 1976). During mineral separation of Kenna, a large amount of magnetically separable material was obtained but not consumed for oxygen isotope analysis. A small quantity of fine powder was used as a “whole-rock” fraction. We combined the magnetic fractions to obtain Kenna ABC and analyzed this and the WR fraction separately. Kenna WR was run on the small ion exchange column, while Kenna ABC ( ~0.7 g) was separated on the large column. PCC- 1 was run for comparison since this peridotite has an almost identical potassium content, but Mg was found in sufficient amount in the separated K fraction of PCC-I to require a second cleanup step on the small column. Thus, PCC-I and Kenna ABC were not handled identically in the chemical procedure. The higher purity of PCC-I resulting from the second cleanup is to a large part responsible for the better accuracy. The two Kenna fractions contained significant Cr. compromising the accuracy of the results.

Lunar samples

Samples were chosen so as to represent the entire spectrum of lunar reservoirs recognized in samples returned by the Apollo missions. Mare basal& orange soil (74220), and three pristine highlands rocks were analyzed as representative of all major lunar reservoirs. An analysis of vesicular mare basalt 15556.63 was isotopically heavy compared with microgabbro, I5495 ( Humayun and Clayton, 1993 ) . This sample was reanalyzed in duplicate, with chips selected from the interior and exterior of the recovered sample, 15556,O. Other vesicular and nonvesicular samples were analyzed to determine whether loss of alkalis from lunar basahs was important. A 1.6 g sample of ferroan anorthosite, 60015,742, was analyzed. The glassy coating on this breccia, 60015,747, and on two other Apollo 16 anorthosite breccias were also analyzed, but most of the potassium in these samples is from the ubiquitous KREEP, and may contain contributions of meteoritic and lunar soil potassium. These results were reported by Humayun and Clayton ( 1995). but have not been included in the lunar database, even though no isotopic fractionations have been found. There are very few examples of pristine crystalline rocks that match the description of KREEP and 15386 is one such candidate (Dowty et al., 1976; Warren et al., 1978). Some of the samples had Cr and Al contamination, and were cleaned up by a second ion exchange step on the small column, while corrections were applied to the remainder.

RESULTS

Potassium Isotopic Composition of Terrestrial Materials

Isotopic results and chemical compositions of the analyzed terrestrial samples have been presented elsewhere (Humayun and Clayton, 1995). An average isotopic composition of b4’K = +0.28 2 0.21%0 (2a,) was obtained. The weighted mean of all seventeen terrestrial samples is isotopically distinct from the reagent Suprapure KN03 at the 20, level and, because

there is no intrinsic reason to assume that these are indeed equal, we compare fractionations of other natural samples against the terrestrial mean by defining

A“‘K = (6“‘K,, - S4’K-) z ha; + 30&,,,

where A represents differences between two 6 values, and errors are given as 30 to allow only those differences to appear significant which are larger than the 99% confidence level. This does not significantly affect any of the results or conclusions presented here, but is carried out for exac- titude.

Potassium Isotopic Composition of Chondrites

Chondrite analyses are given in Table 1 and shown in Fig. 2. The chondrites appear to be isotopically homoge- neous, and the standard error of the mean of eighteen samples ( +-0.30%0) is within a factor of two of the weighted

Chondrites

Chondritic Average

Orgueil Cl

Murchison CM2

Murchison fusion crust

Allende CV3

AL6-S4 chondrule

Vigarano CV3 ‘93

Vigarano CV3 ‘92

Vigarano CAI

Leoville CV3 (a)

Leoville CV3 (b)

Leoville CV3 ‘92

lndarch EH4

Allegan H5

Mezij Madaras LL3

Krymka LL3

Bishunpur LL3

Bemarkona LL3

Krahenberg (host) LL5

Krahenberg (vein) LL5

-4 -2 0 2 4

S4’K (%e) f 2 (T

FIG. 2. Potassium isotopic composition of chondrites relative to SuprapuP KNO,. The data for Murchison fusion crust were not used in calculating the mean value.


mass spectrometric errors (~llC[llaf] = ?0.14%0). There is no systematic offset between any of the classes of chondrites. If the somewhat discrepant Leoville analyses are ig- nored, the standard error of the mean becomes t0.19%0, comparable to the weighted mass spectrometric errors, +-0.14%0. A sample of Murchison fusion crust was analyzed to see if fusion crusts had any isotopic effects in potassium, and none was found. This is hardly surprising, since melt formed at the surface is ablated by the atmosphere almost immediately.

It was noted from the first set of chondrite measurements (1992) that the two CV3 chondrites, Leoville and Vigar- ano, appear to be isotopically heavy by about 1%0, and this was investigated in more detail in the next set of analyses. Replicate measurements of Leoville and Vigarano were carried out and the set of carbonaceous chondrites analyzed was expanded to include Allende CV3 and Murchison CM2. Isotopic results from Table 1 show that Vigarano ‘93, Allende WR, and Murchison WR samples are unfractionated, and correction of Vigarano ‘92 data for Cr removes the entire isotope effect. The correction for Leoville ‘92 is much larger, but is beyond the range of prepared standards, and was not attempted (but would remove most of the + 1.7%0 effect). The new analyses of Leoville still indicate high S4’K, even though Leoville (a) was cleaned up for Cr impurities. It is not yet possible to dismiss analytical arti- facts completely.

An effort was made to analyze individual components of the CV3 chondrites, since CAIs are known to carry most of the isotopic anomalies and severe mass fractionations (e.g., FUN inclusions) reported in meteorites (Clayton et al., 1988; Wasserburg, 1985 ) . The Allende chondrule and Vig CA1 are isotopically normal. The small quantity of potassium available from L-6 made this a very difficult measurement, and traces of Al (well below detection limits of the AAS) in the eluted K fraction exceeded the amount of K by an order of magnitude. This was recognized only after the ion probe mounting, by which time little could be done to rectify the problem. A large isotope effect was measured (Table 1 ), but a significant, if not the entire, amount is attributable to the difference in matrix composition between the standard and L-6 mounts. Figure 3 shows the L-6 result to be collinear with two Al- doped Ba borate glasses, but beyond the range in which the controls apply. Extrapolation gives a correction of -6.5 5 2.6%0, but the result should be viewed with caution. At present, it is not justifiable to claim any isotopic effect, but an effect of the order of (or smaller than) the difference between L-6 and the highest Al standard cannot be precluded by the present analysis. This is unfortunate since this is the same magnitude of effect that would be expected for L-6 should live 4’Ca (T,,* = 130,000 years) have been present in the inclusion with 4’Ca/“‘?a = 2-3 X lo-’ (Goswami, un- publ. results presented at the 56th Meteoritical Society Meet- ing, 1993; Srinivasan et al., 1993). Even if the entire +7%0 effect were real, it would not account for the observed a 1%0 effect in bulk Leoville, and would imply =25% evaporative loss of K from the CA1 melt. This is a surprisingly small amount of loss of a volatile element, and is a significant result, given that this potassium does not appear to be of secondary origin.

Potassium Isotopic Composition of Achondrites

Isotopic data are presented in Table 2. Most achondrites have extremely low abundances of potassium with attendant complications in obtaining accurate data. None of the samples analyzed has any large isotopic fractionation effects. The Ib- itira result is at most an upper limit to the amount of fractionation, and indicates 5 10% relative loss of K from the sample during vesiculation. There is no evidence for loss from the entire parent body of the eucrites as evidenced by the unfractionated nature of Juvinas, contrary to expectation for the mechanism proposed by Mittlefehldt ( 1987) for the alkali depletion observed in the eucrites. This is independently supported by the correlation of K with La and other incompatible LIL elements in eucrites observed by Warrke et al. ( 1974), Dreibus et al. ( 1977), and Jochum and Palme ( 1990).

The K/La ratio in the vesicular eucrite Ibitira (40-55) is lower than that of the other eucrites (140; Wanke et al., 1974). Stolper (1977) drew attention to the low potassium content of Ibitira, relative to other noncumulate eucrites, and discussed the possibility of selective loss of alkalis by vesiculation. The measurement of K isotopes gives an upper limit of 5 10% loss of K, insufficient to disturb the correlation of alkalis with refractory LIL. Our aliquot of 57 mg of Ibitira gave a potassium abundance of 160 ppm, larger than the 119 ppm reported by Wanke et al. ( 1974), but in agreement with 160 ppm reported by Boctor et al. ( 1994). Thus, contamination of our sample is not a likely explanation. Given the lack of evidence for alkali loss from vesicular lunar basalts, we conclude that Ibitira is unlikely to have lost potassium during vesiculation, and leave the enigma of the low K/La (but normal Cs/Nb, Jochum and Palme, 1990) to future investigation.

Potassium Isotopic Composition of Lunar Rocks

Isotopic results are given in Table 3 and shown in Fig. 4 for mare and highlands rocks. The Moon is a highly differentiated body (Taylor, 1975), which is believed to be the result of the crystallization of a lunar magma ocean (Warren, 1985). This magma ocean is estimated to have involved the outer a400 km of the Moon, about half the lunar mass, to account for trace element enrichments seen in lunar rocks and estimates of the thickness of the lunar crust (Taylor, 1992). The anorthositic crust is underlain by a cumulate mantle, melting of which provides the mare basalts. As the cumulate layers grew, residual magma enriched in incompatible trace elements formed. This magma is identified with KREEP basalt (Warren and Wasson, 1979).

A later intrusive phase in the highlands involves magmas derived from below the cumulate levels of the Moon (Warren, 1985 ), although Taylor et al. ( 1993 ) argue for a different interpretation, where these magmas are derived from late accreted moonlets that are swept up by the Moon following the giant impact. Regardless of the origin of this particular suite of rocks, the Mg-gabbronorites, e.g., troctolite 76535 (Haskin et al., 1974), do not show the Eu anomalies or Fe/Mg ratios characteristic of other lunar samples and appear to have es- caped the intense lunar differentiation evident in KREEP, ferroan anorthosites, and mare basalts. Both troctolite 76535 and


KREEP basalt 15386 have isotopically unfractionated potassium.

The pristine anorthosite 60015, 742 (a monomict breccia) is anomalous with 64’K = -4%0. A possible explanation is laboratory sources of error, although it should be noted that there was no indication prior to analysis of incomplete ex- traction or other complications. Another possible explanation is the deposition of vapor produced by breccia-forming impacts. Tera et al. (1974) found lunar breccias, particularly pristine, monomict anorthosites ( 15415 “genesis rock” ), to have high abundances of unsupported radiogenic Pb and to define an isochron of 3.9 Ga. The effects of volatilization in Rb are significantly smaller, 120%. If volatilized potassium were to have condensed in anorthositic breccias at 3.9 Ga along with Pb, about 15% of K with b4’K = -25%0 would be required. This is not an unreasonable proposition, but further investigation is required.

A more comprehensive coverage of mare basalts than highlands rocks was necessitated by the first measurements of potassium isotopes on 15556 and 15495, which indicated that vesicular lunar basalts may be isotopically fractionated (Hu- mayun and Clayton, 1993). To test both the genuine nature of the original 15556 measurement and to provide adequate understanding of the processes, a selection of mare rocks involving a variety of textures from highly vesicular samples, 15556 and 150 16, vitrophyric basalt 12009, to gabbroic rocks 15065 and 15555, was made. A high-Ti mare basalt, the py- roclastic orange glass deposit 74220, was also analyzed. Ding et al. ( 1983) found sulfur to be depleted in the interiors of the orange glass spheres by about a factor of four compared with Apollo 17 basalts, and the surface is coated with isotopically light sulfur by recondensation of the vapor. Potassium is

Table 2. Potassium isotopic composition of achondrites.

Sample N K@pm) K(pg) @lKf2%

Ihitira (eucrite) 12 160 9.2 +2.1* l.O%o

Juvinas (eucrite) 55 216 606 + 0.9 f 0.4 %o

Shergotty (SNC) 51 1480 190 + 0.5 f 0.4 %o

Zagami @NC) 28 1600 180 + 0.5 f 0.4 %o

Kenna ABC (ureilite) 13 9 6.3 + 1.2 f I.0 x0

Kenna WR (ureilite) 12 41 3.4 +2.1*0.7%0

Corrections applied for matrix interferences from Al: Ihitira= -1.7kO.7%00; and from Cr: Ihitira= -0.4iO.3%0. Kenna ABC has an order of magnitude more Cr than the Cr standard, and so no correction was applied, but it is at least 1%0 lighter.

not depleted in the bulk glass, however, probably because the volatiles coating the glass were not outgassed during the free flight of the glass spheres, but rather prior to eruption in the magma chamber. This is consistent with the 634S = +0.7%0 of the orange glass interiors, comparable to other mare basalts.

All the lunar samples (excluding 60015, 742) are isotopi- tally indistinguishable within errors. The mean and standard error of the mean of eleven lunar rocks is &“K = 0.50 ? 0.29%0 (20,) and A4’K = +0.22 ? 0.54%0 (3a,). The quality of measurement is lower for 15016 and 12009, which were analyzed with fewer points (II = 13). Excluding these samples, the remaining lunar rocks give &z,, ( +-O.l8%0,2u, for n = 9) comparable to the mass spectrometric errors (v?/Z[l/af] = ?0.19%0).

6

0

?? Lunar samples

+ Meteoritic samples

0 Al-standards

-I

0 5 10 15

27A,+/136ea+

FIG. 3. Matrix effect of Al on Z?‘K. The glassy impact coatings on anorthosite boulders, 60015,747 and 61016,418, are the two lunar samples with Al contamination. Other meteoritic and lunar samples are clustered at the origin. Ihitira has a heavy isotopic composition even after adjustment for matrix effects from Al and Cr.


Table 3. Potassium isotopic composition of lunar rocks.

Sample N K(ppm) K@g) S4’K*2~rn

Lunar IIlghlandsz

15386 KREEP 20 5045 236 + 0.8 + 0.6 %o

60015,742 Anorthosite 6 46.3 76.8 - 3.9 f 0.9 %o

76535 Mg-gabbronorite 20 232 331 + 0.4 * 0.5 %0

enough to form glassy textures. The amount of loss can be calculated by knowing the vapor pressure of K above lunar mare basalt at the appropriate temperatures (De Maria et al., 1971; Gooding and Muenow, 1976), and the cooling rate. Gooding and Muenow ( 1976) concluded that the vapor pressures measured by them were too low to allow for significant losses of alkalis. Our measurements are consistent with this.

DISCUSSION

Lunar Mare: Potassium Isotope Fractionation of Planetary Reservoirs

12009 vitrophyric basalt 13

15016 vesicular basalt 13

15065 (a) microgabbro 29

15065 @) microgabbro 21

15495 gabbro 41

15555 microgabbro 28

15556.63 (1992)* 14

15556,191 interior 29

15556,190 exterior 28

74220 orange soil 14

453

346

441

752

298

544

371

378

618

496 +l.lf0.8%0

323 + 1.5 * 0.9 %o

394 + 0.2 * 0.7 %a

72 + 0.3 f 0.7 700

707 + 0.3 * 0.5 %o

307 + 0.3 f 0.6 %o

231 t 3.3f1.0%0

424 +0.1+0.7%0

392 + 0.7 f 0.7 %o

868 - 0.2 * 0.9 lo

The mean isotopic compositions of the various planetary reservoirs are shown in Fig. 5, and compared with the isotopic composition that may be calculated by assuming Rayleigh distillation of a chondritic source. The justification for the use of Rayleigh fractionation is that the vaporization process must involve the removal of vapor as soon as it is produced, further details of which will be considered below. The large chemical

Lunar samples

Weighted mean

Average of samples

256 + 0.46 f 0.19 %o

11 + 0.50 + 0.29 %o 4

* First measurement of 15556, superceded by duplicates below. This sample and anorthosite 60015 have been excluded from the mean. Corrections applied for matrix interference from Cr: 15065(a)= -0.7M.5%0, 15556, 191= -0.7kO.5%0.

Alkali degassing in hard vacuum

These isotopic results do not support the notion that mare basalts outgassed significant quantities of alkalis and other volatiles into the lunar vacuum (O’Hara et al., 1970; Big- gar et al., 1971; Storey, 1973). This hypothesis, in its most extreme form as proposed by O’Hara et al. (1970), has been refuted by geochemical studies of lunar samples from subsequent Apollo missions that include a significant number of plutonic rocks never exposed to hard vacuum at the lunar surface. Since these samples have K contents that correlate with La and U (Wanke et al., 1972, 1973, 1974), then to the extent of this correlation, volatilization at the lunar surface can be excluded. Our analyses of vesicular lunar mare basalts can sensitively detect even a few percent relative losses of K into the lunar vacuum, which would not be obvious from chemical analyses alone. A limit of s 1% loss of K is indicated by the experiments on 15556. The lower precision of 15016 sets a limit of ~7% (using bulk Moon = +0.5%0 relative to laboratory KNOX standard ) .

The amount of K lost from samples has been investigated experimentally by a number of workers. Experiments carried out by O’Hara et al. ( 1970) involved steadily heating the samples for fourteen days in vacuum; those of Storey ( 1973) involved heating for several hours. Under these experimental conditions, it is not surprising that large losses of K took place. Natural basalts, however, cool rapidly

Lunar Average

Mare:

12009

15016

15065 (a)

15065 (b)

15495

15555

15556,63

15556,191

15556,190

74220

Highlands:

15386 KREEP

60015,742 Anorthosite

76535 Norite

Impact melts:

60015,747

61016,418

64435,297

1

--I

-4 -2 0 2 4

S4’K (k) f 2 o

FIG. 4. Potassium isotopic composition of lunar samples relative to SuprapuP KNOB. Open symbols were excluded from the mean.


25

0 0.2 0.4 0.6 0.6 1

Fraction of potassium remaining

0 0.2 0.4 0.6 0.8 1

FIG. 5. A comparison of calculated and measured potassium isotopic compositions of bulk planets and chondrites. Top panel shows calculated fractionation assuming a Rayleigh distillation with a - 1 = -0.0247 and a Cl starting composition compared with measured b4’K. The fraction of potassium remaining is determined by the depletion calculated as in Fig. 1. The bottom panel is an expanded scale version of the experimental data with 30, error bars normalized to terrestrial isotopic composition. The ureilite data are only upper limits, as discussed in the text.

)

depletion of the volatile elements in general, and of K in par- titular, if accomplished by a vaporization process, must leave an isotopic signature that is a function of the extent of the process. It can be seen from Fig. 5 that there is no evidence of an increase in mean h4’K with increasing depletion of K from Cl chondrites to the Moon and eucrites, a factor of about 30 decrease in K abundance. In terms of the relative loss of K from the planetary body, the 3a errors allow for no more than l-2% vaporization losses. In Fig. 6, this is contrasted

with the ~15% chemical loss of K in the Apollo 14 lunar soils with heavy b4’K of = IO%0 (Humayun and Clayton, 1995).

Ruyleigh fractionation

Consider the escape of K atoms from either the surface of a vaporizing object (e.g., a molten sphere) or from a hot vapor atmosphere. If the surface has a temperature T, then molecules

0 vaporized pUmens

?? 14163 Lunarsoil

+ 14259Lunarsoil

0.2 0.4 0.6 0.6 i

Fraction of potassium remaining

FIG. 6. A comparison of the calculated planetary compositions and observed effects in the lunar soils (for details, see Humayun and Clayton, 1995). The soils form a linear array above the Rayleigh curve, as would planets formed by two- component mixing of a partially devolatilized component and a chondritic component (Wtike et al., 1984). and as would bulk chondrites formed by mixing partially devolatilized chondrules with matrix (Ganapathy and Anders, 1974; Wolf and Anders, 1980).


leaving the surface have average energy 3kTl2, which must equal the kinetic energy of the molecules mu2/2. For two isotopic species:

1 1 -m,v, 2__ -

u1 m2v~~-==

m2

2 2 -.

v2 $ ml

The ratio of the fluxes (Ji ) can be written as:

JI dN,ldt N,v, dN, N,v, N, -= -=-*-=-=- Jz dN21dt N2v2 dN2 N2v2 N2

where Ni is the molar concentration of the heavy isotope, N2 is the molar concentration of the light isotope, u, and ua are the mean molecular velocities, and ml and m2 are the masses. It can be seen that the ratio of fluxes differs from unity by the inverse square root of the masses, and this having been thus specified, the effect of continued loss from the reservoir can be calculated by integrating the fluxes to give the well-known Rayleigh fractionation equation:

(0-l)

where F is the fraction remaining and is actually an approx- imation valid for small N, , and a = ( m2/ml ) u2 is the isotopic fractionation factor, the inverse square root of the ratio of the masses of the escaping species. For K atoms, (Y = 0.9753 or =25%00, and F is determined from the ratio of K to a refractory element such as Ba, La, or U, with which it is well correlated during geochemical differentiation, compared with the same ratio in Cl chondrites or whatever else is regarded as the original composition. Although this is a good description of single objects undergoing vaporization, it is not adequate for discussing the compositions of composite bodies, e.g., chondrites consisting of CAIs, chondrules, and volatile-rich matrix; planets accreted from volatile depleted material and chondritic material, etc. For such objects, the effect of mixing fractionated and unfractionated materials must be taken into account.

Effect of mixing

The lunar soil is a useful analog for planetary accretion, in that the material accreted to planets would have included high-velocity bodies that were subjected to impact vaporization and low-velocity bodies that are not vaporized during accretion, and are the principal sources of volatile elements ( Ahrens, 1990). Thus, if devolatilization can be conceived of as taking place during accretion, or if devolatilized and volatile-rich bodies are both accreted, the chemical composition of a growing planet can be written as a mixture of these components. Given a vaporization process that produces isotopic fractionation of potassium, the isotopic composition of the growing planet can be written as a mixture of fractionated material with unfractionated material (which for simplicity will be taken as the original starting composition for planetary material, i.e., Cl chondrites). Such a mixture has the following remarkable property: for a significant degree of vaporization the isotopic composition of the fractionated component is exponentially related to the chemical loss (Figs. 5,6), while

mixing linearly reduces both the isotopic and chemical effects. The net result of mixing is to produce compositions along a line which lies above the Rayleigh fractionation curve, joining the pure fractionated endmember and the origin (Figs. 5, 6). For all exponential functions of loss, the effect of mixing is to underestimate the isotopic composition of K.

Value of the fractionation factor, (Y

Could a calculation based on Rayleigh fractionation over- estimate the isotopic effect? Since it is almost impossible to escape the Rayleigh formulation (because each increment of vapor must be removed from the system to space), either the chemical loss (F) or the fractionation factor ( QI = 0.975) may not be accurately prescribed. The chemical loss is well established. The choice of (r used here, though appropriate, is cer- tainly an upper limit, the square root of the masses of the atomic species of K. Results summarized by Brewer (1953) indicate that alkalis evaporate predominantly as the elements, and at the boiling point for potassium oxide, K20/K = 10e4. Most plausible molecular species are likely to be within a factor of 2 or 3 (e.g., KO, KOH, K20, KSi03, etc.) of the mass of the atomic species, and are extremely low in abundance at the high temperatures required for escape from planetesimals. A value of (Y = 0.9998 would be sufficient to prevent detectable b4’K enrichments. If taken to be the square root of the isotopic ratio of a massive species, this value of (Y implies a mass of =5000 a.m.u., compared with the molecular weight of phlogopite = 420 a.m.u.

A more effective means of diminishing the value of a is by back reaction or diffusion of vapor, which does not apply to escape from atmospheres, but might apply to evaporation from a hot surface. In carrying out the Rayleigh calculations, it is assumed that the vapor produced never reacts with the surface once produced, i.e., every increment of vapor formed is permanently lost from the system. For a system where back reaction takes place at a fractional rate k (0 5 k 5 1) , the net flux (J) from the surface may be described as:

J, = N,u,(l - k,) and J2 = N2u2( 1 - k2),

where N is the molar concentration, u is average velocity, and k specifies the relative back reaction rate (ratio of back reaction to forward reaction rate) for the isotopic species. The ratio of the fluxes is:

JI NIV, (1 -k,) -=--- 52 N2u2(1 -k2)

= R~ (1 - k,) (1 - k2) = Ra’7

where (Y’ = 4’K/39K ( vapor)/4’K/39K (phase) is the effective fractionation factor, and (Y = ( m2/mI ) 1’2. There are two spe- cial limiting cases: when k, = k2 = 0, (Y’ = a is the free (Langmuir) evaporation case, and when k, = k2 = 1 the dif- ferential isotopic flux is zero, i.e., equilibrium (Knudsen) evaporation. For all values of 0 < k < 1 the effective fractionation factor gives rise to an isotopic effect that is an exponential function of the chemical loss. In the limit where the kinetic fractionation vanishes, a small equilibrium fractionation factor will control the isotopic fractionation. We have estimated this factor as (Y = 0.99998 from the vapor pressure of argon. The b4’K data establish a limit to the largest possible


value of (Y 5 0.9998, about an order of magnitude larger than the equilibrium fractionation factor.

The values for k may be obtained from kinetic theory of gases for specified boundary conditions, e.g., volatilization of K from chondrules in a gas of solar composition. For this case, using solar abundances (Anders and Grevesse, 1989) P,IP,,, = 2NKINH = 3 X 10 -‘, while the vapor pressure of K (Good- ing and Muenow, 1977) over chondrule melt compositions (T = 1000 K) is Pvap = 10e9 atm. For nebular pressures < lo-* atm, the evaporation rate of K from the surface of the chondrule will exceed the back reaction rate, such that at ca- nonical P = 10e6 atm this will be by a factor of 104. If the nebular gas is enriched in volatile metals by a similar factor, only then will the condensation of K to the surface equal or exceed the rate of loss from the surface, at which point no isotopic effects are likely, but at the expense of no chemical fractionations, either.

This illustrates the power of the isotopic experiment: it is virtually impossible to lose significant quantities of K and other volatile elements by vaporization without producing detectable isotopic fractionations. The result of this experiment is that there are no detectable isotopic fractionations of K isotopes during a process of volatile element fractionation that produced a factor of 30 loss of potassium (Cl chondrites to the Moon). The simple conclusion is that volatile element depletion is not the effect of vaporization of condensed material, but that the chemical fractionations were produced during condensation, such as may have taken place from a hot solar nebula (Grossman and Larimer, 1974).

Site and Processes of Volatile Element Depletion

It is useful to examine the proposed physical mechanisms by which evaporative volatile element depletions could have occurred. Figure 7 shows the processes that take place in sum- mary form as a function of increase in size from submicron grains to planets. There are essentially two size regions with different physics applying in each: dust grains and small clumps in the range < 1 cm, and planetesimals and planets in the range > lo5 cm. In the former, volatilization from a surface is involved and can be studied in the laboratory, but the presence of a gravitational field in the latter gives rise to a vapor atmosphere, loss of which is required in order to leave a permanent chemical imprint in the form of volatile element depletion on the object. The gravitational fields of bodies < 100 km (10’ cm) are weak enough to be overcome by thermal escape at temperatures comparable to the eruption temperatures of basalts ( = 1500 K). Bodies significantly larger than a few hundred kilometres require thermal energies well in excess of the total vaporization of the planetary surface to allow significant mass fluxes out of the gravitational field. These three cases will be discussed separately below.

Nebular processing-chondrules and dust grains

Perhaps the most likely site for evaporative volatile element depletion is the scale of a small grain or a melt droplet. There have been many proposed means of accomplishing this, including loss from heated grains (Hashimoto et al., 1979; Hashimoto, 1983), loss from molten chondrules (Larimer and

Anders, 1967; Ganapathy and Anders, 1974; Wolf and An- ders, 1980), and loss from impact-disrupted planetesimals (Zook, 198 1; Wgnke et al., 198 1, 1984; cf. Anders, 1964, for earlier versions), which is sometimes taken to be identical to chondrule formation (Zook, 1981). The case of a small molten droplet is relatively easy, if the dimensions of the droplet are = 1 cm or smaller. Loss of K from the surface takes place with accompanying isotopic fractionation, while diffusion ho- mogenizes the melt, resulting in Rayleigh fractionation. This has been demonstrated experimentally for evaporation of melts with the compositions of forsterite (Mg, Si, 0), chondrite (Mg, Si, 0), and Fe0 (Fe, 0) by Davis et al. (1990) and Wang et al. ( 1994a,b). Since the physics is the same for the evaporation of K, the results of these experiments are taken to indicate that potassium, too, will fractionate following the Rayleigh law with an exponent that is the inverse square root of the masses.

Chondrules

Any evaporation from molten chondrules would leave an indelible mark on the isotopic composition of bulk chondrites (exclusive of C 1) since:

1) bulk chondrites contain =30-75% chondrules, and 2) bulk chondrites show volatile element depletions to the

extent of =lO-15% for enstatite and ordinary chondrites, 50% for C2 chondrites and 67% for C3 chondrites (Fig. 1). No isotopic fractionation has been found for whole-rock measurements of any type of chondrite (Table 1, Fig. 5, 6) or for the individual large chondrule from Allende (AL6-S4).

It is useful to consider the chemical evidence for volatile loss of alkalis from chondrules. Grossman and Wasson (1983), Grossman (1988), and Wasson (1992) have argued that chondrules do not show large depletions of volatiles (Na, K, S) and so must have cooled fast enough to prevent significant chemical change due to selective volatilization. Sears et al. ( 1991) argue that Group A chondrules, which constitute 35% of the chondrules, from Semarkona LL3.0 are markedly depleted in Na due to evaporation during melting. This would not go undetected in S“‘K, and yet Semarkona shows no bulk isotopic fractionation (Table 1).

Dust grains

Wang et al. (1991, 1993) have shown theoretically and experimentally that the diffusion rate of Mg in crystalline forsterite is low enough that evaporation of the solid restricts the isotopic fractionation to the surface of the grain. It should be noted that this also restricts the chemical fractionation to the same small volume of the grain. It has been argued that evaporation of solids is a mechanism by which significant transfer of mass from solid to gas may take place without much chemical or isotopic fractionation in the bulk sample (Wang et al., 1993). This conclusion is, however, strongly dependent on the dimensions of the solid, which in the case of the Wang et al. ( 1993) experiment was several millimetres in diameter. Isotopic fractionation was restricted to the outer = 10 pm of the solid. This experiment is an example of a limiting case where the solid can be treated as if it were effectively infinite (Wang et al., 1991).


Giant impacts

Collisional aggregation

Goldreich-Ward instability

Log Radius (m&es)

FIG. 7. This figure shows the sizes of solar system materials from the scale of a dust grain to that of the Sun and planets. The materials are named on the right-hand side and the processes responsible for their formation are given on the left-hand side. The asteroids Ceres and Vesta are shown. The dust grains are gradually built into planets by various accretional processes. Thus, any volatile element depletion produced at any stage of the accretional process will be inherited by the planets.

Evaporation of solid grains in the solar nebula involves

dimensions of =O. 1 pm (interstellar grains) to = 10 pm (nebular grains) and rarely exceeds 100 pm (isolated olivines, Steele, 1986). These grains represent exactly the opposite limiting case of sublimation of solids, that in which the grain interior is likely to be perfectly mixed, since JDt > a, the grain radius. If such (interstellar) grains had been subjected to the “evaporation metamorphism” simulated by the experiments of Wang et al. ( 1993), these would show large isotopic effects, even in the bulk grain or in aggregates of bulk grains. Residues surviving from such grains would show Rayleigh fractionation. Had the “evaporation metamorphism” process envisaged by Hashimoto et al. ( 1979) controlled the chemical signatures of meteorites and planets, a detectable isotopic effect in bulk chondritic or planetary matter would be unavoid- able.

Volatile loss from small planetesimuls

Small bodies do not heat much during accretion, leading to the idea that the parent bodies of the primitive chondrites are in the size range of a few kilometres to a few hundred kilometres. Volatile element loss during metamorphism has been ruled out by the results of Takahashi et al. ( 1978) and Wulf et al. (1995). Wanke et al. ( 1981, 1984) proposed that chondritic planetesimals are melted by radioactivity of short-lived radionuclides and then disrupted by impacts into droplets. Evaporation of melt droplets in (essentially) vacuum would guarantee that Component A of WInke et al. ( 198 1, 1984) would be isotopically fractionated, although these later workers may wish to argue that it is completely devoid of potassium. Such recourse to total devolatilization does not account for the planetary K/Rb and Rb/Cs ratios which are a factor of 1.5 and 2 higher than the Cl value, respectively (Mc-

Donough et al., 1992). This requires that nearly equal amounts of alkalis are contributed by the two components (the Cl chondritic component B is present in much smaller amounts of = 10%). There are also practical limitations to removing all the potassium from chondritic matter by evaporation without fractionating Mg and Si isotopes. Thus, the two-component model of W&nke et al. ( 1984) and Wtinke and Dreibus ( 1988) in its present formulation implies several tens of permil enrichment in b4’K, which is incompatible with the results of this study.

Volatile loss from > 100 km bodies

Volatile element loss from a body the size of a planet, if at all possible, must involve first the production of a vapor atmosphere followed by one of three mechanisms by which the atmosphere may be lost. These include ( 1) Jeans escape, (2) hydrodynamic escape, and (3) impact erosion. The source of energy for such heating is provided by accretional energy re- leased during planetesimal collisions, the physics of which has been treated by Melosh ( 1990). The larger the planet, the greater the potential energy gained by an infalling planetesi- ma1 of mass m (AE = mu&,/2), so that the surface temperature of an accreting planet increases with planetary radius (Kaula, 1979). It is important to note that in the simulations of Kaula ( 1979) the surface temperature is at least an order of magnitude lower than the exobase temperature required for atmospheric escape of heavy metals. The matter of interest to the present study is the nature of the chemical and isotopic effects resulting from atmospheric loss.

Jeans escape. In a hot, isothermal atmosphere, with an exponential decrease in pressure with altitude, the pressure eventually falls to a point where the density is so low that molecules suffer few if any mutual collisions. Any molecule


travelling with a velocity comparable to the escape velocity of the planet (uesc = J(2GMIR), where M, R are the mass and distance from the center of the planet) in a direction away from the surface of the planet is likely to escape into space, forever. The molecules have a maxwellian distribution (due to mutual collisions) at the limiting density level in the atmosphere where the gas becomes collisionless, which is re- ferred to as the exobase (base of the exosphere). This distribution is not modified beyond that point (ignoring the radia- tion field), and thus, the velocity distribution is determined by the atmospheric temperature at the exobase, i.e.,

For such escape to be an effective process, Texoba$c must be large so that u,,,,. is comparable to u,, The exobase temperature is shared by all isotopic species of a given molecule, so that the relative rates at which two isotopic species escape is given by,

where J, = Niui is the flux of the ith isotopic component of concentration N moles/litre, and R = N,lN, is the isotopic ratio in the atmosphere. Following Spitzer ( 1952), the escape flux is given by:

where Y = 1.5 ( ucselui )2, vi = d2kT,lmi, T, is temperature at the exobase, R is radius from the center of the planet to the exobase as indicated by subscript e, or to the surface as indicated by subscript 0, and n is the number density of the species concerned at the surface. If the atmosphere is well- mixed below the exobase the conditions for Rayleigh fractionation are satisfied, and the atmospheric isotopic composition will change as a function of time. This process proceeds to the greatest extent for the lightest species.

It should be noted that Rayleigh fractionation has been suc- cessfully applied to the (albeit, nonthermal; Fox and Dal- gamo, 1983) escape of hydrogen and nitrogen from the Mar- tian atmosphere, with an accompanying increase in the D/H ratio of 6000 t 3000%0 (Owen et al., 1988) and in the “N/ 14N ratio of 650%0 recognized by the Viking mission (Nier et al., 1976; McElroy et al., 1976). The excess D/H ratio of the atmosphere is recorded by martian crustal rocks (SNC meteorites) due to atmosphere-lithosphere exchange processes (Watson et al., 1994). This example is an analogue for the production and recording of isotopic effects during escape of volatile element species in a circum-lunar or circum-terrestrial silicate vapor atmosphere.

Hydrodynamic escape. If the flow rate of a light gaseous species out of the atmosphere is large enough to drive a wind (i.e., the flow is supersonic before reaching the exobase), the description of the outflow given above has to be modified to include entrainment of heavier species. Mass fractionation of the entrained species arises from the balance of forces: all

particles are subjected to the same upward drag force, F = (p)odnldt, where (p) is the average momentum imparted per collision, 0 is the cross-section which is relatively constant for different molecules in a neutral gas, and dnldt is the collision rate, and the drag is resisted by the weight of the particle. Thus, heavier particles are less entrained, and this is particularly true of isotopes of an element which have identical cross-sections but different masses. Hunten et al. ( 1987) have derived the equations for mass fractionation of heavy gases in an isothermal, neutral atmosphere. The effect of hydrodynamic escape of hydrogen on noble gas abundances and isotopic compositions in planetary atmospheres has been given by Pepin ( 1991). In the case of an extremely hot silicate vapor atmosphere, the principal light gas would be an oxygen plasma and the atmosphere would have a mean molecular weight of ~8 (Cameron and Benz, 1991). If the Jeans escape rate of this plasma were sufficiently high to drive a wind, the entrainment of species such as potassium is possible, with attendant mass fractionation.

Impact erosion. Impactors with dimensions comparable to or larger than the scale height (H = kT/mg, is the height over which the atmospheric pressure falls by 1 le) of a silicate vapor atmosphere can erode amounts of gas comparable to the mass of atmosphere traversed by the impactor (Walker, 1986; Melosh and Vickery, 1989). To efficiently lose the transient vapor atmosphere produced by a large impact, it must be followed by more large impacts (enough to cover the entire surface of the planet) on a timescale short compared to the condensation timescale of the atmosphere. This is a restrictive criterion. The efficiency of impact erosion must scale with accretion rate and with increasing planetary size, making the large depletions in achondrites relative to the Earth difficult to account for.

Chemical and isotopic consequences of atmospheric loss

It was observed by Humayun and Clayton ( 1993) that the loss of a vapor atmosphere would be a strongly mass dependent process, so that the chemical effects would involve light species/heavy species depletion, rather than the thermody- namically controlled volatile fractionations observed. This can be shown to be the case for noble gas loss from planetary atmospheres (Pepin, 199 1 ), which involves the preferential depletion of light noble gases, e.g., He, Ne, and Ar, relative to Kr and Xe. There is, likewise, a large accompanying isotopic fractionation (Hunten et al., 1987 ) which may have enriched the 38Ar/36Ar ratio in the Martian atmosphere. The largest chemical effect on planetary compositions would be severe depletion of the Li/Yb ratio, which has been found to be relatively constant between various planetary reservoirs (Nor- man and Taylor, 1992). Isotopic fractionation of any affected element is expected. No isotopic fractionation has been found in Mg (Esat et al., 1979), Si (Epstein and Taylor, 1970; Molini-Velsko et al., 1986), K (this study), or Ca (Russell et al., 1978), between chondrites and the Earth or Moon. Likewise, sulfur isotopes vary by only a few permil between Canyon Diablo Troilite, various chondrites, iron meteorites, the Earth and Moon (Des Marais, 1983; Gao and Thiemens, 1991, 1993a,b; Sakai et al., 1984), due to internal fractionations.


Cosmochemical Implications

Volatile element depletion is important for understanding the chronology of formation, bulk chemical composition, and the thermal history of the solar nebula and its products, asteroids and planets. The nature of the volatile depletion process (condensation or evaporation) is critical to the interpretation of the chemical and isotopic record preserved in chondrites and bulk planetary compositions. Having considered above the various mechanisms by which vaporization of matter may have taken place and found that each of these, if effective at chemical fractionation, must produce large S4’K variations that correlate with chemical depletion of volatiles (e.g., K/La), it is concluded from Fig. 5 that fractional vaporization did not play a role in developing the chemical record in the meteorites and planets. That is not to say that vaporization did not take place, just that it must have been effective at completely removing all K and other volatile elements from grains. Large degrees of vaporization of planetary material must have taken place during the accretionary stage, but the vapor was retained by planetary gravitational fields, a result that is not surprising. By elimination of partial vaporization, it is concluded here that condensation in the solar nebula was responsible for the volatile element depletion observed in chondrites, achondrites, and planets. The exact nature of the process is difficult to fathom, but some clues are afforded by the pattern of depletion. Significantly, refractory element ratios are unfractionated (Taylor, 1992), but elements more volatile than the major elements (Mg, Si, Fe, etc.) are progressively depleted as a function of condensation temperature. This implies some form of coagulation or accretion process, since the accumulation of condensates into large clumps and ultimately into planetesimals, whether driven by gravity or by other physical properties, will operate once the Mg-silicates and Fe-Ni metal have condensed. To a first order, volatile element depletion is a statement of hot accretion.

Thermal history of matter and oxygen isotopes

Since the interstellar medium from which the solar system formed must have contained abundant dust grains, each from a particular stellar source, if meteorites were little more than composites of presolar grains it would be expected that meteorites should show large isotopic anomalies, reflecting the diversity of nucleosynthetic processes (Burbidge et al., 1957; Trimble, 1975) involved in the chemical makeup of solar system matter. That this record may have been largely erased by “extraordinarily thorough isotopic mixing in the solar system” was appreciated by Reynolds ( 1967) based on the absence of any known isotopic anomalies at the time, excluding extinct radioactivities. He further concluded that “ (t)his mixing is the most important single thing to be said about isotopes.” It has been construed as evidence for a high temperature event in the inner solar system that eradicated all presolar records (Suess, 1965; Podosek, 1978), and taken to imply that the chemical evolution of material ending up in the chondrites was controlled by the cooling of a gas of solar composition from a temperature well above that required to vaporize all initial dust F= 2000 K (Larimer and Anders, 1970; Grossman and Larimer, 1974). Calculations of the collapse

of a molecular cloud core have provided evidence both for and against an initially hot solar nebula depending on the physics used and the boundary conditions assumed (Ruden and Lin, 1986; Cameron, 1988; Boss, 1993).

The discovery of isotopic anomalies in oxygen (the most abundant element in meteorites and planets) by Clayton et al. ( 1973 ), and the subsequent recognition that isotopic anomalies in oxygen are present at all scales including between planetary objects (Clayton, 1993; Clayton and Mayeda, 1975, 1983; Clayton et al., 1977) is difficult to reconcile with an initially hot nebula, but requires the survival of interstellar grains in sufficient abundances to account for detectable isotopic anomalies. Is this consistent with the observed large- scale depletion of volatile elements? With the exception of the Cl chondrites, all primitive meteorites have large numbers of chondrules, which indicate brief localized melting events (T = 1700-2000 K), during which volatile element depletion could have taken place (Larimer and Anders, 1967; Gana- pathy and Anders, 1974). Hashimoto et al. ( 1979) and Hash- imoto ( 1983) proposed that thermal metamorphism of dust in the solar nebula, resulting in evaporation of the more volatile elements, could produce chemical fractionations identical in most respects to those produced by condensation from a cooling gas. Thus, a cold nebular model with an initial dust content inherited from the interstellar medium, accompanied by rapid localized heating events, could produce all the observed volatile depletions while preserving the oxygen isotopic anomalies, isotopic anomalies in other elements, and presolar grains (Sic, graphite, diamonds, etc.). In view of the b4’K homogeneity observed, this can no longer be considered to be the case.

The absence of evidence for partial vaporization in potassium, and the implication of condensation, is an argument for the existence of a hot solar nebula. Boss ( 1993) has calculated the thermal structure of a thin accretion disk formed around a central = 1Mo star, with compressional heating, and obtains temperatures of 1000-1400 K “thermostatically regulated” by the evaporation and recondensation of Fe-Ni metal ( = 1400 K), the dominant source of opacity, in the region inside of 2.5 AU. The calculated temperatures drop off rapidly to -160 K around 5 AU. This model agrees well with meteoritic evidence for high temperatures (Palme and Boynton, 1993) responsible for large-scale volatile element depletion, while allowing for the existence of low-temperature (ther- mally unprocessed) material (Cl chondrite) in the outer regions of the asteroid belt. It is also consistent with evidence for hot ( = 1300 K) circumstellar dust shells around T Tauri stars (Rydgren et al., 1982).

There are certain specific implications for the oxygen isotopic compositions of the planets that arise from this. Even if grains do not completely vaporize, oxygen isotope exchange rates are sufficient to isotopically exchange gas and dust (Yu et al., 1995) in the hot <2.5 AU region which comprises the bulk of the material accreted to the terrestrial planets. If this were the case, then the A”0 of the Earth, Moon, aubrites, and enstatite chondrites should be identical to the bulk nebular value, which may eventually be obtained by measurement of oxygen isotopes in the solar wind (R. Wiens and D. S. Bur- nett, pers. commun., 1994). The terrestrial planets should all be identical in oxygen isotopic composition, although this can


also be achieved by mixing and does not require a hot solar nebula. An exception can be made for Mars (the SNC parent body), since the material accreted by Mars may include larger contributions from regions beyond 2.5 AU ( Wetherill, 1994). Oxygen isotopic variations in chondrites can arise from dust- gas exchange during brief melting events (chondrule formation, Yu et al., 1995) or by mixing volatile depleted (with terrestrial oxygen isotope composition) material with cold, interstellar dust. This latter process does not appear to be dominant since the mixing lines in carbonaceous chondrites do not pass through the terrestrial composition (Clayton, 1993 ) . The presence of materials above the terrestrial fractionation line (e.g., ordinary chondrites) may require additional processes, e.g., gas-dust fractionation. This is not entirely ad hoc since ordinary chondrites are depleted in refractory lithophile elements by a factor of aO.75 relative to Si and Cl compositions’(Larimer and Anders, 1970; Kallemeyn et al., 1989). Nebular thermal structure of the type proposed by Boss ( 1993) offers an acceptable compromise between pervasive depletions of elements more volatile than Mg-silicates and Fe- Ni metal (Taylor, 1992; Palme and Boynton, 1993) and the presence of isotopic anomalies in oxygen (Clayton, 1993 ) .

Chronology

The chemical fractionations produced by refractory/volatile separation involve two radiometric parent/daughter systems: U/PI, and Rb/Sr. The U/PI, ages of meteorite parent bodies and planets date the major separation of volatile Pb from planetary matter, so that the age of the Earth essentially reflects a nebular process at 4.55 Ga (Patterson, 1956). The 87Rb/87Sr ages of chondrites reflect the timing of Rb/Sr fractionation in the nebula (Minster et al., 1982).

Likewise, initial 87Sr/86Sr ratios in primitive solar system objects date the time of separation of these objects from the Rb/Sr environment of the solar nebula (Papanastassiou and Wasserburg, 1969; Gray et al., 1973; Podosek et al., 1991; Stewart et al., 1993). The use of the chondritic Rb/Sr ratio of 0.30 2 0.03 (Anders and Grevesse, 1989; Humayun and Group, 1991) gives timescales for the separation of the eucrite parent body of 13 Ma relative to CAIs, generally considered to be long for a “free-fall” nebular timescale, although the timescale for disk evolution may be longer (Podosek and Cas- sen, 1994). This has been taken by Tilton ( 1988) and Lug- mair and Galer ( 1992) to indicate Rb/Sr separation during planetesimal accretion of the eucrite parent body, since that accretionary timescale ( lo’-lo8 years: Wetherill, 1989) is significantly longer than the nebular disk accretion timescale (alO years).

The results of the present study using K isotopes do not support vaporization during accretion for at least one primitive basaltic achondrite, Juvinas (Allegre et al., 1975). Nor do the results support magmatic or metamorphic devolatilization proposed by Mittlefehldt ( 1987). It is thus concluded that a nebular origin of the Rb/Sr fractionation should be entertained and the ramifications considered. Observational evidence indicates that dust disks with excess IR emission persist around pre-main sequence stars for periods of -10’ years (Strom et al., 1989), and so the BABI-CA1 age inferred above may not be unreasonable. The timescales for the Angra

dos Reis meteorite, LEW 86010 (angrite), and cumulate eucrites are all shorter than BABI, well within a possible nebular timescale.

Chondrites as building blocks of the planets

It has frequently been assumed since Ringwood ( 1966) that Cl chondrites or certain other kinds of chondrites constitute the building blocks of the terrestrial planets. Attempts to match the bulk composition of planets like the Earth with the observed compositions of chondrites involve a strong Pro- crustean aspect in that:

1) all known classes of chondrites have higher volatile element contents, and

2) oxygen isotopic compositions of these chondrites differ from those of the Earth, with the exception of enstatite chondrites, which otherwise have a distinctly nonterrestrial chemistry.

The oxygen isotopes are easily fixed since a triangle drawn between C3 chondrites (high I60 endmember), C 1 chondrites (high I80 endmember), and ordinary chondrites (low I60 endmember) includes the oxygen isotopic compositions of all achondrite types, the SNC meteorites (Mars?) and the Earth- Moon system (Clayton, 1993). The volatile elements were easy to fix, too, by vaporization of the unwanted excess. This should prove to be a more difficult exercise in the future.

Taylor and Norman ( 1990) argued for the accretion of the Earth from planetesimals that were volatile depleted and differentiated. This proposal has considerable merit both for ac- counting for the low volatile content of the Earth and for core formation, but the former is the concern of this study. There is ample evidence that volatile-depleted planetesimals existed prior to the complete accretion of the Earth as exemplified by angrites, eucrites and the Moon-forming impactor. If the giant impact theory is correct, as much as lo-20% of the mass of the Earth, depleted in potassium by a factor of five relative to the present bulk composition, was added in a single step from the impactor (Cameron and Benz, 199 1) The building blocks of the terrestrial planets must be composed of materials that have partially recondensed volatile elements from an initially hot state. It is interesting to note from Fig. 1 that the large planets Earth, Venus, and Mars, all have volatile element depletions intermediate between those of chondrites and the extremely depleted compositions of the basaltic achondrites and the Moon, a strong indication of mixing of depleted and un- depleted materials. This is an expected consequence of the accretion of the Earth and planets (Wetherill, 1990, 1994). The net contributions of chondrites can be estimated from the abundances of the most volatile elements to be <5% by mass Cl equivalent using Cs abundances. Lower, but less reliable estimates are obtained from other volatiles (see Anders and Owen, 1977, for a discussion of these results).

CONCLUSIONS

The most remarkable finding of this study is the uniformity of potassium isotopic composition of most solar system material. Reliable measurements of 6“‘K reported in this paper and by Humayun and Clayton ( 1995) indicate a range in S4’K of -4%0 to +13%0, with fractionations restricted to a few


natural materials, e.g., lunar anorthosite, 60015,742, silicates from Colomera IIE iron meteorite, and lunar soils. Measure- ments of ??‘K in whole-rock chondrites are identical within errors for all major classes of chondrites, including H, L, LL, E, Cl, CM, and CV, and are independent of metamorphic type. There is no evidence of isotopic variation between the Earth and chondrites with A4’K = -0.08 5 0.55 (3a,), or between the Earth and Moon with A4’K = 0.22 +. 0.54%0 (3a,). Isotopic measurements of a CAI-rich fragment of Vi- garano, a compact-type A CA1 from Leoville (L-6), and an olivine chondrule from Allende indicate no isotopic fractionation from terrestrial values. The large errors associated with L-6 allow several permil differences, which are insufficient to affect the bulk isotopic composition of Leoville. The achondrites Shergotty, Zagami, and Juvinas are unfractionated at the 3u level. An upper limit has been derived for Ibitira and Kenna of =2%0.

These isotopic results are inconsistent with nearly all proposed and conceivable mechanisms for the depletion of volatile elements by evaporation of condensed material, including :

1)

2)

3)

4)

loss from (interstellar) dust grains heated on entry into the solar nebula (Hashimoto et al., 1979; Hashimoto, 1983), loss from chondrules upon melting (Ganapathy and An- ders, 1974), loss from disrupted planetesimals (Wtike et al., 1981, 1984; Zook, 1981), and loss during accretionary heating of planetesimals and planets (Ringwood, 1966; Ahrens, 1990).

There are several reasons why potassium isotopic fractionation is the exception rather than the rule:

1)

2)

3)

Large bodies have gravitational fields that confine any vapor produced, small objects that have demonstrably experienced high temperatures must have had sufficiently high cooling rates to prevent enough chemical loss of potassium to produce detectable isotopic fractionation in potassium, e.g., impact melts, chondrules, etc., and partial vaporization is apparently restricted to the lunar soil, probably due to the tiny sizes of the vaporized particles involved.

We do not find support for the popular two-component models proposed by Wtike et al. ( 1984), WSnke and Dreibus ( 1988 ) , or for the Ganapathy and Anders ( 1974) model. This can be extended to any model using chondrites or chondritic components as the building blocks of the terrestrial planets. It is suggested that condensation from a hot solar nebula as envisaged by Grossman and Larimer ( 1974) qualitatively ac- counts for the volatile depletion observed in the components of meteorites and planets. A plausible nebular thermal structure has been found by Boss ( 1993), who included compressional heating of the infalling gas, and obtained temperatures sufficient to completely devolatilize any presolar grains. A specific prediction for the oxygen isotopic composition of the bulk solar nebula can be made: the A”0 is identical to the terrestrial composition, which can be tested by future oxygen isotope measurements of the solar wind if these are precise to

-tO.1%0. There are several issues remaining to be solved con- cerning the processes by which refractory dust grains take up volatiles on cooling, and the constraints this may impose upon coagulation of particles to form planetesimals.

Acknowledgmenrs-This study was carried out as part of the Ph.D. Thesis of MH, and supported by NASA grants NAG 9-51 and NAGW-3345 to RNC. Samples were generously provided by J. Gooding and the curatorial staff of JSC, L. Grossman, G. R. Huss, R. S. Lewis, H. Palme, and E. J. Olsen. G. R. Huss patiently dissolved seven primitive chondrites. 0. Draughn ably prepared polished mounts for ion probe analysis. Discussions with E. Anders, A. P. Boss, D. S. Burnett, L. Grossman, R. S. Lewis, M. C. Monaghan, H. Palme, E. N. Parker, F. M. Richter, E. Stolper, S. R. Taylor, .I. Truran, P. H. Warren, G. J. Wasserburg, J. T. Wasson, R. Wiens, D. G. York, and many other colleagues are gratefully acknowledged. We thank the reviewers. T. M. Esat, H. Palme, and D. Mittlefehldt. and the editor, C. Koeberl, for their efforts.

Editorial handling: C. Koeberl

REFERENCES

Ahrens T. J. ( 1990) Earth accretion. In Origin of the Earth (ed. H. E. Newsom and J. H. Jones), pp. 21 l-227. Oxford Univ. Press.

Allkgre C. J., Birck J. L., Fourcade S., and Semet M. P. (1975) Rubidium-87/Strontium-87 age of Juvinas basaltic achondrite and early igneous activity in the Solar System. Science 187, 436-438.

Anders E. ( 1964) Origin, age, and composition of meteorites. Space Sci. Rev. 3, 583-714.

Anders E. and Grevesse N. ( 1989) Abundances of the elements: Me- teoritic and solar. Geochim. Cosmochim. Actu 53, 197-214.

Anders E. and Owen T. ( 1977) Mars and Earth: Origin and abundances of volatiles. Science 198,453-465.

Biggar G. M., O’Hara M. J., Peckett A., and Humphries D. J. ( 197 1) Lunar lavas and the achondrites: Petrogenesis of protohy- persthene basalts in the maria lava lakes. Proc. Lunar Sci. Con5 2,617-643.

Boato G., Casanova G., and Vallauri M. E. ( 1960) Vapor pressure of isotopic liquids. II-Ne and A above boiling-point. Nuovo Ci- mento 14,505 -5 19.

Boctor N. Z., Palme H., Spettel B., El Goresy A., and MacPherson G. J. ( 1994) Caldera: A second unbrecciated noncumulate eucrite. Meteoritics 29, 445.

Boss A. P. ( 1993) Evolution of the solar nebula. II. Thermal structure during nebula formation. Astrophys. J. 417, 351-367.

Brewer L. ( 1953) The thermodynamic properties of the oxides and their vaporization processes. Chem. Rev. 52, l-75.

Burbidge E. M., Burbidge G. R., Fowler W. A., and Hoyle F. ( 1957) Synthesis of the elements in stars. Rev. Mod. Phys. 29, 547-650.

Cameron A. G. W. (1988) Origin of the solar system. Ann. Rev. Asrron. Asrrophys. 26,441-472.

Cameron A. G. W. and Benz W. (1991) The origin of the Moon and the single impact hypothesis IV. Icarus 92,204-216.

Clayton R. N. ( 1993) Oxygen isotopes in meteorites. Ann. Rev. Earth Planet. Sci. 21, 115-149.

Clayton R. N. and Mayeda T. K. (1975) Genetic relations between the Moon and meteorites. Proc. 6th Lunar Sci. ConfI, 1761- 1769.

Clayton R. N. and Mayeda T. K. ( 1983) Oxygen isotopes in eucrites, shergottites, nakhlites, and chassignites. Earth Planet Sci. Z&r. 62, 1-6.

Clayton R. N., Grossman L., and Mayeda T. K. ( 1973) A component of primitive nuclear composition in carbonaceous meteorites. Sci- ence 182,485-488.

Clayton R. N., Mayeda T. K., Onuma N., and Shearer J. (1976) Oxygen isotopic composition of minerals in the Kenna ureilite. Geochim. Cosmochim. Acra 40, 1475-1476.


Clayton R. N., Onuma N., Grossman L., and Mayeda T. K. ( 1977) Distribution of the pre-solar component in Allende and other carbonaceous chondrites. Earrh Planer. Sci. L&t. 34,209-224.

Clayton R. N., Hinton R. W., and Davis A. M. ( 1988) Isotopic variations in the rock-forming elements in meteorites. Phil. Trans. Roy. Sot. Load. A325,483 -501.

Davis A. M., Hashimoto A., Clayton R. N., and Mayeda T. K. ( 1990) Isotope mass fractionation during evaporation of Mg,SiO.+. Nature 347,655-658.

De Maria G., Balducci G., Guido M., and Piacente V. ( 1971) Mass spectrometric investigation of the vaporization process of Apollo 12 lunar samples. Proc. 2nd Lunar Sci. Conj, 1367- 1380.

Des Marais D. J. ( 1983) Light element geochemistry and spallo- genesis in lunar rocks. Geochim. Cosmochim. Acfa 47, 1769- 1781.

Ding T. P., Thode H. G., and Rees C. E. ( 1983) Sulphur content and sulphur isotope composition of orange and black glasses in Apollo 17 drive tube 7400211. Geochim. Cosmochim. Acta 47, 491-496.

Dowty E., Keil K., Prim M., Gros J., and Takahashi H. ( 1976) Me- teorite-free Apollo 15 crystalline KREEP. Proc. 7rh Lunar Sci. Conj, 1833- 1844.

Dreibus G., Kruse H., Spettel B., and Wlnke H. (1977) The bulk composition of the moon and the eucrite parent body. Proc. 8rh Lunar Sci. ConjI. 211-227.

Epstein S. and Taylor H. P., Jr. ( 1970) The concentration and isotopic composition of hydrogen, carbon and silicon in Apollo 11 lunar rocks and minerals. Proc. 1st Lunar Sci. Conf., 1085- 1096.

Ganapathy R. and Anders E. ( 1974) Bulk compositions of the Moon and Earth, estimated from meteorites. Proc. 5th Lunar Sci. ConJ, 1181-1206.

Esat T. M., Brownlee D. E., Papanastassiou D. A., and Wasserburg G. J. ( 1979) Magnesium isotopic composition of interplanetary dust particles. Science 2,06, 190- 197.

Fox J. L. and Dalgamo A. ( 1983) Nitrogen escape from Mars. J. Geophys. Rex U&9027-9032.

Gao X. and Thiemens M. H. ( 1991) Systematic study of sulfur isotopic composition in iron meteorites and the occurrence of excess “S and ?S. Geochim. Cosmochim. Acta S&2671 -2679.

Gao X. and Thiemens M. H. (1993a) Isotopic composition and concentration of sulfur in carbonaceous chondrites. Geochim. Cos- mochim. Acta 57,3159-3169.

Gao X. and Thiemens M. H. ( 1993b) Variations of the isotopic composition of sulfur in enstatite and ordinary chondrites. Geochim. Cosmochim. Acta 57,3 17 1-3 176.

Gast P. W. ( 1960) Limitations on the composition of the upper mantle. J. Geophys. Res. 65, 1287- 1297.

Gooding J. L. and Muenow D. W. ( 1976) Activated release of alkalis during the vesiculation of molten basalts under high vacuum: Im- plications for lunar volcanism. Geochim. Cosmochim. Acta 40, 675-686.

Gooding J. L. and Muenow D. W. ( 1977) Experimental vaporization of the Holbrook chondrite. Meteoritics 12,401-408.

Gray C. M., Papanastassiou D. A., and Wasserburg G. J. ( 1973) The identification of early condensates from the solar nebula. Icarus 20,213-239.

Grossman J. N. ( 1988) Formation of chondrules. In Meteorites and the Early Solar System (ed. J. F. Kenidge and M. S. Matthews), pp. 680-696. Univ. Arizona Press.

Grossman J. N. and Wasson J. T. ( 1983) The compositions of chondrules in unequilibrated chondrites: An evaluation of models for the formation of chondrules and precursor materials. In Chon- drules and their Origins (ed. E. A. King), pp. 88-121. Lunar Planet. Inst.

Grossman L. (1972) Condensation in the primitive solar nebula. Geochim. Cosmochim. Acta 36,597-619.

Grossman L. and Larimer J. W. ( 1974) Early chemical history of the Solar System. Rev. Geophys. Space Phys. 12,7 1 - 101.

Hashimoto A. (1983) Evaporation metamorphism in the early solar nebula-evaporation experiments on the melt FeO-MgO-SiO,- CaO-AlzOj and chemical fractionations of primitive materials. Geochem. J. 17, Ill- 145.

Hashimoto A., Kumazawa M., and Onuma N. ( 1979) Evaporation metamorphism of primitive dust material in the early solar nebula. Earth Planet. Sci. Lett. 43, 13-21.

Haskin L. A., Shih C.-Y., Bansal B. M., Rhodes J. M., Wiesmann H., and Nyquist L. E. (1974) Chemical evidence for the origin of 76535 as a cumulate. Proc. 4th Lunar Sci. ConJ, 1213- 1225.

Humayun M. ( 1994) Potassium isotope cosmochemistry: Implica- tions for volatile depletion and origins of the Earth, Moon and meteorites. Ph.D. dissertation, Univ. Chicago.

Humayun M. and Clayton R. N. ( 1993) Potassium isotope cosmochemistry, volatile depletion and the origin of the Earth. Lunar Planet. Sci. XXIV, 685-686.

Humayun M. and Clayton R. N. ( 1994) The non-terrestrial origin of the Moon. Lunar Planet. Sci. XXV, 579-580.

Humayun M. and Clayton R. N. (1995) Precise determination of the isotopic composition of potassium: Application to terrestrial rocks and lunar soils. Geochim. Cosmochim. Acfa 59, 2115-2130.

Humayun M. and Group C. C. ( 1991) Solar Rb/Sr: Is it non-chondritic? Lunar Planet Sci. XXII, 595-596.

Hunten D. M., Pepin R. O., and Walker J. C. G. ( 1987) Mass fractionation in hydrodynamic escape. Zcarus 69, 532-549.

Huss G. R. ( 1989) Ubiquitous interstellar diamond and SIC in primitive chondrites: Abundances reflect metamorphism. Nature 347, 159-162.

Jagoutz E. et al. ( 1979) The abundances of major, minor and trace elements in the Earth’s mantle as derived from primitive ultra- mafic nodules. Proc. 10th Lunar Planet. Sci. Conj, 2031- 2050.

Jochum K. P. and Palme H. ( 1990) Alkali elements in eucrites and SNC meteorites: No evidence for volatility related losses during magma eruption or thermal metamorphism. Meteoritics 25, 373- 374.

Kaula W. M. (1979) Thermal evolution of the Earth and Moon growing by planetesimal impacts. J. Geophys. Res. 84, 999- 1008.

Kallemeyn G. W., Rubin A. E., Wang D., and Wasson J. T. ( 1989) Ordinary chondrites: Bulk compositions, classification, lithophile-element fractionations, and composition-petrographic type relationships. Geochim. Cosmochim. Acta 53, 2747-2767.

Larimer J. W. ( 1967) Chemical fractionations in meteorites-I. Con- densation of the elements. Geochim. Cosmochim. Acta 31, 12 15- 1238.

Larimer J. W. and Anders E. ( 1967) Chemical fractionations in meteorites-11. Abundance patterns and their interpretation. Geo- chim. Cosmochim. Acta 31, 1239- 1270.

Larimer J. W. and Anders E. ( 1970) Chemical fractionations in meteorites-111. Major element fractionations in chondrites. Geo- chim. Cosmochim. Acta 34,367-387.

Lugmair G. W. and Galer S. J. G. ( 1992) Age and isotopic relationships among the angrites Lewis Cliff 86010 and Angra dos Reis. Geochim. Cosmochim. Acta 56, 1673-1694.

McDonough W. F., Sun S-S., Ringwood A. E., Jagoutz E., and Hof- mann A. W. ( 1992) Potassium, rubidium, and cesium in the Earth and Moon and the evolution of the mantle of the Earth. Geochim. Cosmochim. Acta 56, 100 I- 10 12.

McElroy M. B., Yung Y. L., and Nier A. 0. ( 1976) Isotopic composition of nitrogen: Implications for the past history of Mars’ atmosphere. Science 194, 70-72.

Melosh H. J. ( 1990) Giant impacts and the thermal state of the early Earth. In Origin ofthe Earth (ed. H. E. Newsom and J. H. Jones), pp. 69-83. Oxford Univ. Press.

Melosh H. J. and Vickery A. M. (1989) Impact erosion of the primordial atmosphere of Mars. Nature 338, 487-489.

Minster J. F., Birck J. L., and Allegre C. J. ( 1982) Absolute age of formation of chondrites studied by the 87Rb-87Sr method. Nature 300,414-419.

Mittlefehldt D. W. ( 1987) Volatile degassing of basaltic achondrite parent bodies: Evidence from alkali elements and phosphorus. Geochim. Cosmochim. Acta 51,267-278.

Molini-Velsko C., Mayeda T. K., and Clayton R. N. ( 1986) Isotopic composition of silicon in meteorites. Geochim. Cosmochim. Acta 50,2719-2726.


Nier A. O., McElroy M. B., and Yung Y. L. ( 1976) Isotopic composition of the Martian atmosphere. Science 194, 68-70.

Norman M. D. and Taylor S. R. ( 1992) Geochemistry of lunar crustal rocks from breccia 67016 and the composition of the Moon. Gee- chim. Cosmochim. Actu 56, 10 13 - 1024.

Notsu K., Onuma N., Nishida N., and Nagasawa H. ( 1978) High temperature heating of the Allende meteorite. Geochim. Cosmo- chim. Acta 42, 903-907.

O’Hara M. J., Biggar G. M., Richardson S. W., Ford C. E., and Jamieson B. G. ( 1970) The nature of seas, mascons and the lunar interior in the light of experimental studies. Proc. 1st Lunar Sci. Conf., 695-710.

Owen T., Maillard J. P., De Bergh C.. and Lutz B. L. (1988) Deu- terium on Mars: The abundance of HDO and the value of D/H. Science 240, 1767- 1770.

Palme H. and Boynton W. V. (1993) Meteoritic constraints on conditions in the solar nebula. In Protosturs and Plawts III (ed. E. H. Levy and J. I. Lunine), pp. 979-1004. Univ. Arizona Press.

Papanastassiou D. A. and Wasserburg G. J. ( 1969) initial strontium isotopic abundances and the resolution of small time differences in the formation of planetary objects. Em-01 Plunet. Sci. Lrtt. 5, 361-376.

Patterson C. C. ( 1956) Ages of meteorites and the Earth. Grochirn. Cosmochim. Acta 10, 230-237.

Pepin R. 0. ( 1991) On the origin and early evolution of terrestrial planet atmospheres and meteoritic volatiles. Icurus 92, 2-79.

Podosek F. A. ( 1978) Isotopic structures in solar system materials. Ann. Rev. Aaron. Astrophys. 16, 293-334.

Podosek F. A. and Cassen P. ( 1994) Theoretical, observational, and isotopic estimates of the lifetime of the solar nebula. Meteoritics 29, 6-25.

Podosek F. A. et al. ( 1991) Correlated study of initial X7Sr/XhSr and AI-Mg isotopic systematics and petrologic properties in a suite of refractory inclusions from the Allende meteorite. Geochim. Cos- mochim. Acta 55, 1083- 11 IO.

Reynolds J. H. ( 1967) Isotopic abundance anomalies in the solar system. Ann Rev. Nucl. Sri. 17, 253-316.

Ringwood A. E. ( 1966) Chemical evolution of the terrestrial planets, Geochim. Cosmochim. Acta 30, 4 I - 104.

Ringwood A. E. ( 1989) Signiticance of the terrestrial Mg/Si ratio. Earth Planet. Sci. L&t. 95, l-7.

Ringwood A. E. ( 1992) Volatile and siderophile element geochemistry of the Moon: a reappraisal. Earth Planet. Sci. Len. 111, 537- 555.

Ruden S. P. and Lin D. N. C. (1986) The global evolution of the primordial solar nebula. Asrrophys. J. 308, 883-90 I

Russell W. A., Papanastassiou D. A., and Tombrello T. A. ( 1978) Ca isotope fractionation on the Earth and other solar system materials. Geochim. Cosmochim. Acta 42, 1075- 1090.

Rydgren A. E., Schmelz J. T., and Vrba F. J. ( 1982) Evidence for a characteristic maximum temperature in the circumstellar dust associated with T Tauri stars. Astrophys. J. 256, 168- 176.

Sakai H., Des Marais D. J.. Ueda A., and Moore J. G. (1984) Concentrations and isotope ratios of carbon, nitrogen and sulfur in ocean-floor basal&. Geochim. Cosmochim. Acta 48, 2433- 2441.

Sears D. W. G., Jie L., and Benoit P. H. ( 1991) Volatile loss during chondrule formation. Meteoritics 26, 394.

Spitzer L. ( 1952) The terrestrial atmosphere above 300 km. In The Atmospheres of the Earth und Planets (ed. G. P. Kuiper), pp. 213- 249. Univ. Chicago Press.

Srinivasan G., Ulyanov A. A., and Goswami J. N. ( 1993) Search for 4’K excess in Efremovka CAls. Meteoritics 28,442.

Steele 1. M. ( 1986) Compositions and textures of relic forsterite in carbonaceous and unequilibrated ordinary chondrites. Geochim. Cosmochim. Acta 50, 1379- 1395.

Stewart B. W., Papanastassiou D. A., Capo R. C., and Wasserburg G. J. ( 1993) Fine resolution chronology based on initial X7Sr/XhSr. Lunar Planet. Sci. XXIV, 1357- 1358.

Stolper E. (1977) Experimental petrology of eucritic meteorites. Geochim. Cosmochim. Acttr 41, 587-61 I.

Storey W. C. ( 1973) Volatilization studies on a terrestrial basalt and their applicability to volatilization from the lunar surface. Nuture Phys. Sci. 241, 154- 157.

Strom K. M., Strom S. E., Edwards S., Cabrit S., and Skrutskie M. F. (1989) Circumstellar material associated with solar-type pre-main-sequence stars: A possible constraint on the timescale for planet building. Astron. J. 97, 145 1 - 1470.

Suess H. E. ( 1965 ) Chemical evidence bearing on the origin of the solar system. Ann. Rev. Astron. Astrophys. 3, 217-234.

Sylvester P. J.. Simon S. B., and Grossman L. (1993) Refractory inclusions from the Leoville, Efremovka, and Vigarano C3V chondrites: Major element differences between Types A and B, and extraordinary refractory siderophile element compositions. Gm- chim. Cosmochim. Acta 57, 3763-3784.

Takahashi H., Gros J., Higuchi H., Morgan J. W., and Anders 8. ( 1978) Volatile elements in chondrites: metamorphism or nebular fractionation? Geochim. Cosmochim. Actu 42, I859- 1869.

Taylor S. R. ( 1975) Luntir Science; A Post-Apollo View. Pergamon. Taylor S. R. (1979) Lunar and terrestrial potassium and uranium

abundances: Implications for the fission hypothesis. Proc. 10th Lunur Plunet. Sci. Conf, 2017-2030.

Taylor S. R. ( 1992) Solur System Evolution: A New Perspective. Cambridge.

Taylor S. R. and Norman M. D. ( 1990) Accretion of differentiated planetesimals to the Earth. In Origin <f the Eurth (ed. H. E. New- som and J. H. Jones), pp. 29-43. Oxford Univ. Press.

Taylor S. R., Norman M. D., and Esat T. M. (1993) The Mg-suite and the highland crust: An unsolved enigma. Lunrrr Planet. Sci. XXIV, 1413-1414.

Tera F., Papanastassiou D. A.. and Wasserburg G. J. ( 1974) Isotopic evidence for a terminal lunar cataclysm. Earth Planet. Sci. Lett. 22, l-21.

Thode H. G. and Rees C. E. ( 1979) Sulphur isotopes in lunar and meteorite samples. Proc. 10th Lunur Plunrt. Sci. Corlf:. 1629- 1636.

Thode H. G., Monster J., and Dunford H. B. ( 1961) Sulphur isotope geochemistry. Geochim. Cosmochim. Acta 25, 159- 174.

Tilton G. R. ( 1988) Age of the solar system. In Meteorites and the Early Solar System (ed. J. F. Kerridge and M. S. Matthews), pp. 259-275. Univ. Arizona Press.

Trimble V. ( 1975) The origin and abundances of the chemical elements. Rev. Mod. Phys. 47, 877-976.

Wai C. M. and Wasson J. T. ( 1977) Nebular condensation of moderately volatile elements and their abundances in ordinary chondrites. Earth Planet. Sri. Left. 36, I - 13.

Walker J. C. G. ( 1986) Impact erosion of planetary atmospheres. Icarus 68, 87-98.

Wang J., Davis A. M., Hashimoto A., and Clayton R. N. ( 1991 ) The role of diffusion in the isotopic fractionation of magnesium during the evaporation of forsterite. Luncrr PIunet. Sci. XXII, 1461-1462.

Wang J., Davis A. M.. Hashimoto A., and Clayton R. N. ( I993 ) Diffusion-controlled magnesium isotopic fractionation of a single crystal forsterite evaporated from the solid state. Lunur Planet. Sci. XXIV, 1479- 1480.

Wang J., Davis A. M., Clayton R. N., and Mayeda T. K. (1994a) Chemical and isotopic fractionation during the evaporation of the Fe0-Mg0-Si02-Ca0-Al~OI-TiOZ_REE melt system. Lunur Plunrt. Sci. XXV, 1457- 1458.

Wang J., Davis A. M.. Clayton R. N., and Mayeda T. K. ( 1994b) Kinetic isotopic fractionation during the evaporation of the iron oxide from liquid state. Lunar Planet. Sci. XXV, 1459- 1460.

WPnke H. and Dreibus G. ( 1988) Chemical composition and accretion history of terrestrial planets. Phil. Trurts. Roy. Sot. Land. A325,545 -557.

Wanke H. et al. ( 1972) Multielement analyses of lunar samples and some implications of the results. Proc. 3rd Lunar Sci. Conj:, 125 I - 1268.

Wanke H. et al. ( 1973) Multielement analyses of Apollo 15, 16, and 17 samples and the bulk composition of the Moon. Proc. 4th Lurrrrr Sci. Co+ 1461-1481.


Wlnke H. et al. ( 1974) Chemistry of Apollo 16 and 17 samples: Bulk composition, late stage accumulation and early differentiation of the moon. Proc. 5th Lunur Sri. Conj, 1307- 1335.

Wanke H., Dreibus G., Jagoutz E., Palme H., and Rammensee W. ( I98 1) Chemistry of the Earth and the significance of primary and secondary objects for the formation of planets and meteorite parent bodies. Lunar Plunef. Sci. XII, I 139- 1 141.

Wlnke H., Dreibus G., and Jagoutz E. ( 1984) Mantle chemistry and accretion history of the Earth. In Archaean Geochemistry (ed. A. Kriiner et al.), pp. l-24. Springer-Verlag.

Warren P. H. ( 1985) The magma ocean concept and lunar evolution. Ann. Rev. Earth Planet. Sci. 13, 201-240.

Warren P. H. and Wasson J. T. ( 1979) The origin of KREEP. RL’Y. Geophys. Space Phys. 17, 73-88.

Warren P. H., Afiattalab F., and Wasson .I. T. ( 1978) Investigation of unusual KREEPy samples: Pristine rock 15386, Cone Crater soil fragments 14143, and 12023, a Typical Apollo 12 soil. Proc. Lunar Planet. Sci. ConjI 9, 653-660.

Wasserburg G. J. ( 1985) Short-lived nuclei in the early solar system. In Procostars and Planets ZZ (ed. D. C. Black and M. S. Matthews), pp. 703-737. Univ. Arizona Press.

Wasserburg G. J., MacDonald G. J. F., Hoyle F., and Fowler W. A. ( 1964) Relative contributions of uranium, thorium, and potassium to heat production in the Earth. Science 143, 465-467.

Wasserburg G. J., Tera F., Papanastassiou D. A., and Huneke J. C. ( 1977) Isotopic and chemical investigations on Angra dos Reis. Earth Planet. Sci. Lett. 35, 294-3 16.

Wasson J. T. ( 1992) Constraints on chondrule origins. Meteorirks 27, 304.

Wasson J. T. and Kallemeyn G. W. ( 1988) Compositions of chondrites. Phil. Trans. Roy. Sot. Land. A325,535-544.

Watson L. L., Hutcheon I. D., Epstein S., and Stolper E. M. ( 1994) Water on Mars: Clues from deuterium/hydrogen and water contents of hydrous phases in SNC meteorites. Science 265, 86-90.

Wetherill G. W. ( 1989) The formation of the Solar System: Consen- sus. alternatives, and missing factors. In The Formation and Evo- lution of Planetary Systems (ed. H. A. Weaver and L. Danly), pp. l-30. Cambridge Univ. Press.

Wetherill G. W. (1990) Formation of the Earth. Ann. Rev. Earth Planet. Sri. 18, 205-256.

Wetherill G. W. ( 1994) Provenance of the terrestrial planets. Geo- chim. Cosmochim. Acta 58,45 13 -4520.

Wlotzka F. et al. ( 1983) Alkali differentiation in LL-chondrites. Geo- chim. Cnsmochim. Acta 47,743-757.

Wolf R. and Anders E. ( 1980) Moon and Earth: compositional differences inferred from siderophiles, volatiles, and alkalis in basalts. Geochim. Cosmochim. Acts 44, 2 1 I I-21 24.

Wulf A. V., Palme H., and Jochum K. P. ( 1995) Fractionation of volatile elements in the early solar system: Evidence from heating experiments on primitive meteorites. Planet. Spuce Sci. (in press).

Yu Y., Hewins R. H., Clayton R. N., and Mayeda T. K. (1995) Experimental study of the high temperature oxygen isotope exchange during chondrule formation. Geochim. Cosmochim. Acra 59,2095-2104.

Zook H. A. ( 198 1) On a new model for the generation of chondrites. Lunar Planet. Sci. XII, l242- 1244.

Documents

Humayun M. and Clayton RN