Chronology, geochemistry, and petrology of a ferroan ...€¦ · Descartes terrain may, therefore, provide a glimpse of a more representative region of the lunar crust than the near

Meteoritics amp Planetary Science 38 Nr 4 645ndash661 (2003)Abstract available online at httpmeteoriticsorg

645 copy Meteoritical Society 2003 Printed in USA

Chronology geochemistry and petrology of a ferroan noritic anorthosite clast from Descartes breccia 67215 Clues to the age origin

structure and impact history of the lunar crust

Marc D NORMAN1 Lars E BORG2 Lawrence E NYQUIST3 and Donald D BOGARD3

1Research School of Earth Sciences Australian National University Canberra ACT 0200 Australia2Institute of Meteoritics University of New Mexico Albuquerque New Mexico 87131 USA

3NASA Johnson Space Center Houston Texas 77058 USACorresponding author E-mail MarcNormananueduau

(Received 15 October 2003 revision accepted 28 January 2003)

AbstractndashThe petrology major and trace element geochemistry and Nd-Ar-Sr isotopic compositionsof a ferroan noritic anorthosite clast from lunar breccia 67215 have been studied in order to improveour understanding of the composition age structure and impact history of the lunar crust The clast(designated 67215c) has an unusually well preserved igneous texture Mineral compositions areconsistent with classification of 67215c as a member of the ferroan anorthositic suite of lunarhighlands rocks but the texture and mineralogy show that it cooled more rapidly and at shallowerdepths than did more typical ferroan anorthosites (FANs) Incompatible trace element concentrationsare enriched in 67215c relative to typical FANs but diagnostic signatures such as TiSm ScSmplagiophile element ratios and the lack of ZrHf and NbTa fractionation show that this cannot be dueto the addition of KREEP Alternatively 67215c may contain a greater fraction of trapped liquid thanis commonly present in lunar FANs 147Sm-143Nd isotopic compositions of mineral separates from67215c define an isochron age of 440 plusmn 011 Gyr with a near-chondritic initial e143

Nd of +085 plusmn 053The 40Ar-39Ar composition of plagioclase from this clast records a post-crystallization thermal eventat 393 plusmn 008 Gyr with the greatest contribution to the uncertainty in this age deriving from a poorlyconstrained correction for lunar atmosphere 40Ar Rb-Sr isotopic compositions are disturbedprobably by the same event recorded by the Ar isotopic compositions Trace element compositions ofFANs are consistent with crystallization from a moderately evolved magma ocean and do not supporta highly depleted source composition such as that implied by the positive initial e143

Nd of the ferroannoritic anorthosite 62236 Alternatively the Nd isotopic systematics of lunar FANs may have beensubject to variable degrees of modification by impact metamorphism with the plagioclase fractionbeing more strongly affected than the mafic phases 147Sm-143Nd isotopic compositions of maficfractions from the 4 ferroan noritic anorthosites for which isotopic data exist (60025 62236 67016c67215c) define an age of 446 plusmn 004 Gyr which may provide a robust estimate for the crystallizationage of lunar ferroan anorthosites

INTRODUCTION

Lunar ferroan anorthosites (FANs) are relicts of anancient primary feldspathic crust that is commonly thought tohave formed by accumulation of plagioclase from a globalmagma ocean Compositions and ages of ferroan anorthositesprovide fundamental information about the evolution of themoon the structure and impact history of the lunar crust andthe timescales of planetary differentiation in the inner solarsystem Here we report the results of petrologicgeochemical and isotopic (Nd-Ar-Sr) studies of a ferroan

noritic anorthosite clast from lunar breccia 67215 to improveour understanding of the chronology structure and origin ofthe lunar crust

Lunar sample 67215 is a feldspathic fragmental brecciathat was collected from the rim of North Ray Crater Apollo16 Although classified by Lindstrom and Lindstrom (1986)as a granulitic breccia the presence of aphaniticmicroporphyritic melt breccia clasts in 67215 (McGee 1988)demonstrates a lithologic affinity with the feldspathicfragmental breccia suite These breccias have aluminous bulkcompositions (28ndash30 Al2O3) that are poor in KREEP low in

646 M D Norman et al

meteoritic siderophiles and lack solar wind carbon comparedto lunar impact melts or regolith breccias (Norman 1981)They appear to represent a regionally significant unit exposedin the Descartes terrane of the lunar highlands and their bulkcompositions are broadly similar to those of large regions ofanorthositic crust discovered on the far side of the moon bythe Clementine and Lunar Prospector missions TheDescartes terrain may therefore provide a glimpse of a morerepresentative region of the lunar crust than the near sideKREEP-rich breccias that dominate the Apollo samplecollection The Descartes breccias also contain magnesianand ferroan components that may represent importantlithologies of the ancient lunar crust (Lindstrom andLindstrom 1986) The antiquity of at least some of theselithologies is demonstrated by the 453 plusmn 012 Gyr 147Sm-143Nd isochron age obtained for a ferroan noritic anorthositeclast from the Descartes breccia 67016 (Alibert et al 1994)Breccia 67215 consists predominantly of lithic clasts andmineral fragments derived from ferroan noritic anorthosite(Lindstrom and Lindstrom 1986 McGee 1988) and one ofthe clasts in this breccia is the subject of this study

We also address the recent controversy concerning the ageand magmatic source composition of lunar ferroananorthosites Conventional magma ocean models predictcrystallization of FANs and related rocks early in lunar historyfrom a moderately evolved parental magma with near-chondritic relative abundances of LREE and Nd isotopiccompositions (ie initial e143

Nd ~0) However previous Ndisotopic studies of lunar FANs (Carlson and Lugmair 1988Alibert et al 1994 Borg et al 1999) do not provide strongsupport for the classical view of lunar crustal genesis and infact provoke significant challenges to the magma oceanparadigm Specifically the young Sm-Nd isochron age of 429Gyr combined with a remarkably positive initial 143Nd isotopiccomposition (e143

Nd = +3) for ferroan noritic anorthosite 62236(Borg et al 1999) is difficult to accommodate within aconventional magma ocean interpretation This has led toalternative proposals for lunar crustal petrogenesis involvingremelting of mafic cumulates (Borg et al 2002 Longhi 2002)or early depletion of the lunar magma ocean by a proto-crustenriched in LREE (Warren 2001)

Sample Preparation and Analytical Methods

During examination of breccia 67215 in the PlanetaryMaterials Curatorial Laboratories of the NASA JohnsonSpace Center a coarse-grained clast was identified andextracted from the breccia This clast was subsequentlyallocated for chemistry as 67215 46 (hereafter 67215c) and apolished thin section (67215 55) prepared from a small chipcontaining a fragment of the clast and adhering host brecciaA whole rock sample of host breccia (67215 39) was alsoallocated for chemistry

Clast 67215c was prepared for analysis in a manner

similar to that used in our previous study of 62236 (Borg et al1999) After separating the clast from adhering breccia matrix~280 mg of uncontaminated clast material was recoveredGrain size fractions of 100ndash200 200ndash325 and gt325 meshwere produced by gentle crushing in a boron carbide mortarPlagioclase and mafic mineral concentrates were obtainedusing heavy liquids with density cuts at lt285 285ndash332 andgt332 gcm3 and the mineral separates were further purifiedby handpicking Prior to crushing 2 small fragments thatappeared to be representative of the clast were separated forwhole rock trace element analysis these were crushedseparately with an agate mortar and pestle A small number ofgrains were also taken from the density separates forpetrography and mineral analysis including a fewpolymineralic fragments with abundant fine-grained opaquesThese ldquorockletsrdquo were thought to offer the best possibility ofsampling minor phases such as phosphates that might bepresent interstitially Petrographic observations were obtainedon the polished thin section and the grain mounts using opticalmicroscopy and backscattered electron imaging Mineralcompositions were determined by wavelength-dispersiveelectron microprobe using count times of 20ndash40 sec for majorelements and 40ndash60 sec for minor elements Plagioclase wasanalyzed using a 10 micron diameter defocused beam allother phases were analyzed using a focused beam at 15 keVBackgrounds were counted on both sides of the analyte peakNatural mineral standards were used for calibration

Major and trace element data were obtained on the twosamples of 67215 (clast and breccia) by solution ICPMS(Table 1) Rock powders were dissolved in HF + HNO3 andbrought to final volume with 2 HNO3 (Norman et al 1998a)For trace element analysis the sample was diluted 1000times Forthe major element analysis an additional 100times dilution wasapplied to the trace element solution Potassium abundanceswere measured in ldquocool plasmardquo mode using a low forwardpower setting (700 W) which reduces the background on 39Kcaused by the argon plasma For comparison we also reportmajor and trace element compositions on aliquots of wholerock powders and mineral separates from 62236 The majorelement composition of 62236 was determined by fusing 15mg of rock powder to a glass on an Ir strip and analyzing theglass by electron microprobe Trace element abundances weremeasured on whole rock samples by solution ICPMS asdescribed above and by INAA (Mittlefehldt and Lindstrom1993) Trace element abundances in splits of plagioclase andmafic silicate fractions used for the isotopic analysis were alsomeasured by INAA

Sm-Nd and Rb-Sr isotopic data were obtained on wholerock and mineral separates of 67215 (Table 2 and 3) followingprocedures similar to those used for our previous study of62236 (Borg et al 1999) Before isotopic analysis Nd and Srconcentrations were measured by isotope dilution on smallaliquots of each sample to enable optimum amounts of themixed Rb-Sr and Sm-Nd spike to be added The Sm isotopic

Descartes breccia 67215 Clues to the history of the lunar crust 647

Table 1 Major and trace element compositions of 67215 62236 and BHVOndash2 67215 31 67215 46 62236 43 62236 21 BHVOndash2breccia clast WR WR PLAG MAFIC

wt ICPMS ICPMS EMP INAA INAA INAA ICPMS

SiO2 4542a 4510a 4389 ndash ndash ndash 5057a

TiO2 024 018 ndash ndash ndash ndash 277Al2O3 2736 2826 2763 ndash ndash ndash 1376FeO 573 548 616 809 115 326 1126MnO 008 008 005 ndash ndash ndash 017MgO 463 429 596 ndash ndash ndash 726CaO 1621 1626 1577 157 196 053 1143Na2O 031 032 018 0184 0228 00084 223K2O 0019 0018 ndash lt0019 lt0028 lt0005 056Sum 10000 10000 9964 ndash ndash ndash 10000

ppm ICPMS ICPMS ICPMS INAA INAA INAA ICPMS

Li 31 32 15 47K 156 147 ndash ndash ndash ndash ndashSc 134 112 52 346 103 107 323Ti 1470 1183 341 ndash ndash ndash 17338V 20 16 14 ndash ndash ndash 327Cr ndash ndash ndash 492 153 864 naCo 10 16 85 187 28 748 46Ni 28 49 87 23 lt10 lt70 123Cu 13 09 32 ndash ndash ndash 123Zn 12 10 18 ndash ndash ndash 101Ga 28 31 27 ndash ndash ndash 206Rb 09 05 065 ndash ndash ndash 94Sr 139 156 149 128 157 lt80 387Y 63 51 095 ndash ndash ndash 278Zr 116 109 190 ndash ndash ndash 178Nb 057 053 010 ndash ndash ndash 193Mo 001 002 nd ndash ndash ndash 421Sn 011 005 002 ndash ndash ndash 21Cs 0099 0049 0114 0087 0092 012 010Ba 170 224 730 lt15 15 lt30 131La 088 095 0188 0139 0162 0017 153Ce 232 231 0464 037 042 ndash 368Pr 032 031 0060 ndash ndash ndash 516Nd 160 143 0286 ndash ndash ndash 241Sm 055 045 0088 0055 0055 0031 610Eu 075 082 0685 0593 0742 lt002 204Gd 075 063 0109 ndash ndash ndash 612Tb 013 011 0020 0021 00076 lt003 095Dy 093 075 0145 ndash ndash ndash 521Ho 021 017 0034 ndash ndash ndash 099Er 063 052 0106 ndash ndash ndash 253Yb 064 054 0117 0077 0037 018 194Lu 0096 0078 0018 0012 00044 0031 027Hf 031 026 0049 lt005 lt0028 lt01 428Ta 0033 0029 0006 lt002 lt001 lt006 124Pb 039 020 015 ndash ndash ndash 151Th 0070 0068 0011 lt0035 lt0015 lt006 121U 0019 0024 00023 lt006 lt004 lt021 041

aSiO2 by difference

648 M D Norman et al

composition of a split of 67215 host breccia was measured toevaluate the effects of neutron exposure on the lunar surfaceTwo splits of the 67215 breccia matrix were analyzed for Rb-Sr isotopic composition (Table 2)

40Ar-39Ar isotopic data were obtained on a 21 mg sampleof plagioclase from the 67215 clast and on a 63 mg wholerock sample from the 67215 breccia matrix These samplesand the NL-25 hornblende monitor were irradiated with fastneutrons (J-value of 00314) at the University of Missouri Arwas extracted in stepwise temperature release and its isotopiccomposition was measured on a mass spectrometer followingthe procedures of Bogard et al (1995) The isotopic data werecorrected for system blanks radioactive decay and reactor-produced interferences For most extractions 40Ar and 39Arblanks were 2ndash3 and 3ndash4 respectively Because of thelarge CaK ratio in the plagioclase corrections for thereaction 42Ca (n a) 39Ar were 15ndash29 for the variousextractions We used a 39Ar37Ar correction factor of 745 plusmn020 times 10-4 This factor was obtained by irradiating severalsamples of pure CaF2 at different times in the same reactorposition as used for 67215 and we believe the uncertaintygiven for this factor is realistic Where we report an averageAr-Ar age with an uncertainty that uncertainty includesconsideration of uncertainties in isotopic ratio measurementblank decay reactor corrections and uncertainty inirradiation constant However the Ar-Ar age spectrapresented for stepwise temperature extractions do not containthe uncertainty in the J value

RESULTS

Petrography and Mineral Compositions

The noritic anorthosite clast 67215c has an unbrecciatedigneous texture that is easily distinguished from thefragmental host breccia (Fig 1) Plagioclase (An96ndash98) grainsare euhedral to subhedral and range in size from ~01ndash07 mmlong with a median length of ~05 mm Interstitial pyroxeneand olivine (Fo493ndash541) take irregular shapes with low-Capyroxene gt high-Ca pyroxene gtgt olivine Thin (pound5 micronswide) exsolution lamellae are abundant in both low-Ca andhigh-Ca pyroxene with compositions ranging fromWo14En583 to Wo445En398 (Fig 2) Chromite ilmenite FeNimetal and troilite are present as minor phases and often occuras complex fine-grained intergrowths interstitial to theplagioclase and pyroxene Although the clast has not beenbrecciated shock effects are apparent in the irregularextinction and micro-fracturing that are especially visible inplagioclase (Fig 1) Modal analysis of the clast givesplagioclase 705 vol mafic silicates 282 vol ilmenite +chromite 07 vol and FeNi metal 07 vol (610 points)Plagioclase and mafic silicate compositions in 67215c aresimilar to those of lunar ferroan anorthosites (Fig 3) Al andTi contents of pyroxenes in 67215c fall along a 21 cation

trend (Fig 4) indicating a TiVIAl2IV substitution mechanismconsistent with crystallization from a plagioclase-saturatedmagma FeNi metal compositions have relatively high Ni (3ndash5 wt) and Co (1 wt) contents and low NiCo ratios whichis atypical for lunar ferroan anorthosites (cf Ryder et al1980)

Major and Trace Elements

The clast and host breccia samples of 67215 and the splitof 62236 analyzed for this study all have similar bulkcompositions that reflect the relatively high modal abundanceof pyroxene and olivine in these rocks (27ndash28 wt Al2O3 5ndash8 wt FeO Table 1) The fused bead prepared from 62236was internally heterogeneous with Al2O3 ranging from 255ndash312 wt and other elements showing correlated variationsindicating incomplete mixing between plagioclase and maficminerals Nonetheless the average composition of the glassbead probably provides a reasonable estimate of the bulkcomposition of 62236 considering the close agreement withCaO and Na2O contents measured independently by INAA(Table 1) The similarity in composition between the 67215clast and the host breccia also extends to trace elementabundances which are almost identical for manyincompatible elements (Table 1 Fig 5) The major and traceelement data therefore support previous suggestions basedon mineral chemistry that breccia 67215 is composedpredominantly of a single ferroan noritic anorthosite lithology(McGee 1988)

Incompatible trace element concentrations in 67215c areelevated compared to 62236 and most other lunar ferroananorthosites (cf Norman and Ryder 1979 James 1980) butbelow those of 67016c (Fig 6) As expected from the wholerock compositions the plagioclase and the mafic fractions of67215c also have high concentrations of Nd and Sm (Table2) In this respect 67215c is similar to 67016c (Alibert et al1994) and distinct from both 62236 (Borg et al 1999) and60025 (Carlson and Lugmair 1988) which have trace elementcompositions more typical of lunar FANs Strontiumconcentrations in plagioclase from 67215c (176 ppm) arewithin the range observed for lunar ferroan anorthosites(Hubbard et al 1971 Floss et al 1998 Papike et al 1997Carlson and Lugmair 1988 Alibert et al 1994 Borg et al1999) The INAA data for mineral separates taken from62236 show that the mafic fraction of this rock is stronglydepleted in LREE and enriched in HREE relative toplagioclase with the whole rock split having a compositionconsistent with a mixture of ~80ndash85 wt plagioclase and 15ndash20 wt mafics

The Ni and Co contents of 67215c exceed thosecommonly found in more plagioclase-rich samples of ferroananorthosite but similar values were also measured in 62236(Table 1) and probably relate to the greater abundance ofmafic phases in these noritic anorthosites The low NiCo

Descartes breccia 67215 Clues to the history of the lunar crust 649

ratios of the 67215 clast (NiCo = 31) and host breccia (NiCo= 28) are similar to that of metal grains in the clast The twosplits of 62236 analyzed by ICPMS and INAA have disparateNi (23 versus 87 ppm) and Co (187 versus 85 ppm) contentsbut similar NiCo ratios (10ndash12)

Sm-Nd Isotopes

The 147Sm-143Nd isotopic compositions of mineralseparates from 67215c yield an isochron age of 440 plusmn 011Gyr with an initial e143

Nd of +085 plusmn 053 (Fig 7) Correctionsfor neutron exposure at the lunar surface were not necessarybecause Sm isotopic compositions measured on a split of thehost breccia were normal The positive initial Nd indicates asource region which experienced long-term LREE-depletionrelative to chondritic reference values (CHUR 147Sm144Nd =01967 143Nd144Nd = 0511847 Wasserburg et al 1981)67215c is the fourth lunar ferroan noritic anorthosite to yield

a Sm-Nd internal isochron age Other previously datedsamples include 60025 (444 plusmn 002 Gyr Carlson andLugmair 1988) a clast from breccia 67016 (453 plusmn 012 GyrAlibert et al 1994) and 62236 (429 plusmn 006 Gyr Borg et al1999) The 147Sm143Nd ratios of mineral separates from67215c are similar to those in 67016c (Alibert et al 1994)with the mafic fractions having less extreme compositionsthan 62236 (Borg et al 1999) and 60025 (Carlson andLugmair 1988) Notable is the presence of a leachablecomponent with highly radiogenic 143Nd144Nd isotopiccompositions in both the plagioclase and the pyroxenefractions of 67215c (Table 2) which was not included in theisochron calculation

39Ar-40Ar Isotopes

For 67215c the Ar-Ar ages generally increase slowly asthe Ar extraction proceeds (Fig 8a) A few of the early

Fig 1 Photomicrographs of the ferroan noritic anorthosite clast from breccia 67215 a) thin section 67215 55 showing the contact betweenthe clast analyzed for this study on the left and the host fragmental breccia on the right b) a plagioclase grain in crossed-nichols showing theshock features and igneous twinning c) close up view of the clast illustrating the well preserved igneous texture d) backscatter image of apyroxene grain showing the fine-scale exsolution lamellae indicating relatively rapid cooling and shallow depth of emplacement

650 M D Norman et al

extractions show higher apparent ages probably due to therelease of adsorbed atmospheric 40Ar and 40Ar released duringmelting of the Al foil that contained the sample This slopedage spectrum is what would be expected if the clast had lost asmall amount of its 40Ar by diffusion The clast age prior tosuch diffusive loss would be given by the average age of 393plusmn 002 Gyr defined by the four extractions releasing ~62ndash100 of the 39Ar This is identical to the Ar-Ar age of 62236(393 plusmn 004 Borg et al 1999) Concentrations of K and Ca forboth are essentially identical at 140 ppm K and 110ndash112 Caand agree with the determinations given in Table 1

The Ar-Ar age spectrum for the matrix sample (Fig 8b)is more complex While the matrix suggests older ages thanthe clast at high extraction temperatures it also shows evengreater diffusive loss of 40Ar The shape of the matrix agespectrum suggests that two different 40Ar diffusion-lossprofiles produce these sloped ages one occurring over ~8ndash68 39Ar release and another occurring over ~75ndash10039Ar These diffusion profiles correlate with 2 apparent peaksin the rate of release of Ar as a function of temperature andsuggest the presence of 2 populations of K-bearing grains thatdiffer slightly in their Ar diffusion properties The heatingevent that produced 40Ar diffusive loss in the matrix samplewas later than 38 Gyr ago and possibly much later Thisheating event may not have been experienced by the clastThe time of assembly of the 67215 breccia is unknown andthe greater 40Ar loss shown by the matrix sample may haveoccurred prior to this brecciation event

Yet another uncertainty affects the Ar-Ar ages For bothclast and matrix samples the 36Ar38Ar ratios for mostextractions are larger than the ratio of ~07 expected fromcosmic-ray production and indicate the presence of solar-wind 36Ar and probably lunar-atmosphere 40Ar The trapped40Ar36Ar ratio in 67215 is unknown and the Ar-Ar ldquoplateaurdquoage of 393 Gyr for the clast was determined by correcting fortrapped 40Ar36Ar = 50 Trapped 40Ar36Ar ratios for several

Apollo 16 regolith breccias lie in the range of lt1 to ~12 andthis ratio tends to be significantly larger for lunar samples thatacquired their trapped gases early in lunar history (McKay etal 1986) An isochron plot (R2 = 0994) of 40Ar36Ar versus39Ar36Ar for 9 extractions of the clast releasing ~17ndash100 ofthe 39Ar gives an age of 406 Gyr and a trapped 40Ar36Arintercept of ndash52 plusmn 28 This negative intercept (and theassociated age) is without merit and gives no insight as to thetrapped Ar composition If we assume a trapped 40Ar36Arratio of 1 the age plateau for the clast would be 399 Gyr Ifwe assume trapped 40Ar36Ar = 10 this plateau age becomes385 Gyr Ar-Ar ages of the matrix would require similarcorrections except that the few highest temperatureextractions require no correction and thus indicate thatmatrix ages of 39 Gyr are real

All of these Ar-Ar ages are younger than the Sm-Ndisochron age because of the impact heating history of 67215We conclude that the 67215 clast was last completely degassed393 plusmn 008 Gyr ago where the greatest contribution to theuncertainty in this age derives from a correction for lunaratmosphere 40Ar The Ar-Ar age spectra for the matrix and theclast suggest that subsequent milder heating events also

Table 2 Sm-Nd isotopic compositions for lunar sample 6721546

SampleWeight(mg)

Sm (ppm)

Nd (ppm) 147Sm144Nd 143Nd144Nd

WR 11999 03938 1237 019258 plusmn 19 0511767 plusmn 10WR1 4393 09286 2812 019967 plusmn 20 0511928 plusmn 10WR2(lt325mesh)

932 1294 3900 020056 plusmn 20 0511982 plusmn 12

Plag1 3078 ndash ndash ndash ndash(residue) ndash 02698 1355 012042 plusmn 12 0509695 plusmn 14(leachate) ndash ndash ndash 014532 plusmn 29 0510851 plusmn 20Plag2 2266 ndash ndash ndash ndash(residue) ndash 04295 1748 014863 plusmn 15 0510500 plusmn 17(leachate) ndash ndash ndash 015689 plusmn 37 0511490 plusmn 11Px + Ol 940 ndash ndash ndash ndash(residue) ndash 1466 3697 023977 plusmn 24 0513204 plusmn 14(leachate) ndash ndash ndash 016516 plusmn 55 0515928 plusmn 17Px 698 ndash ndash ndash ndash(residue) ndash 3886 9152 025677 plusmn 26 0513660 plusmn 13(leachate) ndash ndash ndash 020532 plusmn 168 0512532 plusmn 10

Fig 2 Line scan electron microprobe traverses taken at 1 micronsteps across two pyroxene grains in 67215c demonstrating theexistence of fine (2ndash5 micron) exsolution lamellae in both (a) low-Caand (b) high-Ca pyroxenes

Descartes breccia 67215 Clues to the history of the lunar crust 651

occurred This Ar-Ar age of the clast could be consistent withthe formation of one of several large lunar basins For examplepreferred (but not uncontested) lunar basin ages are ~392 Gyrfor Nectaris ~385ndash389 Gyr for Serenitatis and Crisium and~385 Gyr for Imbrium (Stoumlffler and Ryder 2001)

Rb-Sr Isotopes

Rb-Sr isotopes in 67215c have been disturbed and thedata do not form an isochron The whole rock (WR) +plagioclase compositions are consistent with an event at 393plusmn 006 Gyr and an initial 87Sr86Sr of 0699104 plusmn 10 in therock at that time (Fig 9) Rb-Sr systematics of the maficfractions in 67215c fall to the left of the 393 Gyr referenceline indicating unsupported radiogenic Sr possibly due to lossof volatile Rb during impact metamorphism (Borg et al1999) The 87Sr86Sr ratios measured in the mafic fractions of67215c are similar to those of 67016c (Alibert et al 1994) andlower than the values measured for mafic fractions in 62236(Borg et al 1999) suggesting that the extent of Rb mobility in67215c was less than that proposed for 62236 (Borg et al1999) 87Rb86Sr ratios of plagioclase separates from 67215care similar to those of 62236 and 67016c and less than thoseof 60025 In addition to having highly radiogenic 143Nd144Ndcompositions the leachable component in 67215c also hasradiogenic 87Sr86Sr (Table 3)

DISCUSSION

Petrological and Geochemical Affinity of 67215c with theFerroan Anorthosite Suite

To extract information about the magmatic and thermalhistory of the lunar crust understanding the petrological and

geochemical context of the ferroan noritic anorthosites isnecessary 60025 and 62236 have long been recognized asmafic members of the ferroan anorthositic suite of lunarhighlands rocks and are therefore presumed to be linkedgenetically to the more common plagioclase-enrichedmembers of the suite such as 15415 and 60015 (Dixon andPapike 1975 Dymek et al 1975 Warren and Wasson 1977)Major element compositions of plagioclase and pyroxene alsoclassify 67215c and 67016c as members of the ferroananorthositic suite of lunar highlands rocks (Fig 3 and 4)Whole rock trace element compositions of the ferroan noriticanorthosites have TiSm and ScSm ratios that fall within therange defined by other lunar ferroan anorthosites and whichare distinct from the compositions of KREEP and Mg-suitenorites and troctolites (Fig 10) Plagiophile elementcompositions (eg AlEu SrGa SrEu) of the ferroan noriticanorthosites are also more like those of ferroan anorthositesthan either Mg-suite norites and troctolites or KREEP (Fig11) This is consistent with our previous conclusion that theelevated abundances of incompatible trace elements in67215c and 67016c cannot result from an admixed KREEPcomponent (Fig 6)

The possibility that the ferroan noritic anorthosites aremixed rocks is further mitigated by the unusually low NiCoratios of the metal and the bulk rocks compositions which aredistinct from those of most meteorite-contaminated lunarimpact melts (Hewins and Goldstein 1975 Ryder et al 1980)This appears to be a primary magmatic feature of the ferroannoritic anorthosites Nickel contents comparable to those ofthe ferroan noritic anorthosites (10ndash50 ppm) are notuncommon in monomict lunar rocks (Haskin and Warren1991) and similar values for indigenous lunar crustalabundances have been inferred based on mixing relations of

Fig 3 Compositions of plagioclase (An) and low-Ca pyroxene (En)in the principal suites of lunar highlands igneous rocks All four of theferroan noritic anorthosites for which Sm-Nd isochrons have beenobtained have mineral compositions consistent with a classificationof these rocks as members of the ferroan anorthositic suite

Table 3 Rb-Sr isotopic compositions for splits of lunar sample 67215

SampleWeight (mg)

Rb (ppm)

Sr (ppm) 87Rb86Sr 87Sr86Sr

6721546 (clast)WR 713 0758 1308 001676 plusmn 8 0700066 plusmn 14WR1 258 0396 1292 000887 plusmn 4 0699718 plusmn 16WR2(lt325 mesh)

932 0385 1324 000841 plusmn 4 0699677 plusmn 11

Plag1 175 0183 1760 000301 plusmn 2 0699273 plusmn 12Plag1 (leachate)

ndash ndash ndash 003048 plusmn 15 0703522 plusmn 11

Plag2 129 0165 1756 000272 plusmn 1 0699260 plusmn 11Plag2 (leachate)

ndash ndash ndash 003507 plusmn 18 0703933 plusmn 14

Px + Ol 940 ndash ndash ndash ndash(residue) ndash 00689 4839 000412 plusmn 3 0699514 plusmn 16(leachate) ndash ndash ndash 007946 plusmn 40 0704495 plusmn 29Px 698 ndash ndash ndash ndash(residue) ndash 00238 1078 000638 plusmn 11 0700231 plusmn 17(leachate) ndash ndash ndash 004158 plusmn 60 0709544 plusmn 42

6721539 (host breccia)Mag 9602 07404 1134 001890 plusmn 9 0700170 plusmn 12Non-Mag 10448 07275 1549 001359 plusmn 7 0699873 plusmn 12

652 M D Norman et al

polymict highlands breccias (Palme 1980 Korotev 1987)From the combination of mineral compositions anddiagnostic trace element signatures we conclude that thepetrological and geochemical features of 67215c and theother ferroan noritic anorthosites represent primary magmaticcharacteristics and that the petrogenesis of these rocks isclosely linked with more plagioclase-rich varieties of lunarferroan anorthosites

Compared to other ferroan anorthositic suite rocks67215c and 67016c are unusual in having more abundantaugite ilmenite and chromite raising the possibility that67215c and 67016c are part of a common magmatic systempreserved in the Descartes breccias Jolliff and Haskin (1995)showed that a set of soil particles collected from the rim ofNorth Ray Crater belong to a coherent ferroan magmaticsuite that produced bulk compositions ranging fromanorthosite to noritic anorthosite Although most of theseparticles are fragmental or impact-melt breccias themonomict varieties include noritic anorthosites with exsolvedpyroxenes and trace quantities of both Cr-spinel and ilmeniteand 67016c and 67215c are both possibly related to thismagmatic system In contrast to 67016c however thesubhedral granular texture of 67215c shows that this clast wasnot brecciated or metamorphosed significantly after itcrystallized from a melt The TiVIAl2IV substitution inferredfor pyroxenes in 67215c (Fig 4) shows that this magma wasalready saturated in plagioclase and the presence ofexsolution lamellae in both high-Ca and low-Ca pyroxene in67215c suggests that both pigeonite and augite (plus minorolivine) were magmatic phases

Thermal modeling of pyroxene compositions in 67215cindicates emplacement of the magma at very shallow (pound05km) depths in the lunar crust (McCallum et al 2002) This ismuch more shallow than previously inferred depths of other

ferroan anorthositic suite rocks such as 67075 and 60025which have pyroxene compositions consistent with cooling at14ndash20 km depth (McCallum and OrsquoBrien 1996) Lithologiesrelated to the ferroan anorthositic suite of lunar highlandsrocks appear to be distributed through the middle and uppercrust of the moon

The primary magmatic lunar crust appears to have beengrossly stratified with a relatively mafic upper crustcontaining both ferroan and more magnesian lithologiesunderlain by relatively pure ferroan anorthosite at depth Thisgeneralized view of the lunar crust is supported by remotesensing observations demonstrating regionally extensivelayers of relatively pure anorthosite at mid-crustal depth(Hawke et al 1993 2002) the compositions of lunarmeteorites (Korotev 2000) and recent studies of crustalstratigraphy based on lithologic units exposed in lunar craters(Wieczorek and Zuber 2001) In this context 67215c and67016c may represent samples of relatively shallow noriticanorthosite crust enriched in trapped liquid while otherferroan anorthositic rocks such as 60025 and 62236 mayrepresent adcumulates derived from greater depths andcontaining very little trapped liquid If all of these samplescrystallized from a common magmatic system as suggested bytheir coherent mineralogical and trace element characteristicsit must have been at least 20 km deep and probably gt45ndash60km deep to account for the lack of complementary mafic andultramafic cumulates in the lunar crust

Ba-Sr in Plagioclase Evidence Against a Depleted ParentMagma for Lunar Ferroan Anorthosites

One of the more provocative conclusions to come fromprevious Sm-Nd isotopic studies of lunar ferroan anorthositicsuite rocks is that their parental magmas were derived from

Fig 4 Cation proportions of Al and Ti in pyroxenes (per 6 oxygens) in 67215c fall along a 21 correlation line indicating crystallization ofthe pyroxenes from a magma saturated in plagioclase

Descartes breccia 67215 Clues to the history of the lunar crust 653

source regions that experienced long term depletions ofLREE (ie high SmNd) as indicated by their highly positiveinitial e143

Nd compositions (Borg et al 1999) 62236 has themost extreme initial e143

Nd composition (e143Nd = +31 at 429

Gyr) but all of the ferroan noritic anorthosites for which Sm-Nd isochrons are available show at least a modestly positiveinitial e143

Nd that is resolvably higher than the nominalchondritic reference values (Carlson and Lugmair 1988Alibert et al 1994 Borg et al 1999 Fig 7) The initial 143Ndof 62236 implies a source region with a 147Sm144Nd of 0287(Borg et al 1999) similar to that of the high-Ti mare basaltcumulate source regions (Snyder et al 1994 Nyquist et al1995) The conclusion that lunar ferroan anorthosites werederived from a fractionated and highly depleted source regionwould if confirmed provide a serious challenge to themagma ocean paradigm for lunar evolution

In contrast to the highly depleted source compositionimplied by the apparent initial 143Nd composition of 62236whole rock trace element compositions of 62236 and 67215cprovide little evidence for fractionated parental magmacompositions or severe depletions of the highly incompatibleelements (Fig 5 and 6) In addition REE compositions ofparental liquids inferred from plagioclase and pyroxenecompositions in lunar ferroan anorthosites typically show flatto slightly LREE-enriched chondrite-normalized patterns(Hubbard et al 1971 Phinney 1991 Papike et al 1997 Flosset al 1998 James et al 2002) rather than the LREE-depletionimplied by the positive initial e143

Nd compositionsTo investigate the composition of ferroan noritic

anorthosite parental magmas in greater detail we examinedthe Sr and Ba contents of plagioclase in lunar ferroananorthosites During the generation and evolution of basalticmagmas Ba behaves as a highly incompatible element(similar to La) while Sr is somewhat more compatible(similar to Nd) so that a LREE-depleted source would beexpected to also have a subchondritic BaSr ratio while

crystallization of a lunar magma ocean would be expected toproduce residual melts with a near-chondritic or slightlysuper-chondritic BaSr ratio Barium and Sr are both morecompatible in plagioclase than the REE and mineral-meltdistributions are relatively well known (Blundy and Wood1991 Bindeman et al 1998) so this approach may provide amore robust indication of parental magma compositions thanthose based on highly incompatible elements

Barium and Sr concentrations of plagioclase from lunarferroan anorthosites were compiled and melt compositions inequilibrium with the plagioclase were estimated usingmineral-melt distribution coefficients (D) calculated from theregressions of Blundy and Wood (1991) assuming T = 1500K and a plagioclase composition of An96 The resulting valuesare DSr = 1098 and DBa = 0120 Plagioclase compositionswere obtained primarily from published ion microprobestudies (Papike et al 1997 Floss et al 1998) Both of thesestudies include data for 60025 Floss et al (1998) report ionmicroprobe data for five lithic clasts from breccia 67215 butdid not include the specific clast described here Howeverbreccia 67215 is dominated by a specific ferroan noriticanorthosite lithology (McGee 1988) and the materialanalyzed by Floss et al (1998) appears to be similarpetrologically and geochemically to the clast that we studiedFor 62236 laser ablation ICP-MS data were used (Norman etal 1998b and unpublished data) For 67016c plagioclasecompositions were estimated from whole rock Sr and Ba data(Norman and Taylor 1992) assuming 70 plagioclase(Norman et al 1995) and that all of the Sr and Ba in theserocks are contained in the feldspar Neutron activation datawere not used because Ba in many lunar FANs is near thedetection limit for this technique (~10 ppm Warren andKallemeyn 1984 Palme et al 1984) and therefore subject tolarge uncertainties

For comparison a simple crystallization model wascalculated to illustrate the compositional evolution of a

Fig 5 Trace element compositions of 67215c (open circles solid line) the 67215 host breccia (open circle dashed line) 62236 (filled circles)and 67016c (open squares Norman and Taylor 1992) normalized to chondritic abundances (Anders and Grevesse 1989) and compared withthe compositions of Apollo 17 KREEP (data from Norman et al 2002)

654 M D Norman et al

chondritic magma ocean A magma ocean with initialconcentrations 3times the CI-chondritic values of Anders andGrevesse (1989) was assumed (234 ppm Sr 702 ppm BaSrBa = 333) and residual melt compositions werecalculated for 1 fractional crystallization incrementsassuming bulk distribution coefficients of DSr = 001 and DBa

= 0001 This approximates the formation of mafic cumulatesin the lunar mantle Melt compositions in equilibrium withplagioclase from lunar FANs intersect the trend for residualliquids predicted by this model at about 75 crystallization(100 ppm Sr 305 ppm Ba) and extend to higher Ba contentsand higher BaSr ratios (Fig 12) Melts with high Bacontents and high BaSr ratios can be produced by continuedevolution of the magma during which plagioclase enters thecrystallization sequence with ~25 residual liquid remainingand forms 50 of the crystallizing assemblage This isconsistent with the crystallization sequence and relativephase proportions calculated by Snyder et al (1992) for alunar magma ocean

Plagioclase compositions in most lunar FANs areconsistent with crystallization from a lunar magma ocean(Fig 12) Melt compositions inferred from the plagioclase in60025 and 62236 fall along the trend predicted for

equilibrium crystallization after plagioclase has joined thecrystallization sequence (Fig 12) In contrast melts inequilibrium with plagioclase in the 67215 clasts and 67016capparently had much higher Ba contents and very high BaSrratios unlike those of the other FANs Floss et al (1998) alsorecognized that the clasts in 67215 are not related in anysimple way to the other lunar FANs If all of the plagioclase in67215c and 67016c is a cumulus phase the melts from whichthese rocks crystallized would have been highly evolved withsome characteristics of KREEP (Fig 12) although a simplemixture of FAN with KREEP can be ruled out by the bulkrock and mineral compositions (Floss et al 1998 Fig 6)Alternatively plagioclase in 67215c and 67016c may not becompletely cumulus in origin but may contain a greaterfraction of trapped liquid than most other lunar FANsUnfortunately in this case distribution coefficients cannot beused to infer trace element concentrations in the parentalliquid without an independent estimate of the amount oftrapped melt Regardless of their exact origin thesuperchondritic BaSr of 67215c and 67016c shows that theycannot have been derived from a subchondritic or depletedsource

Although it is apparent that formation of lunar ferroananorthosites and related rocks was a complex process weconclude that 1) there is no compelling evidence in the traceelement compositions of these rocks for a source that washighly depleted in incompatible trace elements and 2) the Ba-Sr relations in plagioclase from lunar ferroan anorthosites andrelated rocks are largely consistent with crystallization of anear-chondritic magma ocean The fact that Ba and Srcontents of melts parental to FANs cluster between theequilibrium and fractional crystallization trajectories (Fig12) may indicate complexities in magma ocean evolution thatcould be traced through additional petrological andgeochemical studies (Raedeke and McCallum 1979 Longhi2002)

Disturbance in the Plagioclase Fraction Effects on Sm-NdIsochrons

Accepting the nominal Sm-Nd isochron ages for theferroan noritic anorthosites at face value would imply anextended formation interval for ferroan anorthosites thatspans the first 250 million years of lunar history (454ndash429Gyr) Thermal models of lunar evolution do not rule out thispossibility but do suggest that this is near the limit expectedfor crystallization of a magma ocean even when theinsulating effects of a megaregolith on the primordial crustare considered (Warren et al 1991 Longhi 2002) This rangeof ages would imply that crystallization of ferroananorthosites continued until the time of Nd isotopic closure ofthe lunar interior as indicated by the isotopic compositions oflunar mare basalts (~432 Gyr Nyquist et al 1995) andKREEP model ages (~436 Gyr Carlson and Lugmair 1979)

Fig 6 A comparison of ZrHf and NbTa ratios versus Laconcentrations predicted for mixtures of a ferroan anorthosite withlow incompatible element concentrations (62236) and KREEP withthose measured in the 67215 clast and host breccia and 67016c(Norman and Taylor 1992) The relatively unfractionated ZrHf andNbTa ratios in 67215 and 67016c show that the elevatedconcentrations of incompatible elements in these samples cannot bedue to admixture of a KREEP component The KREEP compositionis the average of 11 Apollo 17 poikilitic impact melt rocks given byNorman et al (2002) The compositions of 62236 and 67215 are fromthis study

Descartes breccia 67215 Clues to the history of the lunar crust 655

The range of ages and the LREE-depleted nature of theferroan anorthosite source region that is implicit in thepositive initial 143Nd of these rocks clearly would require anew view of magmatism and crustal evolution on the moonThis interpretation has not been universally accepted Forexample Shearer et al (2002) explored various alternativesand suggested that the young age and elevated initial 143Nd of62236 could be ldquototally meaninglessrdquo due to re-equilibrationmixing or disturbance during large impact or thermalmetamorphic events

Borg et al (1999) considered it unlikely that disturbanceor re-equilibration of the Sm-Nd isotopic systematics in 62236would produce either an isochron or an elevated initial e143

NdHowever examples of unrealistically positive initial e143

Nd

isotopic compositions indicated by 147Sm-143Nd mineralisochrons have also been found in primitive meteorites(Prinzhofer et al 1992 Yamaguchi et al 2001) In the study byPrinzhofer et al (1992) well-behaved Sm-Nd isochrons gaveinitial 143Nd values of +16 at 446 Gyr and +21 at 447 Gyrfor the eucrite Ibitira and silicate phases from the mesosideriteMorristown respectively These initial 143Nd values wouldimply source compositions even more fractionated and LREE-depleted than that inferred for 62236 despite the relativelyunfractionated SmNd ratio and REE pattern of the eucrite andrelatively modest LREE depletion in Morristown Prinzhoferet al (1992) concluded that the 147Sm-143Nd mineral isochronsof Ibitera and Morristown were disturbed and they present amodel in which plagioclase partially equilibrated withcogenetic phosphate at a younger time while pyroxenesretained their primary isotopic compositions This mechanismis capable of producing partially rotated isochrons that yield

Fig 7 Mineral separates from the 67215 ferroan noritic anorthosite clast form a 147Sm-143Nd isochron indicating an igneous crystallizationage of 440 + 011 Gyr and an initial e143

Nd isotopic composition of +085 plusmn 053

Fig 8 39Ar-40Ar ages (rectangles) and KCa ratios (stepped line)plotted against cumulative release of 39Ar from stepwise temperatureextractions of samples of (a) the 67215 clast and (b) a matrix samplefrom the 67215 fragmental host breccia The release profile from the67215 clast is consistent with an impact heating event at 393 plusmn 008Gyr with the greatest contribution to the quoted error arising fromuncertainty over the composition of trapped lunar argon

656 M D Norman et al

shallower slopes and therefore artificially younger ages withpositive apparent initial 143Nd values Evidence for thisprocess in Ibitira and Morristown include a wide scatter of147Sm-143Nd model ages in the plagioclase fractions (418ndash463 Gyr) compared to the pyroxene fractions (456ndash459 GyrPrinzhofer et al 1992) Yamaguchi et al (2001) alsoconcluded that the positive initial 143Nd for eucrite EET 90020(+1 plusmn 05) reflects transfer of radiogenic 143Nd into theplagioclase during a subsequent heating event

We suggest that a similar process may have affected the147Sm-143Nd systematics of at least some of the ferroan noriticlunar anorthosites and that the young age and highly positiveinitial 143Nd of 62236 in particular may reflect open systembehavior of plagioclase subsequent to its crystallizationEvidence for this is provided by a comparison of 147Sm-143Ndmodel ages in coexisting plagioclase and pyroxene For thiscomparison we normalized the Sm-Nd isotopic data for67215c (this study) 62236 (Borg et al 1999) 67016c (Alibertet al 1994) and 60025 (Carlson and Lugmair 1988) to acommon basis and calculated model ages for the plagioclaseand pyroxene fractions relative to the Murchisoncarbonaceous chondrite analyzed by Alibert et al (1994)Model ages for plagioclase in the noritic anorthosites span agreater range (394ndash483 Gyr) than the pyroxenes (441ndash460Gyr) (Fig 13) but the average model age is similar for boththe plagioclase (442 Gyr) and pyroxene (447 Gyr) fractionsof these rocks 62236 and 67016c show relatively largedifferences in model ages for plagioclase compared to themafic fractions of these rocks with plagioclase model agesdisplaced to younger and older values respectively (Fig 13)

In contrast model ages for plagioclase from 67215c and60025 agree reasonably well with those of the pyroxenesfrom these samples

One possibility is that the wide range of model ages in theplagioclase fraction of these rocks reflects multi-stageevolution of the ferroan noritic anorthosites and derivationfrom sources that ranged from strongly LREE-depleted(62236) to relatively chondritic (67215c 60025) to LREE-enriched (67016c) However there is no compelling evidencefor such a diversity of source compositions in the mineralogyor trace element compositions of lunar FANs Alternativelythe 147Sm-143Nd systematics of plagioclase in 67016c and62236 may have been disturbed producing isochron ageswhich are too old and too young respectively The presenceof disturbed components in plagioclase from 67016c wasrecognized by Alibert et al (1994) who showed that if theywere included in the isochron age calculation someplagioclase fractions from this clast have anomalously low143Nd144Nd compositions producing an unrealistically oldage and a negative initial e143

Nd isotopic composition (e143Nd =

ndash13 at 466 Gyr) while the pyroxenes were only marginallyaffected by this disturbance

However the comparison between primitive meteoritesand the lunar noritic anorthosites may be flawed because themeteorites contain REE-enriched phosphates that provideconsiderable leverage for partial re-equilibration with theplagioclase (Prinzhofer et al 1992 Yamaguchi et al 2001) Incontrast no phosphate has ever been found in any lunarferroan anorthosite including 67215c despite a specific anddetailed search for such phases in the interstitial regions of

Fig 9 Rb-Sr isotopic compositions of mineral separates in the 67215 ferroan noritic anorthosite have been disturbed and do not form anisochron A plagioclase-whole rock tie line indicates an event at 393 plusmn 006 Gyr consistent with the Ar-Ar age

Descartes breccia 67215 Clues to the history of the lunar crust 657

this clast using backscatter imaging and electron microprobeanalysis In addition Prinzhofer et al (1992) assumed thatIbitira had been shock heated at a time much later than thetrue crystallization age (ie lt4 Gyr ago) while this meteoriteactually shows an Ar-Ar age of 449 Gyr (Bogard andGarrison 1995) An extended discussion of the origin of theNd isotopic disturbance in the primitive meteorites is beyondthe scope of the current study but evidence exists for thepresence of cryptic trace element-enriched phases orcomponents in at least some of the lunar noritic anorthositeswhich might provide the necessary leverage to account for aNd isotopic disturbance of the plagioclase in these rocksAlthough the brecciated nature of 62236 makes a detailedsearch for such phases difficult laser ablation ICPMS studiesof mineral separates from this sample show thatclinopyroxenes in this rock are highly enriched in REE andother incompatible elements compared to plagioclase(Norman et al 1998b) These trace element-enriched

pyroxenes cannot represent cumulus phases thatcocrystallized with the plagioclase in this rock Alternativelythey may reflect late-stage crystallization of more evolvedmelt or exsolution from original pigeonite rather than primarycumulate phases (James et al 2002) If these pyroxenes havehighly radiogenic 143Nd144Nd as predicted from their REEpatterns partial equilibration might account for the disturbedNd isotopic compositions of plagioclase in 62236 The textureof 67106c also indicates a complex history involvingbrecciation and recrystallization (Norman et al 1991 1995)and a cryptic phase with elevated Sm-Nd concentrations wasinvoked by Alibert et al (1994) to explain the disturbedplagioclase compositions in 67016c

There may also be evidence of a suitable crypticcomponent in the Nd isotopic data from 67215c Allplagioclase and mafic fractions from this clast contain aleachable component that is more LREE-enriched than thewhole rock with the leachate from the Px + Ol fractionhaving a highly radiogenic 143Nd144Nd (Table 2) The originof this leachable component and possible relations to otherphases in this rock are unclear but isotopic exchange orphysical transfer of a small amount of such material into theplagioclase during an impact or thermal metamorphic eventmight be sufficient to produce a rotated isochron with anelevated apparent initial 143Nd isotopic composition such asthat observed in 62236 Unfortunately the lack ofconcentration data for this component in 67215 makes itdifficult to model such a process quantitatively

Formation Age of Lunar Ferroan Anorthosites A MaficArray

The narrow range of model ages for pyroxenes in thenoritic anorthosites raises the possibility that all of these

Fig 10 A comparison of the diagnostic trace element rations TiSmand ScSm versus FeOMgO in 67215c 60025 62236 and 67016cwith the compositions of the principal suites of lunar highlandsigneous rocks The trace element ratios are consistent with aclassification of these rocks as relatively mafic members of theferroan anorthositic suite The trace element compositions of theferroan noritic anorthosites demonstrate a lack of KREEP componentor affinity with the Mg-suite of lunar highlands cumulates

Fig 11 Plagiophile trace element ratios such as AlEu SrEu and SrGa are consistent with a geochemical affinity of 67215c 6002562236 and 67016c with the ferroan anorthositic suite of lunarhighlands rocks

658 M D Norman et al

samples may be related to a common crystallization event A147Sm-143Nd ldquoisochronrdquo based only on the mafic fractions ofthe four ferroan noritic anorthosites gives an apparent age of446 plusmn 004 Gyr (Fig 14) This approach is valid provided thatall of these rocks crystallized from a coherent magmaticsystem with a common initial isotopic composition Althoughthe MSWD for this array is larger than would be accepted fora true isochron (MSWD = 65) we propose that this resultprovides a reasonable estimate for the crystallization age ofthe ferroan noritic anorthosites and the best estimate currentlyavailable of the time at which the ferroan anorthositic crustformed on the moon The Sm-Nd mineral isochron ages of60025 (444 plusmn 002 Gyr) 67215c (440 plusmn 011 Gyr) and67016c (454 plusmn 012 Gyr) are all consistent with this agewithin the error limits while the nominal isochron age of62236 is resolvably younger (429 plusmn 006 Gyr) The relativelylarge variation of SmNd in the mafic fractions from theserocks would not be unusual for a cogenetic suite of igneouscumulates as shown by the diverse LREE patterns observedin clinopyroxene separates from anorthosites in the Stillwaterigneous complex (Salpas et al 1983) and may reflect variableproportions of cumulus phases and trapped liquid The higherMg and more fractionated SmNd in pyroxenes from 60025and 62236 compared to 67215c and 67016c may be consistentwith a greater proportion of cumulus pyroxene in the former

The initial e143Nd indicated by the mafics-only isochron

is indistinguishable from the chondritic value albeit with a

large uncertainty due to lack of control at low SmNd ratios(e143

Nd = +08 plusmn 14 Fig 14) Using a similar approachNyquist et al (2002) showed that including recent data froman anorthositic clast in lunar meteorite Yamato-86032 (clastGC) in the array yielded a similar age (449 plusmn 009 Gyr) andan initial e143

Nd identical to the chondritic value with areduced uncertainty (e143

Nd = 00 plusmn 08) The initial e143Nd

values indicated by these arrays are consistent with thoseexpected for crystallization of a chondritic magma ocean andsuggest that positive e143

Nd values were not a universalfeature of the early anorthositic lunar crust The e143

Nd valuesof 60025 (+09 plusmn 05 Carlson and Lugmair 1988) and67215c (+08 plusmn 05 this study) are within error of the initial143Nd isotopic compositions indicated by the mafic-onlyarrays

CONCLUSIONS

The petrology geochemistry and isotope geochronologyof ferroan noritic anorthosite clast from lunar breccia 67215provides crucial clues to the age origin structure and impacthistory of the lunar highlands crust The mineralogy and traceelement composition of this clast (67215c) demonstrate aclear petrologic and geochemical affinity with the ferroananorthositic suite of lunar highlands rocks However the clastis unusual in preserving evidence for a relatively shalloworigin and greater amounts of trapped melt than is commonly

Fig 12 A comparison of the Sr and Ba compositions of melts in equilibrium with plagioclase in ferroan anorthositic suite rocks withcompositional trends predicted by crystallization models of a chondritic lunar magma ocean Most ferroan anorthositic suite rocks including60025 and 62236 (dark circles) have compositions consistent with fractional (FC) or equilibrium (EC) crystallization of an evolved lunarmagma ocean and show no evidence of a fractionated or highly depleted source with low Ba and Sr contents and low BaSr ratios such as thatimplied by the measured initial e143

Nd isotopic composition of 62236 (Borg et al 1999) The compositions of melts inferred to be in equilibriumwith plagioclase in 67215c and 67016c fall off the crystallization trends possibly due to a greater component of trapped liquid in these samplescompared to other ferroan anorthositic suite rocks

Descartes breccia 67215 Clues to the history of the lunar crust 659

found in ferroan anorthosites A 147Sm-143Nd internalisochron obtained from 67215c indicates a primarycrystallization age of 440 plusmn 011 Gyr and an initial e143

Nd of+085 plusmn 053 Crystallization ages of sup344 Gyr have now beendetermined for two ferroan noritic anorthosites from thefeldspathic fragmental breccias that were collected aroundNorth Ray crater (Alibert et al 1994 this study) TheseDescartes breccias appear to carry very old and relativelylittle modified crustal components that were derived from aregion of the moon distinct from the KREEPy near sideterrains and may be more representative of the overallcomposition of the lunar crust than would be indicated by themajority of highlands breccias from other landing sites 40Ar-39Ar and Rb-Sr isotopic compositions of plagioclase from67215c and the host breccia indicate a significant thermalevent at ~393 Gyr although uncertainties in corrections forlunar atmospheric argon preclude assignment of this age to aspecific basin

Of the 4 ferroan noritic anorthosites that have now beendated by Sm-Nd isochrons 67215c (this study) and 60025(Carlson and Lugmair 1988) appear to best retain theirprimary age characteristics while 67016c (Alibert et al 1994)and 62236 (Borg et al 1999) show evidence for isotopicdisturbance The effects of this disturbance are primarily inthe plagioclase fraction and we propose that the young age(429 plusmn 006 Gyr) and elevated initial e143

Nd of 62236 (+31plusmn 09 Borg et al 1999) is due to this disturbance The exactmechanism by which this disturbance occurs is unclear but itmay be related to redistribution of highly radiogeniccomponents such as that found in 67215c during impactmetamorphism of the lunar crust Pyroxenes in the ferroannoritic anorthosites seem to have been more resistant to thisdisturbance and yield an 147Sm-143Nd array indicating an age

of 446 plusmn 004 Gyr This may provide a robust estimate for theprimary crystallization age of ferroan anorthosites andsolidification of the early lunar crust The initial e143

Nd

indicated by the ldquomafics-onlyrdquo array is indistinguishable fromthe chondritic value with a large uncertainty (e143

Nd = 08plusmn 14) Trace element systematics of plagioclase from ferroananorthosites are also consistent with crystallization from anevolved magma ocean having near-chondritic relativeabundances of refractory lithophile elements rather than afractionated and highly depleted source

AcknowledgmentsndashThis work was supported by the NASACosmochemistry Program the Lunar and Planetary Instituteand the Australian Research Council Enlighteningdiscussions with John Longhi Randy Korotev StuMcCallum Paul Warren and Marilyn Lindstrom areappreciated as are journal reviews by Christine Floss and JeffTaylor Craig Schwandt Henry Wiesmann Chi-yu Shih DanGarrison and David Mittlefehldt provided expert laboratorysupport and advice

Editorial HandlingmdashDr Randy Korotev

REFERENCES

Alibert C Norman M D and McCulloch M T 1994 An ancientSm-Nd age for a ferroan noritic anorthosite clast from lunarbreccia 67016 Geochimica et Cosmochimica Acta 582921ndash2926

Anders E and Grevesse N 1989 Abundances of the elementsMeteoritic and solar Geochimica et Cosmochimica Acta 53197ndash214

Bindeman I N Davis A M and Drake M J 1998 Ion microprobestudy of plagioclase-basalt partition experiments at naturalconcentration levels of trace elements Geochimica etCosmochimica Acta 621175ndash1193

Fig 13 A comparison of Nd model ages relative to chondrites forplagioclase and mafic mineral fractions of the noritic ferroananorthosites 67215c 60025 62236 and 67016c Plagioclase andpyroxene fractions from 67215c and 60025 give similar model agesof ~445 Gyr while 62236 and 67016c show large discrepanciesbetween the plagioclase and pyroxene fractions Plagioclase in 62236and 67016c may have been disturbed possibly by impactmetamorphism while the mafic fraction in all of these rocksremained relatively resistant to this disturbance

Fig 14 The Sm-Nd isotopic compositions of mafic fractions from theferroan noritic anorthosites 67215c 62236 60025 and 67016c forman array indicating an age of 446 plusmn 004 Gyr and an initial e143

Ndvalue indistinguishable from chondritic This may represent arelatively robust estimate for the crystallization age of the ferroananorthositic lunar crust The isochron was calculated using theISOPLOT program (Ludwig 1999)

660 M D Norman et al

Blundy J D and Wood B J 1991 Crystal-chemical controls on thepartitioning of Sr and Ba between plagioclase feldspar silicatemelts and hydrothermal solutions Geochimica etCosmochimica Acta 55193ndash209

Bogard D D and Garrison D H 1995 39Ar-40Ar age of Ibitiraeucrite and constraints on the time of pyroxene equilibrationGeochimica et Cosmochimica Acta 594317ndash4322

Bogard D D Garrison D H Norman M Scott E R D and KeilK 1995 39Ar-40Ar age and petrology of Chico Large-scaleimpact melting on the L chondrite parent body Geochimica etCosmochimica Acta 591383ndash1399

Borg L Norman M Nyquist L Bogard D Snyder G Taylor Land Lindstrom M 1999 Isotopic studies of ferroan anorthosite62236 A young lunar crustal rock from a light-rare-earth-element-depleted source Geochimica et Cosmochimica Acta 632679ndash2691

Borg L E Shearer C K Nyquist L E and Noman M D 2002Isotopic constraints on the origin of lunar ferroan anorthosites(abstract 1396) 33rd Lunar and Planetary Science ConferenceCD-ROM

Carlson R W and Lugmair G W 1979 Sm-Nd constraints on earlylunar differentiation and the evolution of KREEP Earth andPlanetary Science Letters 45123ndash132

Carlson R W and Lugmair G W 1988 The age of ferroananorthosite 60025 Oldest crust on a young moon Earth andPlanetary Science Letters 90119ndash130

Dixon J R and Papike J J 1975 Petrology of anorthosites from theDescartes region of the moon Proceedings 6th Lunar andPlanetary Science Conference pp 263ndash291

Dymek R F Albee A L and Chodos A A 1975 Comparativepetrology of lunar cumulate rocks of possible primary originDunite 74215 troctolite 76535 and anorthosite 62237Proceedings 6th Lunar and Planetary Science Conference pp304ndash341

Floss C James O B McGee J J and Crozaz G 1998 Lunarferroan anorthosite petrogenesis Clues from trace elementdistributions in FAN subgroups Geochimica et CosmochimicaActa 621255ndash1283

Haskin L A and Warren P H 1991 Lunar chemistry In Lunarsourcebook edited by Heiken G H Vaniman D T and FrenchB New York Cambridge University Press pp 357ndash474

Hawke B R Peterson C A Lucey P G Taylor G J Blewett D TCampbell B A Coombs C R and Spudis P D 1993 Remotesensing of the terrain northwest of Humorum Basin GeophysicalResearch Letters 20419ndash422

Hawke B R Peterson C A Blewett D T Bussey D B J LuceyP G Taylor G J and Spudis P D Forthcoming The distributionand modes of occurrence of lunar anorthosite GeophysicalResearch Letters

Hewins R H and Goldstein J I 1975 The provenance of metal inanorthositic rocks Proceedings 6th Lunar and Planetary ScienceConference pp 343ndash362

Hubbard N J Gast P W Meyer C Nyquist L E and Shih C 1971Chemical composition of lunar anorthosites and their parentliquids Earth and Planetary Science Letters 1371ndash75

James O B 1980 Rocks of the early lunar crust Proceedings 11thLunar and Planetary Science Conference pp 365ndash393

James O B Floss C and McGee J J 2002 Rare earth elementvariations resulting from inversion of pigeonite and subsolidusreequilibration in lunar ferroan anorthosites Geochimica etCosmochimica Acta 651269ndash1284

Jolliff B L and Haskin L A 1995 Cogenetic rock fragments froma lunar soil Evidence of a ferroan noritic-anorthosite pluton onthe moon Geochimica et Cosmochimica Acta 59

Korotev R L 1987 The nature of the meteoritic components of

Apollo 16 soil as inferred from correlations of iron cobaltiridium and gold with nickel Proceedings 17th Lunar andPlanetary Science Conference Journal of Geophysical Research92E447ndashE461

Korotev R L 2000 The great lunar hot-spot and the composition andorigin of Apollo mafic (ldquoLKFMrdquo) impact-melt breccias Journalof Geophysical Research 1054317ndash4345

Lindstrom M M and Lindstrom D J 1986 Lunar granulites andtheir precursor anorthositic norites of the early lunar crustProceedings 16th Lunar and Planetary Science ConferenceJournal of Geophysical Research 91D263ndashD276

Ludwig K R 1999 Using IsoplotEx Version 201 ageochronological toolkit for Microsoft Excel BerkeleyGeochronology Center Special Publication No 1a BerkeleyUniversity of California 47 p

Longhi J Forthcoming A new view of lunar ferroan anorthositesPost magma ocean petrogenesis Journal of GeophysicalResearch

McCallum I S and OrsquoBrien H E 1996 Stratigraphy of the lunarhighland crust Depths of burial of lunar samples from coolingrate studies American Mineralogist 811166ndash1175

McCallum I S Schwartz J M Camara F and Norman M 2002Sample 67215 An anomalous ferroan anorthosite (abstract1830) 33rd Lunar and Planetary Science Conference CD-ROM

McGee J J 1988 Petrology of brecciated ferroan noritic anorthosite67215 Proceedings 18th Lunar and Planetary ScienceConference pp 21ndash31

McKay D S Bogard D D Morris R V Korotev R L Johnson Pand Wentworth S J 1986 Apollo 16 regolith brecciasCharacterization and evidence for early formation in the mega-regolith Proceedings 16th Lunar and Planetary ScienceConference pp D277ndashD303

Mittlefehldt D W and Lindstrom M M 1993 Geochemistry andpetrology of a suite of ten Yamato HED meteorites Proceedingsof the NIPR Symposium on Antarctic Meteorites 6268-292

Norman M D and Ryder G 1979 A summary of the petrology andgeochemistry of pristine highlands rocks Proceedings 10thLunar and Planetary Science Conference pp 5310ndash559

Norman M D 1981 Petrology of suevitic lunar breccia 67016Proceedings 12th Lunar and Planetary Science Conference pp235ndash252

Norman M D and Taylor S R 1992 Geochemistry of lunar crustalrocks from breccia 67016 and the composition of the moonGeochimica et Cosmochimica Acta 561013ndash1024

Norman M D Keil K Griffin W L and Ryan C G 1995Fragments of ancient lunar crust Petrology and geochemistry offerroan noritic anorthosites from the Descartes region of themoon Geochimica et Cosmochimica Acta 59831ndash847

Norman M D Taylor G J and Keil K 1991 Additional complexityin the lunar crust Petrology of sodic anorthosites and sulfur-richferroan noritic anorthosites Geophysical Research Letters 182081ndash2084

Norman M D Griffin W L Pearson N J Garcia M O andOrsquoReilly S Y 1998a Quantitative analysis of trace elementabundances in glasses and minerals A comparison of laserablation ICPMS solution ICPMS proton microprobe andelectron microprobe data Journal of Analytical AtomicSpectrometry 13477ndash482

Norman M Borg L Nyquist L E Bogard D Snyder G Taylor Land Lindstrom M 1998b Composition and age of the lunarhighlands Petrogenesis of ferroan noritic anorthosite 62236(abstract 1551) 29th Lunar and Planetary Science ConferenceCD-ROM

Norman M D Bennett VC and Ryder G Forthcoming Targetingthe impactors Highly siderophile element signatures of lunar

Descartes breccia 67215 Clues to the history of the lunar crust 661

impact melts from Serenitatis Earth and Planetary ScienceLetters

Nyquist L E Wiesmann H Shih C Y Keith J E and Harper CL 1995 146Sm-142Nd formation interval for the lunar mantleGeochimica et Cosmochimica Acta 592817ndash2837

Nyquist L E Bogard D D Shih C Y and Wiesmann H 2002Negative e143

Nd in anorthositic clasts in Yamato-86032 andMAC88105 Evidence for the LMO (abstract 1289) 33rdLunar and Planetary Science Conference CD-ROM

Palme H 1980 The meteoritic contamination of terrestrial and lunarimpact melts and the problem of indigenous siderophiles in thelunar highland Proceedings 11th Lunar and Planetary ScienceConference pp 481ndash506

Palme H Spettel Wanke H Bichoff A and Stoffler D 1984 Earlydifferentiation of the moon Evidence from trace elements inplagioclase Proceedings 15th Lunar and Planetary ScienceConference Journal of Geophysical Research 89C3ndashC15

Papike J J Fowler G W and Shearer C K 1997 Evolution of thelunar crust SIMS study of plagioclase from ferroan anorthositesGeochimica et Cosmochimica Acta 612343ndash2350

Phinney W C 1991 Lunar anorthosites their equilibrium melts andthe bulk moon Proceedings 21st Lunar and Planetary ScienceConference pp 29ndash49

Prinzhofer A Papanastassiou D A and Wasserburg G J 1992Samarium-neodymium evolution of meteorites Geochimica etCosmochimica Acta 56797ndash815

Raedeke L D and McCallum I S 1979 A comparison offractionation trends in the lunar crust and the Stillwater ComplexProceedings Conference on the Lunar Highlands Crust pp 133ndash153

Ryder G Norman M D and Score R A 1980 The distinction ofpristine from meteorite-contaminated highlands rocks usingmetal compositions Proceedings 11th Lunar and PlanetaryScience Conference pp 471ndash479

Salpas P A Haskin L A and McCallum I S 1983 Stillwateranorthosites A lunar analog Proceedings 14th Lunar andPlanetary Science Conference Journal of Geophysical Research88B27ndashB39

Shearer C K Borg L E and Papike J J 2002 Trace element andpetrologic constraints on the age of lunar ferroan anorthosites(abstract 1517) 33rd Lunar and Planetary Science ConferenceCD-ROM

Snyder G A Taylor L A and Neal C R 1992 A chemical model

for generating the sources of mare basalts Combined equilibriumand fractional crystallization of the lunar magmasphereGeochimica et Cosmochimica Acta 563809ndash3823

Snyder G A Lee C C Taylor L A Halliday A N and Jerde EA 1994 Evolution of the upper mantle of the Earthrsquos moonNeodymium and strontium isotopic constraints from high-Timare basalts Geochimica et Cosmochimica Acta 584795ndash4808

Spudis PD 1984 Apollo 16 site geology and impact meltsImplications for the geologic history of the lunar highlandsProceedings 15th Lunar and Planetary Science ConferenceJournal of Geophysical Research 89C95ndash107

Stoumlffler D and Ryder G 2001 Stratigraphy and isotope ages of lunargeologic units Chronological standard for the inner SolarSystem Space Science Reviews 969ndash54

Warren P H 2001 Early lunar crustal genesis The ferroananorthosite epsilon-neodymium paradox as a possible result ofcrustal overturn Meteoritics amp Planetary Science 36A219

Warren P H and Kallemeyn G W 1984 Pristine rocks (8th foray)ldquoPlagiophilerdquo element ratios crustal genesis and the bulkcomposition of the Moon Proceedings 15th Lunar and PlanetaryScience Conference Journal of Geophysical Research 89C16ndashC24

Warren P H and Wasson J T 1977 Compositional-petrographicinvestigation of pristine nonmare rocks Proceedings 8th Lunarand Planetary Science Conference pp 2215ndash2235

Warren P H Haack H and Rasmussen K L 1991 Megaregolithinsulation and the duration of cooling to isotopic closure withindifferentiated asteroids and the moon Journal of GeophysicalResearch 965909ndash5923

Wasserburg G J Jacobsen S B DePaolo D J McCulloch M Tand Wen T 1981 Precise determination of SmNd ratios Sm andNd isotopic abundances in standard solutions Geochimica etCosmochimica Acta 452311ndash2323

Wieczorek M A and Zuber M T 2001 The composition and originof the lunar crust Constraints from central peaks and crustalthickness modeling Geophysical Research Letters 284023ndash4026

Yamaguchi A Taylor G J Keil K Floss C Crozaz G Nyquist LE Bogard D D Garrison D H Reese Y D Wiesmann H andShih C Y 2001 Post-crystallization reheating and partial meltingof eucrite EET 90020 by impact into the hot crust of asteroid 4Vesta ~450 Gyr ago Geochimica et Cosmochimica Acta 653577ndash3599