Evidence for a predominantly non-solar origin of nitrogen in the lunar regolith revealed by single grain analyses

ELSEVIER Earth and Planetary Science Letters 167 (1999) 47–60

Evidence for a predominantly non-solar origin of nitrogen in the lunarregolith revealed by single grain analyses

Rainer Wieler a,Ł, Franck Humbert b, Bernard Marty b

a ETH Zurich, Isotope Geology, NO C61, CH-8092, Zurich, Switzerlandb Centre de Recherches Petrographiques et Geochimiques, 15 rue Notre-Dame des Pauvres,

BP20, F-54501, Vandoeuvre les Nancy, France

Received 12 August 1998; revised version received 21 December 1998; accepted 9 January 1999

Abstract

We measured the amounts of 14N and 36Ar in single mineral and glass grains from a lunar soil by laser extraction, with thegoal of studying the controversial origin of trapped nitrogen in the lunar regolith. The average 14N=36Ar ratio of 29 ilmenitegrains is 379, similar to the value determined previously on a large ilmenite separate from the same soil and 10 times largerthan the solar ratio of 37. However, the 14N=36Ar ratios in the individual grains vary between 1 and 440 times the solar ratio.36Ar amounts in the ilmenite grains scatter by more than two orders of magnitude, N amounts by less than a factor of 6. Thevariability of the 14N=36Ar ratio forms a striking contrast to the very uniform relative abundances of Ar, Kr, and Xe trappedfrom the solar corpuscular radiation observed earlier in ilmenite and other mineral grains from the same soil. This stronglysuggests that, on average, some 90% of the N in the grains has a non-solar source, contrary to the often expressed view thatessentially all N in the lunar regolith has been trapped from the solar wind. The conclusion that the lunar regolith testifies toa secular variation of the N isotopic composition in the solar wind of ¾30% becomes thus highly questionable. The originof the bulk of trapped lunar nitrogen remains unknown. 1999 Elsevier Science B.V. All rights reserved.

Keywords: lunar soils; N-14; Ar-36; solar wind

1. Introduction

The provenance of nitrogen trapped in lunar soilsis still contentious. The solar corpuscular radiationhas been postulated by many workers to be by far themajor contributor (e.g. [1–7]). We will refer to thissource as ‘solar wind’, although in analogy to thenoble gases [8] it may encompass a higher energycomponent [7], the so-called ‘solar energetic parti-cles’. The main reasons to favor a solar wind source

Ł Corresponding author. Tel.: C41 1 632 3732; Fax: C41 1 6321179; E-mail: [email protected]

have been the facts that nitrogen concentrations inbulk soil samples correlate relatively well with thoseof solar wind implanted noble gases and other ma-turity parameters [6], and that N resides near grainsurfaces [1,9,10], similar to the solar noble gases.However, an observation difficult to reconcile with asolar origin of lunar nitrogen is that N=Ar and N=Xeratios are higher than values inferred for the sunby factors of about ten and three, respectively (cf.[5,6]). Proponents of a solar origin for most of thetrapped lunar N usually explained this by a depletionof the heavy solar noble gases relative to N duringor after trapping, such that Ar would have been lost

0012-821X/99/$ – see front matter 1999 Elsevier Science B.V. All rights reserved.PII: S 0 0 1 2 - 8 2 1 X ( 9 9 ) 0 0 0 1 2 - 6

48 R. Wieler et al. / Earth and Planetary Science Letters 167 (1999) 47–60

more efficiently than Xe. A large enrichment of N inthe solar wind source region has also been discussed[12–14].

A second crucial but ill-understood observation isthat the 15N=14N ratio of trapped N varies by up tomore than 30% in different bulk samples as well asin different extraction steps of a given sample [2,4–7,9,11]. This has often been taken to indicate a secularvariation of the isotopic composition of N in the solarwind [2–4,9], although no generally accepted mech-anism to achieve such a variation has been proposed.More recently, two different solar N components witha composition depending on energy have also beenconsidered [11]. Non-solar trapped N components, eg. meteoritic or lunar indigenous nitrogen, were usu-ally considered to be minor, such that the ‘classical’picture [6] has been that N in the lunar regolith isdominated by the solar wind fraction.

This view was challenged by Geiss and Bochsler[15] and Signer and co-workers [16]. Geiss andBochsler argued that none of the conceived mecha-nisms to produce a 30% variation in the N isotopiccomposition in the solar wind are convincing. Theyexplain the isotopically light N released at interme-diate temperatures by a lunar indigenous componentoutgassed from the Moon’s interior and subsequentlyretrapped in the regolith. Later, Geiss and Bochsler[17] suggested that N from the upper terrestrialatmosphere may have been implanted into the lu-nar regolith by acceleration in the magnetosphere.Signer et al. [16] noted that the N concentrations inthe regolith are uncomfortably high for a solar originand alluded to a possible atmospheric contaminationafter sample return, although they did not addressthe question how this could have led to the observedtrend of 15N=14N with the solar wind irradiation ageof a sample.

Noble gas analyses in single lunar dust grains [18]allowed us recently to corroborate the conclusion[19] that the lunar regolith conserves the true relativeelemental abundances of the heavy noble gases inthe solar wind. Ar=Kr and Kr=Xe ratios are nearlyuniform in grains of a given sample. Hence, theobserved enrichments of Kr and Xe in lunar samplesrelative to Ar and bulk solar abundances are due toa selection process at the solar wind source and nota noble gas loss on the Moon that would fractionatethe elements, as has long been thought. This also

means that N in the regolith is overabundant by afactor of ¾10, as indicated by the N=Ar ratio, andnot merely a factor of ¾3, as deduced previouslyfrom N=Xe. A loss of solar noble gases by an orderof magnitude not affecting their elemental ratiosis difficult to imagine. We therefore extend herethe successful approach of single grain analyses tore-evaluate the provenance of N in the lunar regolith.We present 14N and 36,38Ar data on some 50 singlegrains from sample 71501.

2. Sample and experimental procedure

Soil 71501 has been exposed to the solar wind rel-atively recently. The ‘antiquity’ indicator 40Ar=36Arsuggests a surface exposure during roughly the past100 Ma [8]. Some grains may well have been ex-posed considerably earlier than this, but the constantKr=Xe ratios of all grains indicate that essentiallynone of them should contain solar wind noble gasesolder than perhaps several hundred Ma [18]. Thissoil has been widely studied for both nitrogen andnoble gases (e.g. [5,8,18–23]). For a first series ofanalyses, we chose 24 of the largest mineral andglass grains that appeared relatively clean under thebinocular microscope (plagioclase, pyroxene, blackglass and dark-green glass). These grains had sizesof ¾500–1000 µm and masses of 0.2–1.3 mg. Ina second series, we measured a more homogeneouspopulation, namely 29 ilmenite grains in the sizerange 175–250 µm and with masses on the orderof 0.04 mg, separated as described by Benkert et al.[8]. We also analysed two batches of these ilmenites,each consisting of some 20 grains.

Analyses were performed in Nancy. Seven grainsat a time were loaded into a water-cooled chamberequipped with a ZnSe window and baked overnightat 120ºC. Volatiles were extracted by melting a grainwith a CO2 laser in CW mode [24]. Purification ofthe Ar and N fractions, respectively, as well as massspectrometric procedures are described by Marty andHumbert [25].

In this work we are primarily concerned with de-termining reasonably precise 14N=36Ar ratios. 36Arblanks were about .2–5/ ð 10�16 moles. This corre-sponds mostly to less than a few % of the 36Ar ina grain, and therefore Ar blanks constitute only for

R. Wieler et al. / Earth and Planetary Science Letters 167 (1999) 47–60 49

very few grains a significant source of error. Muchmore critical is the nitrogen blank. Blank analyseswere carried out very frequently, sometimes aftereach sample measurement. N2 blanks ranged be-tween .2–10/ ð 10�12 moles in the first series andmostly between .1:5–2:5/ ð 10�12 moles in the sec-ond. The blanks contribute 10–80% and 20–60% tothe total signal of a grain in series 1 and 2, respec-tively. However, nitrogen blanks were very stableduring a day, usually within š20%, irrespective ofwhether the laser was shot on an empty hole ora previously degassed grain. This is because blanknitrogen is mainly contributed by the purificationprocedure. This contribution has been considerablylowered by optimizing the temperature of the Pt foilthat catalyses oxidation of C-bearing species. TheN amount, and hence essentially the 14N=36Ar ra-tio, could therefore be deduced for most grains towithin better than š30% and often to within š15%(Tables 1 and 2). These numbers are rather conser-vative, assuming the blank uncertainty to be equalto the total spread of blank values within a day.Dilution experiments using calibration gas showedthat no nitrogen was lost in the system [24]. Ar andN amounts are both calibrated by a known amountof air. All volumes of the line have been measured,which allows computation of sample gas amountsfrom volume dilution factors and analyser sensitiv-ity, the latter being determined by the air pipette.In order to check the accuracy of the method, weanalysed mg-sized aliquots of the K–Ar standardMMhb-1. The mean of 1:629ð 10�9 moles of radio-genic 40Ar*=g is just 1.8‰ above the recommendedvalue [26].

3. Results

Nitrogen and argon amounts and 36Ar=38Ar ratiosare given in Tables 1 and 2. 40Ar in each grain is ofthe order of 2 ð 10�14 moles, but uncertainties aretoo large to report 40Ar values. For the same reason,we report only the weighted mean nitrogen isotopiccomposition of the single ilmenites (footnote to Ta-ble 2). Also given is the solar 36Ar fraction (36Artr),corrected for the cosmic ray produced component.The correction is less than 1% for all ilmenites exceptilm. #13, in which 18% of the total 36Ar was cosmo-

genic. The grains of the first series needed on averageslightly higher corrections, but in no case more than12% (plag. #2). Corrections for cosmogenic 14N arenegligible. We will therefore discuss in the follow-ing the ratio 14Nmeas=

36Artr and refer to it simply as14N=36Ar. The stated errors of these values includethe small uncertainty of the correction for cosmogenicAr, but do not take into account the systematic uncer-tainty of the calibration of the N=Ar ratio, which webelieve to be�5% [24]. Pyroxene grain #3 apparentlycontained no N. This cannot be due to a wrong blankcorrection, since the N signal of this sample was closeto the lowest blank values measured in this series. Weconclude that the N of this grain was erroneously lostand discard this data point.

We first compare the single ilmenite data withthose of the two ilmenite grain batches and withdata obtained in a stepwise extraction run of a 29mg ilmenite separate from the same soil 71501 [5].This sample (100–150 µm grain size) had been pre-pared in Zurich by the same techniques and at thesame time as the grains here. The mean Ar and Namounts in the single grains are ¾35–80% higherthan the per-grain values of our two grain batches(Table 2), which is to be expected, since we delib-erately selected the largest crystals for the single-grain measurements. We therefore consider only the14N=36Ar ratios. The weighted average of the singlegrains of 379 is 13% higher than the value of thetotal gas of the large ilmenite separate [5], whereasthe values of our two multi-grain samples are both17% lower than that of the large separate. Since thetwo multi-grain values agree closely, the latter dif-ference might at first sight appear to be significant.However, given the large scatter of 14N=36Ar ratiosin single grains discussed below, the mean valuesof 20–25 grains should scatter much more than theclose agreement of the two analysed batches mightsuggest. The respective errors in Table 2 indicateonly the analytical precision, whereas the uncer-tainty of the mean 14N=36Ar ratio of the single grainsreflects the scatter of their gas concentrations. This1 s uncertainty of the 29 single ilmenites is ¾15%,(Table 2), and therefore, the 14N=36Ar ratios of sam-ples comprising 20–25 grains must scatter by at least15% also. Hence, the weighted mean 14N=36Ar ratioof the single grains and the ratios of the two batchesagree within error limits. Furthermore, the overall


Table 1Nitrogen and argon in single mineral and glass grains from lunar soil 71501

Grain # 14N š 36Armeas š 36Ar=38Ar 36Artr š 14N=36Artr šPlagioclaseplag. 1 10.0 1.3 7.52 0.07 4.69 7.39 0.09 135 17plag. 2 5.0 1.0 1.31 0.04 3.04 1.18 0.05 424 87plag. 3 15.3 1.6 5.33 0.05 4.60 5.21 0.06 294 31plag. 4 9.6 1.1 1.80 0.06 3.37 1.66 0.07 578 75plag. 5 15.9 1.6 3.70 0.05 3.94 3.52 0.06 452 45plag. 6 5.6 1.1 5.50 0.05 4.14 5.29 0.06 106 22plag. 7 22.5 2.3 14.46 0.10 4.64 14.18 0.14 159 16plag. 8 8.7 1.2 7.55 0.08 4.36 7.32 0.10 119 16

Pyroxenepyx. 1 44.2 4.4 13.75 0.07 5.18 13.71 0.08 322 32pyx. 2 4.7 1.1 11.53 0.05 5.17 11.49 0.07 41 9pyx. 3 (�4.4) 4.8 33.92 0.10 5.26 33.88 0.12 – 15pyx. 4 12.0 1.2 4.83 0.04 5.26 4.82 0.04 249 25pyx. 5 23.6 2.4 23.25 0.10 5.18 23.18 0.12 102 10pyx. 6 21.5 2.2 15.69 0.07 5.26 15.68 0.07 137 14pyx. 7 25.0 2.5 22.81 0.08 5.15 22.72 0.10 110 11pyx. 8 50.4 5.0 36.17 0.11 5.20 36.07 0.14 140 14

Black glassbl0gl. 1 10.8 1.2 4.20 0.04 3.40 3.87 0.07 279 30bl0gl. 2 73.1 12.2 30.9 0.15 4.90 30.5 0.25 240 40bl0gl. 3 53.5 5.5 5.53 0.06 5.15 5.51 0.09 971 101bl0gl. 4 23.0 2.4 4.16 0.07 4.90 4.11 0.09 560 59bl0gl. 5 2.8 1.1 1.88 0.06 4.10 1.81 0.08 155 60

Green glassgr0gl. 1 16.0 1.6 3.18 0.06 3.50 2.95 0.07 542 54gr0gl. 2 17.3 1.7 14.92 0.08 4.36 14.47 0.12 120 12gr0gl. 3 12.7 1.3 8.34 0.06 4.31 8.07 0.10 157 17

14N in 10�12 mole=grain, 36Ar in 10�14 mole=grain. Grain size: 500–1000 µm.Trapped 36Artr corrected for cosmogenic component, assuming (36Ar=38Ar)tr D 5.3, (36Ar=38Ar)cos D 0.65.Errors include blank and measurement uncertainty, but not uncertainties of calibration (see text).

mean value of 323 of all ilmenites analysed here iswithin 4% the same as the 14N=36Ar ratio measuredin the large separate [5].

Values of δ15N (per mil deviation of 15N=14N fromthe atmospheric ratio) of individual grains have con-siderable uncertainties, on the order of 100–150‰.Yet, the weighted mean δ15N of the single ilmenitegrains is known to reasonable precision. The value ofC64‰ is close to that of 47.7‰ of the large ilmeniteseparate measured by stepwise extraction [5] as wellas the two values of our multi-ilmenite samples(76‰ and 50‰, respectively). The δ15N measuredby Frick et al. [5] is actually expected to be lowerby about 10‰ than values obtained in this work, be-cause the grains in the large sample had a somewhat

higher surface to volume ratio than those measuredhere, and therefore a somewhat higher contributionof trapped N relative to cosmogenic 15N. Hence, de-spite the large uncertainties of the individual graindata, the mean N isotopic composition determined inthis work is in very good agreement with the Frick etal. value. Nominal δ15N values in the ilmenite grainsrange from �30‰ to C280‰, but due to the largeanalytical uncertainties it is unclear how much ofthis spread is real.

In summary, our data agree very well with thoseobtained on a much larger ilmenite separate by step-wise gas extraction [5]. This is further evidence thatthe data reported here are reliable, despite the oftenlarge N blank corrections.


Table 2Nitrogen and argon in single ilmenite grains from lunar soil 71501

Grain # 14N š 36Armeas š 36Ar=38Ar 36Artr š 14N=36Artr š2 5.5 1.1 2.873 0.020 5.20 2.865 0.020 192 383 3.77 0.94 2.185 0.020 5.17 2.177 0.020 173 444 8.6 1.1 3.261 0.020 5.23 3.255 0.020 264 368 6.86 0.69 2.666 0.013 5.03 2.646 0.013 259 269 4.93 0.69 1.215 0.013 5.03 1.206 0.013 409 57

10 4.4 1.4 2.829 0.014 4.94 2.800 0.014 157 5011 3.50 1.4 2.373 0.013 5.04 2.356 0.013 149 6013 6.69 0.94 0.050 0.006 2.79 0.041 0.006 16300 250014 3.01 0.57 0.541 0.006 5.04 0.537 0.006 561 11015 3.76 0.60 0.443 0.007 4.64 0.435 0.007 864 14516 5.58 0.89 1.642 0.025 4.87 1.622 0.025 344 5517 7.5 1.2 2.920 0.006 4.92 2.889 0.006 260 3918 5.5 1.2 2.039 0.006 4.98 2.021 0.006 273 5820 4.60 0.60 0.844 0.020 4.84 0.833 0.020 552 7222 4.19 0.59 2.170 0.022 5.06 2.156 0.022 194 2723 10.9 0.9 2.304 0.021 5.02 2.286 0.021 476 4024 3.2 1.2 0.798 0.007 4.79 0.786 0.007 410 15525 3.3 1.2 0.268 0.007 4.65 0.263 0.007 1255 44028 17.1 2.1 4.693 0.023 5.08 4.664 0.023 367 4529 6.4 1.7 1.781 0.027 5.01 1.766 0.027 362 9830 12.1 1.9 2.151 0.026 5.10 2.138 0.026 566 9031 11.4 1.9 2.499 0.025 5.09 2.484 0.025 459 7539 6.99 0.98 1.415 0.021 4.90 1.397 0.021 500 7040 6.37 0.95 1.103 0.011 5.00 1.093 0.011 583 8841 12.2 1.0 2.729 0.014 5.03 2.707 0.014 452 3742 6.87 0.96 1.356 0.014 4.87 1.338 0.014 513 7243 10.94 0.44 1.394 0.014 4.87 1.376 0.014 795 3244 4.76 0.38 0.925 0.014 4.94 0.915 0.014 520 4245 5.81 0.41 0.725 0.014 5.07 0.719 0.014 808 59

Mean 6.78 0.63 a 1.80 0.20 a 1.79 0.20 379 50 b

20 grains 76.3 1.6 27.60 0.02 4.96 27.33 0.01 279 6

25 grains 91.2 1.9 32.81 0.01 4.98 32.52 0.01 280 6

Mean of 29 C 20 C 25 grains 323

Large sample [5] 336 19

14N in 10�12 mole=grain, 36Ar in 10�14 mole=grain. Grain size 175–250 µm.δ15N values: mean of single grains: C64‰; 20-grain-batch: C76‰; 25-grain-batch: C50‰; large sample ([5]): C47.7‰.a Standard error of mean 14N and 36Ar amounts.b propagated standard errors of mean gas amounts.

Frick et al. [5] attributed to atmospheric contami-nation only the ¾9% of the total N of their ilmenitesample that was released in the first step at 300ºC.Since this sample had been heated prior to analysisat a similar temperature to our grains, it suggests thatalso only about 9% of the total N in our ilmenitegrains may be loosely bound atmospheric N that sur-vived the overnight pre-heating. This contribution is

negligible. As two reviewers pointed out, it cannotbe rigorously excluded that in lunar dust a largerfraction of N is of atmospheric origin or derivedfrom terrestrial organic contamination, but so tightlybound that it is not released by the conventional mildpre-heating. However, since the mean 14N=36Ar ratioof all grains studied here is very typical for lunarsamples, any such air-N would have affected earlier


Fig. 1. 14N=36Artr ratios vs. amounts of trapped 36Artr for 24 large (500–1000 µm) single mineral and glass grains of lunar soil 71501.Error bars reflect mainly the rather conservative estimate of the uncertainty of the N blank. One grain has a 14N=36Ar ratio identical tothe solar value given by Anders and Grevesse [27]. All other grains have considerably higher ratios which scatter widely. Data point inparentheses reflects an analysis where the N presumably was lost.

analyses to the same extent as the single grain datahere, in particular the stepwise heating run by Frickand coworkers on 71501 ilmenite. It is thus impor-tant to realize that potentially present tightly-boundair-N would not at all compromise the importanceof the present results. We will discuss below thepossibility that (part of) the non-solar trapped N weinfer to be present in lunar soils might be from theterrestrial atmosphere.

Figs. 1 and 2 show 14N=36Ar ratios versus 36Aramounts for the two series of analyses. The ratiosvary widely. The lowest value in the first set ofgrains is close to the solar 14N=36Ar ratio of 37[27], whereas the highest value is 24 times larger.In the ilmenites, the ratios range between 4 and 440times the solar value. In both figures, the highest14N=36Ar ratios are observed among the grains withthe lowest 36Ar amounts. The pronounced variabilityof 14N=36Ar contrasts strikingly with the uniformAr=Kr=Xe ratios in ilmenites and other single grainsfrom the same soil discussed in Section 4.2.

Fig. 3 displays the N and Ar amounts of theilmenites, the more homogeneous of the two grainsets. Argon amounts vary by more than a factor of100, nitrogen amounts by less than a factor of 6.

Without the very Ar-poor grain #13, the spread intrapped Ar reduces to a factor of about 11 only. Notehowever, that the very low trapped Ar concentrationof this grain cannot be due to Ar loss upon analysis.The very low 36Ar=38Ar ratio of this grain clearlyindicates a very low proportion of trapped relativeto cosmogenic Ar. It is also well documented that36Ar amounts in individual lunar ilmenites from thesame soil vary by more than two orders of magnitude[28,22,18]. In any case, it is rather the variable Arthan the more uniform N contents that cause thespread of the 14N=36Ar ratios, which is yet anotherargument that the spread is not due to erroneous Nblank corrections or insufficient removal of looselybound atmospheric nitrogen. N and Ar in Fig. 3are only weakly correlated, which is a remarkablecontrast to lunar bulk soils, for which the 14N=36Arratio is constant to within a factor of about two [6].

4. Nitrogen excess or noble gas depletion in theregolith?

In this section, we discuss possible causes for thehigh 14N=36Ar ratios in the lunar regolith, empha-


Fig. 2. Same as Fig. 1 for ilmenite grains of soil 71501 in the size range 175–250 µm. The weighted mean 14N=36Ar ratio of the singlegrain data agrees well with the mean ratio deduced by Frick et al. [5] in a stepwise heating analysis of a large ilmenite separate from thesame soil. As in Fig. 1, 14N=36Ar ratios scatter widely, the highest ratios being observed among the least Ar-rich grains.

sizing the single grain data presented here. Becker[14] listed several ways to explain the ‘excess’ ni-trogen. Basically, these are: (1) the 14N=36Ar ratio inthe solar wind is ¾10 times higher than in the sun,

Fig. 3. Amounts of 14N vs. 36Artr for the single ilmenite grains. The Ar amounts vary much more than the N amounts. In contrast tolunar bulk soils, N and Ar amounts in single grains hardly correlate with each other.

(2) solar 36Ar becomes depleted in the regolith bya factor of ¾10 relative to solar N, (3) 90% of theN in the regolith represents one or more non-solarcomponents.


In scenarios (2) and (3) the 14N=36Ar ratio insolar wind and sun are assumed to be identical.Combinations are also possible, e.g. 14N=36Ar in thesolar wind might be a few times solar and some 36Ardepletion relative to N might have occurred.

4.1. 14N=36Ar in the solar wind ¾10 times higherthan the bulk solar value?

This possibility was considered by Becker andPepin [12]. However, a ten-fold enhancement of Nover Ar in the solar corpuscular radiation wouldbe most unexpected according to solar physics, al-though elemental fractionations between solar windand sun are well known. Elements with a first ion-ization potential < 10 eV (low-FIP elements) areenriched in the normal, low-speed solar wind byabout a factor of 4 (cf. [29]), but N and Ar are bothhigh-FIP elements and therefore expected to remainunfractionated relative to each other. The 14N=36Arratio of 43:7š 8:0 in solar energetic particles [30] iswithin error limits identical to the solar ratio of 36.7[27]. Since large, gradual solar energetic particleevents sample the low-speed solar wind composition(cf. [30]), this supports the expectation of an unfrac-tionated 14N=36Ar ratio in the solar wind. The highlyvariable 14N=36Ar ratios in individual grains nowalso provide clear evidence against a N enrichmentin the solar wind. We therefore conclude that thehigh and non-uniform 14N=36Ar ratios in individualgrains must be due either to a variable Ar depletionor a variable non-solar nitrogen excess.

4.2. Heavy noble gases in lunar soils stronglydepleted relative to nitrogen?

Trapping probabilities of Ar ions hitting miner-als with solar wind energies are close to unity, ascalculations by the TRIM program [31] indicate.High N=Ar ratios can therefore not be due to in-efficient trapping of solar Ar but would have to becaused by later Ar loss. The ilmenites would havelost between 75% and 99.77% of their solar Ar, ifall trapped nitrogen would be solar and if the solarN would be completely retained. The grains fromthe first series would have lost 60–96% of their Ar,except for the one remarkable pyroxene with a solar14N=36Ar value. Even larger Ar deficits would be

required if not all solar N had been retained. Aresuch large losses compatible with the abundancesof the three heavy noble gases in the grains? Theratios 36Ar=84Kr and 84Kr=132Xe in single ilmenitesfrom soil 71501 are uniform to within š8% in themost precisely measured 30% of the grains [18]. Thegrains from the first series can be assumed to haveequally uniform Ar=Kr and Kr=Xe ratios as the il-menites, since these values are mineral independent[18]. Such constant elemental ratios show that themeasured Ar=Kr and Kr=Xe ratios represent the trueratios in the solar wind [18], and corroborate ear-lier work indicating that the ratios of trapped Ar=Krand Kr=Xe remain constant with depth within lunargrains [19]. This would not be expected if noblegases had been lost in a way that fractionates theelemental abundances, e.g. by diffusion. Therefore,large and highly variable losses of solar Ar wouldhave had to have happened without fractionating theabundances of the heavy solar noble gases amongthemselves. One would thus require a mechanismthat severely and variably fractionates solar noblegases from solar nitrogen but leaves noble gas ratiosunaltered. Note that a similar conclusion had alreadybeen drawn just on the basis of the observation thatmulti-grain samples would have to have lost some90% of their solar Ar [19,18]. The single grain dataprovide us now with the additional evidence that therequired degree of Ar loss would have to vary highlyfrom grain to grain, yet, the noble gases among them-selves would have to remain unfractionated. Can amechanism be conceived to achieve this?

Becker [14] considers fine iron particles that maychemically bind N without retaining noble gases, buthe notes that such particles cannot account for morethan a small fraction of the N in (bulk) lunar samples.For example, Brilliant et al. [32] suggest that 5% ofthe N in a bulk soil may be associated with finelydispersed metallic Fe. Nanometer-sized metallic Fesimilar to that found in composite particles [33] isalso present in rims of mineral grains from the lunarregolith [34], but we agree with Becker [14] that itdoes not appear possible that most of the solar Ncould be bound in such particles.

The relative abundances of solar Ar, Kr, andXe in bulk samples and mineral separates from thesame soils are essentially identical (e.g. [35]) andthe few available data suggest that this holds also


for the abundance ratios of nitrogen to heavy noblegases [5,13,36]. Therefore, any postulated fractiona-tion between N and noble gases would have affectedmineral grains and bulk samples to the same ex-tent, although most of the gases in the latter residein reworked particles which have a more complexhistory than clean mineral grains. It then becomesimportant to ask where the noble gases and possiblythe N lost from the regolith would have gone. Threeconceivable sinks are loss of atoms or ions fromthe lunar atmosphere, loss of SW-bearing dust fromthe lunar gravity field, e.g. by meteorite impact, andgas sequestering in the polar ice deposits. The latterreservoir is grossly inadequate to possibly retain 10times as much solar noble gases as the entire rest ofthe regolith. In the second process N and noble gaseswould not be fractionated by an order of magnitude.

This leaves the possibility that loss from the at-mosphere might have led to a large depletion ofnoble gases relative to N. Most of the solar noblegases now in the regolith would then have to havebeen recycled through the atmosphere, similar to40Ar, which is outgassed from the Moon and sub-sequently reimplanted by the Manka-Michel process[37]. Ar, Kr, and Xe are lost non-thermally from thelunar atmosphere [37], and the loss fraction dependson element specific parameters such as scale height,ionization cross sections, ion-trajectories, and the ef-ficiencies for a gas to be released from grains andto stick again. If such atmospheric loss would haveseverely depleted the solar noble gases, it wouldseem very likely that their relative abundances wouldhave become severely altered also. This has not hap-pened, however, since bulk samples have the sameAr=Kr and Ar=Xe ratios as mineral separates, whichconserve the true ratios in the solar wind, as wediscussed above [18,19]. With respect to this latterargument, it is important to remember that Ar=Krand Ar=Xe ratios in mineral grains remain constantto depths much larger than the implantation range ofreimplanted atmospheric species, in particular 40Ar[19]. This precludes the possibility that solar Ar,Kr, and Xe in mineral grains are mostly reimplantedfrom a possibly fractionated atmospheric reservoir.Consideration of the total amount of solar Xe in thepresent day regolith also argues against the hypothe-sis that the Moon lost most of the heavy noble gasesit ever trapped. Following Geiss [38], Wieler [39]

concluded that Xe amounts in the lunar regolith re-quire an average flux of solar particles in the past ¾4Ga equal to or perhaps a few times higher than today.The data are therefore consistent with the assump-tion that the regolith as a whole has not lost heavysolar noble gases, although most of the gas atomstoday may not reside in their original trapping sites.

In summary, we can conceive neither a convincingmechanism to produce the required large and highlyvariable nitrogen to noble gas fractionation in singlegrains, nor a convincing sink to remove 90% ofthe heavy noble gases from the lunar regolith. Wetherefore strongly suggest that most of the nitrogenin mineral grains as well as in the lunar regolithas a whole has a source other than the solar wind.In Section 4.3, we will discuss a complementaryargument in favor of this conclusion.

4.3. Solar flare tracks, a further argument fornon-solar nitrogen in lunar soil

Solar flare track densities measure the total timespent by a grain in the uppermost few mm of the re-golith. Different grains which spend the same lengthof time in the top few mm may make different num-bers of trips to the regolith surface. The solar windexposure time of grains will therefore vary morethan the exposure time to track-producing more en-ergetic particles. Therefore, comparing track densitydistributions with those of N and Ar amounts, re-spectively, should allow us to judge which of thelatter elements better measures the residence timeof a grain on the immediate regolith surface. Solarflare track densities in plagioclase grains from soil71501 vary by about a factor of 20 [28]. This isin-between the variations of less than a factor of 6for nitrogen and of more than a factor of 100 for Arin the ilmenite grains studied here, which suggeststhat 36Ar amounts are a reasonably good measureof the surface residence time of each grain, but thatnitrogen amounts are not. This is a further hint thatmost of the N is non-solar, and has not been acquiredon the immediate regolith surface. One might arguethat solar nitrogen may be in saturation in manyilmenites, but since saturation of solar wind speciesseems to occur by sputtering or ‘microflaking’ [22],this should affect nitrogen and 36Ar to approximatelythe same extent.


4.4. The siting of lunar nitrogen near grain surfaces

Similar to the solar noble gases, N in the lunarregolith resides near grain surfaces [1,9,10], which issometimes taken as argument in favor of a solar ori-gin of the nitrogen. This argument is not compelling,since it is easy to envisage trapping mechanisms fornon-solar species that also lead to a surface siting. Awell known example is the retrapped radiogenic 40Arin lunar soils, which resides similarly close to grainsurfaces as the solar 36Ar. A more relevant observa-tion may be that the ratio 14N=36Ar stays constantto within less than a factor of about two in stepwisecombustion runs (gas extraction in the presence ofoxygen), except in the first and last few steps [40].This might suggest that both species are implantedto the same depth and hence have the same (so-lar) origin. However, 40Ar again shows that such aconclusion is not warranted, since the only stepwisecombustion run of a lunar sample we are aware ofto have yielded reliable 40Ar data (soil 79035, [5]),yields constant 40Ar=36Ar ratios to within š20%,except for the first three steps. This clearly showsthat a similar gas release pattern of two surface sitedspecies does not imply an identical origin.

Mathew et al. [7] find a negative correlation be-tween 15N=14N and 22Ne=20Ne for part of the gasreleased from aliquots of a 71501 ilmenite separateetched to different degrees. The authors interpret thisas evidence for the existence of an isotopically dis-tinct SEP-N component, analogous to the SEP-Nethought to be from solar energetic particles havingsomewhat higher energies than the solar wind [8],but showing an isotopic fractionation relative to so-lar wind N opposite to that between solar wind Neand SEP-Ne. These data are certainly another impor-tant step towards solving the ‘lunar nitrogen puzzle’.However, in the view preferred here, the correla-tion between N and Ne isotopic compositions mayinstead just imply that solar and non-solar trappedlunar N components have different depth distribu-tions.

5. Origin of non-solar nitrogen

As discussed in the previous Section 4, our sin-gle grain data strongly support earlier evidence that

nitrogen in the lunar regolith is dominated by atleast one non-solar component. Here we discussconstraints that these data, combined with earliermeasurements, impose on the origin of the non-solarnitrogen.

5.1. Constraints from single grain data

Two important observations are the relativelysmall spread of the N amounts and the very weakcorrelation between 36Ar and N in ilmenites (Fig. 3).The most N-rich grain contains only 5.7 times asmuch nitrogen as the least N-rich grain. The correla-tion coefficient in Fig. 3 is only 0.44. N and Ar thuscorrelate much less well in single ilmenite grainsfrom one soil than in bulk samples from differentsoils, which yield a correlation coefficient of r D 0:9[6]. In bulk samples, N also correlates remarkablywell with Is=FeO [6], a measure for the ‘maturity’of a soil, i.e. for the time it spent near the surfaceof the regolith. Single grain and bulk sample datataken together therefore indicate that the non-solarN is presumably trapped close to the lunar surfacebut not only while a grain is at the immediate top ofthe regolith. This conclusion has already been drawnfrom the track data in Section 4.3. and seems toimply that N does not get trapped by ion implanta-tion. An interesting grain to illustrate this is ilm. #13,which has a very low 36Ar amount but an average Namount. The 36Ar concentration per surface area inthis grain is similar to that of many grains of noblegas-poor soil 60051 [28], which may have experi-enced just a single exposure at the regolith top [41].Ilm. #13 may thus also have been just once on thelunar surface, but despite this acquired an averageamount of nitrogen.

Fig. 4 shows the single ilmenite data in a mix-ing diagram. Data points will plot on a straight linefor all grains that contain variable amounts of e.g.a solar component plus constant amounts of a sec-ond, N-only, component. Such a straight line wouldintercept the ordinate at the 14N=36Ar ratio of thesolar component. Fig. 4 does not define a singlestraight line, therefore not all ilmenites contain thesame amount of non-solar N. On the other hand,the 7 data points defining the lower bound of thetriangle encompassing all points (open symbols) fallon a straight line which intersects the ordinate within


Fig. 4. 14N=36Artr vs. 1=36Artr for the ilmenite grains. The solid straight line is the best fit through the data points shown by opensymbols. This line intersects the ordinate at the solar 14N=36Ar ratio. The data are therefore compatible with the hypothesis that all grainscontain a mixture of solar Ar and N plus a N-only component present in the same minimum amount in a sizeable fraction of the grains.In this interpretation, the dotted straight line indicates that the maximum non-solar N amount per grain is about 4.5 times higher than theminimum.

error limits at the solar 14N=36Ar ratio. We cannotexclude that this is fortuitous, i.e. we do not regardthe figure as providing proof that a component witha solar 14N=36Ar ratio is indeed present in the grains.We think, however, that this is a reasonable inter-pretation and we believe that the figure shows thata sizeable fraction of the grains contains a commonminimum amount of non-solar N. The 7 grains definethis minimum to ¾2:8ð 10�12 mole for ilmenites ofsoil 71501 of 175–250 µm diameter. The dotted linein Fig. 4 defines the maximum amount of non-solarN in these grains to be ¾12:5 ð 10�12 mole. Notethat not much of this N should be indigenous to thegrains in the sense of having been present alreadywhen they were still part of a rock, since most, if notall, of the non-cosmogenic nitrogen in lunar mineralgrains must reside near grain surfaces.

5.2. Constraints from bulk sample measurements

As pointed out in the previous section, amountsof 14N and 36Ar in bulk samples correlate quite well.For example, the 14N=36Ar ratio is constant to withina factor of 2.5 in Apollo 16 samples of variable

antiquity of up to 2.5 Ga [6]. This means that thesupply of the non-solar component has remainedapproximately constant over the last ¾2.5 Ga. Thisconstraint is not new for scenarios invoking twoN components, but it becomes more stringent herebecause we postulate that 90% of the nitrogen in theregolith is non-solar.

Probably the least understood fact about lunarnitrogen is the large variability of its isotopic compo-sition. δ15N values vary by up to ¾30% in stepwiseextractions of bulk samples and mineral separates(e.g. [4,5,9,11,40]). Similarly, δ15N values in differ-ent bulk samples vary also by some 30%, and thesevariations correlate with the antiquity of the samples[6]. The classical interpretation that this testifies toa secular change of the N isotopic composition inthe solar wind of some 30% cannot be true if only¾10% of the lunar N is solar. On the other hand,even if lunar N consists of two components, at leastone of them appears to require a temporally variablecomposition. This is because δ15N values do not cor-relate with the 14N=36Ar ratio, which implies that theδ15N variability can only in special circumstances beexplained by variable proportions of two different N


components of fixed isotopic composition [6,14]. Toexplain the observed ¾30% spread of δ15N valuesby a variable solar N component that represents only10% of the total lunar N would require the δ15Nin the solar wind to have varied by no less than¾2000‰. In contrast, the δ15N of the 90% non-solarN would ‘only’ need to have varied by some 330‰.The second alternative seems preferable, although itremains a clear challenge to identify such a compo-nent.

Sugiura et al. [42] recently proposed to explainour single grain data by an unspecified carrier re-sponsible for the trapped N in certain meteorites(‘ALHA77214 type’) added to the lunar regolith.This input could, however, not have been achievedsimply by infalling ordinary meteoritic material, asthe authors also point out, since bulk meteorites ofthe ALHA77214 type contain roughly an order ofmagnitude less N than lunar bulk soils. Only about2% of the regolith consists of non-lunar matter [43].For the sake of the argument, we may assume thatall this material is heavily enriched in the N carrierof ALHA77214. To contribute the 90% non-solar Nin the regolith, this carrier would then need to beenriched in the foreign material about 500 times rel-ative to its concentration in bulk ALHA77214, andcorrespondingly larger enrichment factors would berequired if only part of the meteoritic material inthe regolith is enriched in this carrier. We believethat this requirement makes meteoritic N an unlikelysource for the sought non-solar component.

As mentioned in Section 3, it cannot be rigor-ously excluded that lunar regolith samples containtightly bound terrestrial nitrogen acquired after sam-ple return. However, the ¾30% variation of δ15Nin samples of different antiquity [6] is a clear in-dication that N contamination on Earth cannot bethe main explanation of the nitrogen puzzle, since itseems inconceivable that such contamination coulddistinguish between low-and high-antiquity samples.Another, so far mysterious, non-solar N componentmust exist.

6. Summary and conclusions

The N blank in our extraction line is low and sta-ble enough to allow us to determine with reasonable

precision the 14N amounts, and hence the 14N=36Arratio, in single grains from the lunar regolith. Notyet possible has been a precise determination of theδ15N values for these individual grains. The data herecan therefore be used to provide evidence on the firstof the two most crucial unexplained observations re-lated to lunar N, which is the fact that the abundanceratio of N to heavy noble gases in bulk samples andmineral separates is some ten times higher than thebulk solar value. The second of these crucial obser-vations, the highly variable N isotopic compositionin the samples, will be addressed by future stepwiseheating N and Ar analyses on single grains [44].

The most important observation here is that14N=36Ar ratios in single grains of lunar soil71501 vary by more than two orders of magni-tude, while it has been shown earlier that 36Ar=84Krand 36Ar=132Xe ratios in grains of the same soil arevery uniform and identical to the true ratios in thesolar wind. If the scatter of 14N=36Ar ratios werecaused by highly variable fractional losses of solar36Ar (between 0% and 99.77%), this would havehappened without an expected concomitant highlyvariable fractionation of the heavy noble gas abun-dances. We cannot conceive how this could havehappened, and therefore conclude that the highlyvariable N=Ar ratios in single grains are very prob-ably due to the presence of a major non-solar Ncomponent. In bulk samples, this component ac-counts for about 90% of the total trapped N. Insingle grains, the non-solar N accounts for between0% (in one grain only) and 99.77% of the total N.The absolute amounts of this component in ilmenitegrains in a narrow size range vary by less than afactor of 6, considerably less than the amounts ofsolar 36Ar. This seems to suggest that the non-solarN has not been trapped by ion implantation. Theorigin of the non-solar component remains a puzzle,but it presumably must have changed its isotopiccomposition over the past several billion years.

Acknowledgements

We thank J.-P. Benkert for sample preparationand L. Zimmermann for technical assistance. Wevery much appreciate constructive reviews and com-ments by R.H. Becker, I.A. Franchi, J.F. Kerridge,


K. Kehm, and K. Marti. This work has been sup-ported by CNRS and the Swiss National ScienceFoundation. [FA]

References

[1] O. Muller, Solar wind nitrogen and indigenous nitrogen inApollo 17 lunar samples, Proc. 5th Lunar Sci. Conf., 1974,pp. 1907–1918.

[2] J.F. Kerridge, Solar nitrogen: evidence for a secular increasein the ratio of nitrogen-15 to nitrogen-14, Science 18 (1975)162–164.

[3] R.H. Becker, R.N. Clayton, T.K. Mayeda, Characterisationof lunar nitrogen components, Proc. 7th Lunar Sci. Conf.,1976, pp. 441–458.

[4] R.N. Clayton, M.H. Thiemens, Lunar nitrogen: Evidencefor secular change in the solar wind, in: R.O. Pepin, J.A.Eddy, R.B. Merrill (Eds.), Proc. Conf. Ancient Sun, 1980,pp. 463–473.

[5] U. Frick, R.H. Becker, R.O. Pepin, Solar wind record in thelunar regolith: Nitrogen and noble gases, Proc. 18th Lun.Planet. Sci. Conf., 1988, pp. 87–120.

[6] J.F. Kerridge, Long-term compositional variation in solarcorpuscular radiation — evidence from nitrogen isotopes inthe lunar regolith, Rev. Geophys. 31 (4) (1993) 423–437.

[7] K.J. Mathew, J.F. Kerridge, K. Marti, Nitrogen in solarenergetic particles: isotopically distinct from solar wind,Geophys. Res. Lett 25 (1998) 4293–4296.

[8] J.-P. Benkert, H. Baur, P. Signer, R. Wieler, He, Ne, and Arfrom the solar wind and solar energetic particles in lunarilmenites and pyroxenes, J. Geophys. Res. (Planets) 98 (E7)(1993) 13147–13162.

[9] R.H. Becker, R.N. Clayton, Nitrogen abundances and iso-topic compositions in lunar samples, Proc. 6th Lunar Sci.Conf., 1975, pp. 2131–2149.

[10] P.S. Goel, B.K. Kothari, Total nitrogen contents of someApollo 14 lunar samples by neutron activation analysis,Proc. 3rd Lunar Sci. Conf., 1972, pp. 2041–2050.

[11] D.R. Brilliant, I.A. Franchi, C.T. Pillinger, Nitrogen com-ponents in lunar soil 12023 — complex grains are notthe carrier of isotopically light nitrogen, Meteoritics 29 (5)(1994) 718–723.

[12] R.H. Becker, R.O. Pepin, Solar wind noble gases and nitro-gen in metal from lunar soil 68501, Meteoritics 29 (1994)724–738.

[13] R.H. Becker, R.O. Pepin, Long-term changes in solar windelemental and isotopic ratios: a comparison of two lunarilmenites of different antiquities, Geochim. Cosmochim.Acta 53 (1989) 1135–1146.

[14] R.H. Becker, An evaluation of possible explanations for‘excess’ nitrogen in the lunar regolith, Lunar Planet. Sci.XXVI, Lunar Planet. Institute, Houston, 1995, pp. 89–90.

[15] J. Geiss, P. Bochsler, Nitrogen isotopes in the solar system,Geochim. Cosmochim. Acta 46 (1982) 529–548.

[16] P. Signer, H. Baur, P. Etique, R. Wieler, Nitrogen and noble

gases in the 71501 bulk soil and ilmenite as records of thesolar wind exposure: Which is correct?, in: Workshop ofPast and Present Solar Radiation, Mainz, LPI Tech. Rep.86-02, Lunar Planet. Institute, Houston, 1986, pp. 34–35.

[17] J. Geiss, P. Bochsler, Long time variations in solar windproperties: possible causes versus observations, in: C.P.Sonett, M.S. Giampapa, M.S. Matthews (Eds.), The Sun inTime, Univ. Arizona Press, 1991, pp. 98–117.

[18] R. Wieler, K. Kehm, A.P. Meshik, C.M. Hohenberg, Secularchanges in the xenon and krypton abundances in the solarwind recorded in single lunar grains, Nature 384 (1996)46–49.

[19] R. Wieler, H. Baur, Fractionation of Xe, Kr, and Ar inthe solar corpuscular radiation deduced by closed systemetching of lunar soils, Astrophys. J. 453 (2 Part 1) (1995)987–997.

[20] C. Petrowski, J.F. Kerridge, I.R. Kaplan, Light elementgeochemistry of the Apollo 17 site, Proc. 5th Lunar Planet.Sci. Conf., 1974, pp. 1939–1948.

[21] R. Wieler, H. Baur, Krypton and xenon from the solarwind and solar energetic particles in two lunar ilmenites ofdifferent antiquity, Meteoritics 29 (1994) 570–580.

[22] R.H. Nichols, C.M. Hohenberg, C.T. Olinger, Implantedsolar He, Ne and Ar in individual lunar ilmenite grains:Surface effects and a temporal variation in the solar windcomposition, Geochim. Cosmochim. Acta 58 (1994) 1031–1042.

[23] A.O. Nier, D.J. Schlutter, Helium and neon in lunar il-menites of different antiquities, Meteoritics 29 (1994) 662–673.

[24] F. Humbert, Solubilite de l’azote dans les silicates liquidesen fonction de la fugacite d’oxygene, PhD thesis, UniversiteHenri Poincare, Nancy, 1998.

[25] B. Marty, F. Humbert, Nitrogen and argon isotopes inoceanic basalts, Earth Planet. Sci. Lett. 152 (1234) (1997)101–112.

[26] S.D. Samson, E.C. Alexander, Calibration of the interlab-oratory 40Ar–39Ar dating standard, MMhb-1, Chem. Geol.66 (1987) 27–34.

[27] E. Anders, N. Grevesse, Abundances of the elements: me-teoritic and solar, Geochim. Cosmochim. Acta 53 (1989)197–214.

[28] R. Wieler, P. Etique, P. Signer, G. Poupeau, Record of thesolar corpuscular radiation in minerals from lunar soils: Acomparative study of noble gases and tracks, Proc. 11thLun. Planet. Sci. Conf., 1980, pp. 1369–1393.

[29] J.P. Meyer, Element fractionation at work in the solar at-mosphere, in: N. Prantzos, E. Vangioni-Flam, M. Casse,(Eds.), Origin and Evolution of the Elements, CambridgeUniv. Press, Cambridge, 1993, pp. 26–62.

[30] H.H. Brenemann, E.C. Stone, Solar coronal and photo-spheric abundances from solar energetic particle measure-ments, Astrophys. J. 299 (1985) L57–L61.

[31] J.F. Ziegler, J.P. Biersack, U. Littmark, The Stopping andRange of Ions in Solids, vol. 1, Pergamon, New York, 1985.

[32] D.R. Brilliant, I.A. Franchi, C.T. Pillinger, Release of lightnitrogen from Apollo 12023 < 10 µm fraction by a combi-


nation of pyrolysis — combustion, Lunar Planet. Sci. XXV,Lunar Planet. Institute, Houston, 1994, pp. 171–172.

[33] R.M. Housley, R.W. Grant, N.E. Paton, Origin and char-acteristics of excess Fe metal in lunar glass welded ag-gregates, Proc. 4th Lunar Science Conf., 1993, pp. 2737–2749.

[34] L.P. Keller, D.S. McKay, The nature and origin of rimson lunar soil grains, Geochim. Cosmochim. Acta 61 (11)(1997) 2331–2341.

[35] P. Eberhardt, J. Geiss, H. Graf, N. Grogler, M.D. Mendia,M. Morgeli, H. Schwaller, A. Stettler, U. Krahenbuhl, H.R.von Gunten, Trapped solar wind noble gases in Apollo 12lunar fines 12001 and Apollo 11 breccia 10046, Proc. 3rdLun. Sci. Conf., 1972, pp. 1821–1856.

[36] J.F. Kerridge, J.S. Kim, Y. Kim, K. Marti, Evolution ofisotopic signatures in lunar-regolith nitrogen: noble gasesand nitrogen in grain-size fractions from regolith breccia79035, Lunar Planet. Sci. Conf. 22 (1992) 215–224.

[37] R.H. Manka, F.C. Michel, Lunar atmosphere as a source oflunar surface elements, Proc. 2nd Lunar Sci. Conf., 1971,pp. 1717–1728.

[38] J. Geiss, Solar wind composition and implications about

the history of the solar system, Conf. Pap. Int. 13th CosmicRay Conf., 1973, pp. 3375–3398.

[39] R. Wieler, Why are SEP noble gases so abundant in ex-traterrestrial samples?, Lunar Planet. Sci. XXVIII, LunarPlanet. Institute, Houston, TX, 1997, pp. 1551–1552.

[40] D.R. Brilliant, PhD thesis, Open University, Milton Keynes,UK, 1998.

[41] P.E. Rider, R.O. Pepin, R.H. Becker, Noble gases and ni-trogen released from lunar soils by acid etching, Geochim.Cosmochim. Acta 59 (1995) 4983–4996.

[42] N. Sugiura, K. Kiyota, K. Hashizume, Nitrogen componentsin primitive ordinary chondrites, Meteor. Planet. Sci. 33(1998) 463–482.

[43] J.W. Morgan, U. Krahenbuhl, R. Ganapathy, E. Anders,Trace elements in Apollo 15 samples: Implications formeteorite influx and volatile depletion on the moon, Proc.Lunar Sci. Conf. 3rd (1972) 1361–1376.

[44] K. Hashizume, B. Marty, R. Wieler, Single grain analysesof the nitrogen isotopic composition in lunar regolith, LunarPlanet. Sci. XXX, Lunar Planet. Institute, Houston, TX, inpress.

Documents

Evidence for a predominantly non-solar origin of nitrogen in the lunar regolith revealed by single grain analyses