Technical Comments

Isotopic Variability of Nitrogen in Lunar Regolith

See allHide authors and affiliations

Science  14 Sep 2001:
Vol. 293, Issue 5537, pp. 1947
DOI: 10.1126/science.293.5537.1947a

Hashizume et al. (1) claim that the distribution and isotopic composition of nitrogen in the lunar regolith can be explained by a two-component mixing model. Their report, however, fails to discuss the bulk of existing relevant data—data that show unequivocally that the conclusions in (1) cannot be correct. Although the data, derived by several different groups on many different samples using a variety of analytical approaches, are quite complex, rather simple arguments serve to illustrate the failure of a two-component mixing model such as that presented in (1).

As pointed out in (1), N in the lunar regolith is overabundant relative to the heavy noble gases, which are of dominantly solar origin, by about an order of magnitude compared with their solar proportions. Regolith N is isotopically inhomogeneous; variations both within individual samples and between different samples can exceed 30%. Intersample variations are apparently a function of the epoch during which each sample was exposed on the lunar surface (2), though that dependence is more complex than is suggested by Hashizume et al. (1). Intrasample isotopic variations are at least partly a function of implantation depth within individual mineral grains, as predicted by Kerridge (2), demonstrated by Matthew et al. (3), and confirmed by Hashizume et al. (1). Finally, the abundance of N correlates strongly, on a sample-to-sample basis, with that of all measurable elements implanted in the regolith by the solar corpuscular radiation (2).

Two-component models that seek to explain isotopic variability fall into two groups: (i) category I models, in which the isotopic composition of each end-member is constant and in which intersample variability arises solely from changes in mixing ratio of the components; and (ii) category II models, in which the isotopic composition of one end-member or both end-members varies, with or without a change in mixing ratio. As described, the model of Hashizumeet al. (1) falls into category I; their figure 2 shows the restricted compositions of the two end-members.

Such a model may be subject to a straightforward test. By definition, the “solar” end-member hypothesized by Hashizume et al. (1) is characterized by proportions of N and the heavy noble gases that are solar or close to it. To achieve the observed overabundance of N relative to those noble gases in regolith samples, their “planetary” end-member must be substantially enriched in N relative to the noble gases. It follows that, in order for the model to explain a large variation in 15N/14N ratio, the proportion of N to noble gases in individual samples must covary with the 15N/14N ratio. Given the end-member compositions proposed in (1), the N/noble-gas ratio must increase as the 15N/14N ratio increases, the actual quantitative relationship depending upon the ratio of N to noble gases in the “planetary” component, not stated in (1). In practice, however, the N isotopic composition of regolith samples shows no dependence upon the measured proportion of N to any of the heavy noble gases (2). For example, ilmenite separates from 71501 and 79035 (the two samples from which Hashizumeet al. took their mineral grains) have very similar N/36Ar ratios despite markedly different15N/14N ratios (4). Observations like this effectively rule out any category I model, including that of (1).

Of course, this conclusion might not apply if heavy noble gases, such as Ar, are substantially lost from the lunar regolith. Indeed, Hashizume et al. (1) suggest that N experiences “preferential retention . . . relative to [noble] gases . . . under the lunar reducing conditions.” Isotopic and abundance data for the heavy noble gases, however, argue against any significant loss of those gases from the lunar regolith (5, 6). Furthermore, to bring the model of (1) into consistency with the observational data, the magnitude of such noble-gas loss must be tied in some way to the N isotopic composition—yet the model is incapable of predicting such a connection. Variable loss of heavy noble gases therefore becomes a free parameter in the model.

It follows that any mixing model capable of explaining the lunar regolith N data must be of category II. Indeed, if the N inventory in the regolith is dominated (∼90%) by “planetary” N, as argued in (1) and, previously, in (5–7), large bulk isotopic variations, as observed, obviously must be due to variability within that “planetary” component, absent a much more extreme isotopic composition for the “solar” component than any yet observed. Thus, drawing upon arguments presented above and elsewhere (2,3), a hypothetical “planetary” component capable of explaining the N isotopic systematics observed for the lunar regolith must be derived from a source with the following properties: (i) It must deliver ∼30 μg N per cm2 per million years to the lunar surface; (ii) it must enable that N to be trapped in a way that closely mimics implantation by solar corpuscular radiation; and (iii) its 15N/14N ratio must vary by at least 30% on a billion-year time scale. The possible existence of such a source is problematical, and Hashizume et al. (1) do not identify one.

These arguments have focused only on isotopic data for N, whereas the model of Hashizume et al. (1) addresses both N and H isotopic data in regolith samples. However, although a rival model to that of (1) must explain both N and H data at least as well as (1), a failure to explain the N data alone constitutes grounds for rejecting the model of (1). Indeed, lunar regolith D/H data in the literature fail to show any systematic dependence upon 15N/14N (8).

Beyond these arguments, four additional points should be noted:

1) Hashizume et al. (1) do not mention a recent study by Mathew et al. (3) that used stepwise etching to reveal isotopically distinct N components within a sample essentially identical to one of theirs (i.e., ilmenite from 71501). The two studies yielded similar results: an isotopically heavy, near-surface component [δ 15N > +50‰ in (1); δ 15N > +68‰ in (3)] and an isotopically light, more deeply implanted component [δ15N < –100‰ in (1); δ 15N < –90‰ in (3)]. However, Mathew et al. (3) also analyzed the isotopic composition of neon associated with each N component and found that relatively heavy N correlated with so-called solar wind (SW) Ne (20Ne/22Ne = 13.8), and relatively light N correlated with so-called solar energetic particle (SEP) Ne (20Ne/22Ne = 11.2). This is the opposite relationship to that inferred, without any direct observational basis, by Hashizume et al. (1). Although the nature and origin of SEP Ne are still controversial, Ne with an isotopic ratio of 13.7 has been directly measured in the SW (9).

2) The conclusion of (1) that SW N has δ 15N below –240‰ is inconsistent with the measurement by Kim et al. (10) of N with a δ 15N value of +40‰ in the surface of a recently exposed lunar rock. The stepwise thermal release of N from that sample showed no significant evidence for a –240‰ component, and the ratio of N to noble gases was considerably lower than that in soils, which indicate much less of any putative nonsolar N component.

3) The suggestion (1) that the N isotopic composition of solar system objects reflects a mixing of solar N with N fractionated by interstellar chemical processes has been made previously (11). None of the data presented in (1), however, bear usefully upon this issue, which remains speculative (though not unreasonable).

4) Although interpreting the isotopic systematics of regolith N in terms of isotopic variability within the solar corpuscular radiation suffers from a number of difficulties (1, 2, 5, 6, 11, 12), those difficulties are arguably no more severe than those mentioned above in connection with two-component mixing models.

In summary, the model advanced by Hashizume et al. (1) fails to account for the isotopic variability of lunar regolith N, which, 25 years after its discovery (13), still lacks a viable explanation.


Response: As Kerridge notes in his comment and noted in his 1993 review (1), the interpretation of the 30% isotopic variation observed among the lunar regolith samples has been a matter of debate for three decades. In 1975, Kerridge (2) proposed that the N isotopic variation of lunar regolith was due to a secular change of the nitrogen isotopic composition of the solar corpuscular irradiation, a possibility discarded by Geiss and Bochsler (3) because no known spallation or thermonuclear reaction could produce enough 15N. Instead, Geiss and Bochsler concluded that the N isotope ratio at the solar surface has been constant during the last 4 billion years, and proposed that variation of the 15N/14N ratio is due to mixing between solar and nonsolar N components. More recently, Wieler et al. (4) demonstrated that lunar regoliths do indeed contain a nonsolar N component, based on single-grain analysis of N and Ar, a technique recently developed at the Centre de Recherches Pétrographiques et Géochimiques in Nancy, France. Independently, it has recently been shown (5, 6) that the largest flux of extraterrestrial matter on the Earth is due to the fall of microscopic objects, micrometeorites, and interplanetary dust particles (IDPs), which, as we show below, matches well the required characteristics of the nonsolar component present in the lunar regolith.

We have recently developed a new working hypothesis based on modern results of microscopic lunar regolith study (7,8), quantitative studies on planetary materials accreting to the contemporary Earth and Moon (5, 6,9), and characterization of these materials (10,11). In this new model, our recent study (8) plays a key role, because it has allowed us to identify a15N-depleted component that we argue to be of solar origin and a 15N-rich planetary component. These new observations were obtained by ion probe depth profiling of single grains from lunar regoliths, a technique that has offered the first clear view of the distribution versus depth of N and H isotopes at nanometer-scale resolutions. Three major lines of evidence emerge from these depth profiles: (i) The 15N-depleted N is present in the grains at a depth of ∼50 nm, this depth being characteristic of the implantation depth of the SW having energies of ∼1 KeV/nucleon. (ii) The 15N-depleted N is associated in the grains with D-free hydrogen, which implies a solar origin for H. (iii) Some lunar regolith grains also present a 15N-rich nitrogen component at their surface; this N is associated with D-rich H (akin to meteoritic H) and is present in Si-rich surface deposits akin to those described generally in lunar grains as a result of meteoritic bombardment of the Moon (7).

A brief description of this working hypothesis was recently published as a symposium abstract (12, 13), and a paper discussing this issue in full detail is currently in preparation. In brief, we propose that the isotopic variation of lunar regolith N is the result of contributions in variable proportion of the solar corpuscular flux and of the micrometeoritic flux (12). We argue that nitrogen in micrometeorites was released to lunar atmosphere (14) by impact-degassing, and then reimplanted (15) or chemisorbed to the surface of lunar regolith minerals, where SW components also reside. This model can be classified under Kerridge's category I, notwithstanding his claim that such a model cannot explain the observed lunar N systematics.

Our model also is consistent with the three criteria that, according to Kerridge, must be met by the source of “a hypothetical ‘planetary’ component capable of explaining the N isotopic systematics observed for the lunar regolith.” (i) If the accretion rate of the micrometeorites to Moon is similar to that recorded on Earth (5, 6), and assuming that its N concentration is typical of primitive carbonaceous chondrites, as suggested by Keller et al. (10), the supply rate of micrometeoritic N is about 10 μg N per cm2 per million years, which is of the same order of magnitude as the flux requested by Kerridge. (ii) The micrometeoritic accretion is a continuous event, which parallels that of SW implantation. (iii) The observed δ15N variation of the lunar N can be explained by variation in the flux ratio between the SW and micrometeorites. The difference in the δ15N values between these end-members is enormous—much larger than the range observed among lunar samples. Owen et al. (16), from the isotopic measurement of the Jovian atmosphere, have recently proposed a solar δ15N value of −370 ± 80‰, which is compatible with our upper limit estimate (δ15N <−240‰) for lunar samples (8). Carbonaceous chondrites, which share similarities with micrometeorites [e.g., (7)], show positive δ15N values (around +40‰ for CMs and up to several hundreds of ‰ for CRs and related objects); also, IDPs are enriched in 15N [−90 <δ15N <+800‰ (11, 17)]. To estimate the relative proportions of solar and planetary N, one has to assume δ15N values for the solar and nonsolar end-members that require further analysis and will be developed elsewhere. For the sake of illustration, however, we estimate that a δ15N value increase of 100‰ in a bulk lunar sample can be explained by an increase of only 1.3 to 2 times in the micrometeoritic N fraction in bulk N.

Although Kerridge claims that the N to noble gas ratio is constant among bulk lunar samples, N/36Ar ratios are actually variable by factor of ≤3 among different regolith samples—even in the case of the Apollo 16 samples taken as an example by Kerridge (1), where bulk δ15N variation is less than a half of the full range (300‰). We do not follow the simplistic argument of Kerridge that, in case of binary mixing between solar and planetary N, there should exist a single mixing trend between the δ15N value and the N/36Ar ratio, given that miscellaneous analytical and natural factors might exist that would obscure the relationship, suggested from the observed diversity in the N/36Ar ratio (2). The flux of impacting bodies on the Moon might have varied with time (9), possibly showing a significant increase during the last 400 Ma. The “recent” lunar samples on which Kerridge bases his model for N isotope secular variation of the solar corpuscular irradiation might actually be heavily “contaminated” by planetary N.

Kerridge refers to the study of Mathew et al. (18), who argued that there exists an excellent correlation between the N and the Ne isotopic compositions for three samples of the “recent” regolith 71501, which were acid-leached to different degrees. However, the Ne and N isotopic ratios among the “bulk” data of the surface-correlated components (700 + 1040° C temperature step data) for respective samples define a very poor correlation among them, as Mathew et al. (18) note. The correlation claimed in (18) is created by adding data for different temperature steps obtained during stepwise heating experiments. However, N and Ne released at a given temperature do not necessarily originate from the same site or depth of lunar grains, because N, unlike Ne, is chemically reactive in the reducing conditions characterizing the lunar surface. Finally, the conclusion in (18) that SEP N is depleted in 15N relative to the SW N is not consistent with noble gas systematics. Indeed, the relationship between the isotopic composition of SW and SEP among He, Ne, and Ar [e.g., (19)] suggests that SEP always exhibit enrichment in the heavier isotopes. For a SW N component depleted in15N (δ15N <–240‰), however, the problem vanishes, because the proposed value of SEP N, on which (18) and (8) agree, is around –100‰.

Based on our ion probe and single-grain analyses (8,12), we reinforce our previous conclusion that SW N is depleted in 15N by at least –240‰ and that N isotopic variations observed in the lunar regolith result from variable contributions of a planetary-type N component or components.

*Also Ecole Nationale Supérieure de Géologie–INPL, Rue de Doyen Marcel Roubault, BP 40, 54501 Vandoeuvre-lès-Nancy Cedex, France.


Stay Connected to Science


Navigate This Article