LUNAR ATMOSPHERE

How surface composition andmeteoroid impacts mediate sodiumand potassium in the lunar exosphereA. Colaprete,1* M. Sarantos,2,3 D. H. Wooden,1 T. J. Stubbs,4 A. M. Cook,1,5 M. Shirley1

Despite being trace constituents of the lunar exosphere, sodium and potassium are themost readily observed species due to their bright line emission. Measurements ofthese species by the Ultraviolet and Visible Spectrometer (UVS) on the LunarAtmosphere and Dust Environment Explorer (LADEE) have revealed unambiguoustemporal and spatial variations indicative of a strong role for meteoroid bombardmentand surface composition in determining the composition and local time dependenceof the Moon’s exosphere. Observations show distinct lunar day (monthly) cycles for bothspecies as well as an annual cycle for sodium. The first continuous measurements forpotassium show a more repeatable variation across lunations and an enhancement overKREEP (Potassium Rare Earth Elements and Phosphorus) surface regions, revealing astrong dependence on surface composition.

The Lunar Atmosphere and Dust Environ-ment Explorer (LADEE) mission, which op-erated in lunar orbit between 6 October 2013and 18 April 2014, had the goal to deter-mine the composition of the lunar atmo-

sphere and investigate the processes that controlits distribution and variability, including sources,sinks, and surface interactions (1). The Ultravioletand Visible Spectrometer (UVS) (2) was designedto make observations of the lunar exosphere andsearch for dust. The LADEE orbit was retrogradeand equatorial; thus, observations were restrictedto between about ±20° latitude. However, this orbitdid provide synoptic observations of lunar exo-spheric species with temporal cadence of typicallyless than 12 hours. These observations providenew constraints on the processes, sources, andsinks that govern concentrations of species in thelunar exosphere. UVS observed resonant scatteringfrom sodium (Na) and potassium (K). AlthoughNa and K are minor constituents in the lunar exo-sphere, the brightness of their emission lines makesthem readily observable, and thus they serve asexcellent proxies to understanding the processesthat govern the composition of the lunar exo-sphere (3–6).Potential sources of Na and K in the exo-

sphere include photon-stimulated desorption (PSD),sputtering, and impact vaporization (Fig. 1). Therole of PSD and sputtering have been estimatedby evaluating the change in Na abundances asthe surface is shielded from solar wind sputteringas the Moon passed through Earth’s magneto-

tail (7). From these previous Earth-based observa-tions and modeling, PSD was generally consideredto be the dominant source, by a factor of ~10 to100, over sputtering or impact vaporization (6, 7).However, the question of whether photons simplyre-excite material that has been previously vapor-ized by other processes or whether they act as aprimary source process of atoms from glasses andminerals has not been previously answered. The

only other space-based observations of Na weremade by the Telescope for Visible Light (TVIS) in-strument on board the Kaguya spacecraft (8).Kaguya TVIS observations were limited to thelunar nightside, looking in the antisolar directiontoward the extended Na tail; thus, they prefer-entially observed the “hottest” portion of the Napopulation and were not continuous across theentire lunation (9). These observations showed acontinuous decrease of the inferred Na surfacedensity from first to third quarter, including phaseswhen the Moon passes through Earth's magneto-sphere, with a variation of about 50% across alunation.Observations of K from Earth are more difficult

than for Na because a strong telluric O2 band over-whelms the stronger of the two K lines, whereasthe smaller scale height for K results in relativelysmall concentrations at altitudes where Earth-based observations can be made. Besides its initialdiscovery (3), only two other K observations havebeen published (10, 11), and these were limitedto observations on single days. There are thereforeno observations of K over the course of a lunation.Earth-based observations have detected increases

in the overall concentration of Na in the lunar exo-sphere associated with some meteoroid streams(12–16). However, these observations were inter-mittent and limited in time, often focusing ononly a few measurements about an individualstream; thus, Na variability associated with mete-oroid streams has been inconclusive (17). Annualvariability of lunar-sodium tail brightness has beenfound, with a peak in the Na tail brightness in

SCIENCE sciencemag.org 15 JANUARY 2016 • VOL 351 ISSUE 6270 249

1Space Science Division, NASA Ames Research Center,Moffett Field, Mountain View, CA, USA. 2Heliophysics ScienceDivision, NASA Goddard Space Flight Center, Greenbelt, MD,USA. 3University of Maryland, Baltimore County, Baltimore,MD, USA. 4Solar System Exploration Division, NASA GoddardSpace Flight Center, Greenbelt, MD, USA. 5MillenniumEngineering and Integration Services, Moffett Field, CA94035, USA.*Corresponding author. E-mail: anthony.colaprete-1@nasa.gov

Fig. 1. The primary sources and sinks of the lunar Na and K exosphere.The yellow region representsthe sodium exosphere having a greater extent on the dayside due to higher sodium temperatures. Thesources and sinks include photon-stimulated desorption (from solar ultraviolet), sputtering (from solarwind protons) and meteoroid impact vaporization (e.g., the Geminid meteoroid stream).These processesare affected as the Moon passes through Earth’s magnetosphere (green region) and sheath (red region).

RESEARCH | REPORTSon June 5, 2019

http://science.sciencemag.org/

Dow

nloaded from

February, but no explanation for the annual varia-tions has been provided (9, 17). The UVS observa-tions allow for continuous monitoring of theexosphere, both during and outside intervals ofincreased stream activity, through the course ofmore than five lunations, which allows for a muchless ambiguous measure of immediate Na activityassociated with meteoroid streams. In additionto these Na observations, these observations arethe first to assess the response of K to meteoroidstream activity.The emission line strengths for both Na and

K were derived from UVS limb observations ataround local noon. These noon limb-viewing ac-tivities typically covered spacecraft solar longitudesbetween –25° and +55° with a telescope viewingaltitude of around 40 km (2, 18).The derived tangent column densities of Na

and K show variations by a factor of 2 to 3 overthe course of a lunation (Fig. 2). The minimum-to-maximum ratio of Na column density changes con-tinuously during the five lunations measured byLADEE, but the factor of 2 for Na seen on thelast month is significantly different from the ~50%seen in Kaguya observations (9). A factor of twochange in the lunar Na content was inferred frommodeling ground-based observations taken beforeUVS observations (7) and attributed to the solarwind increasing the PSD rate via ion-enhanceddiffusion inside grains (18). However, the modelin (18) predicted that the exosphere rose contin-uously between exit and subsequent reentry tothe magnetosphere. The Kaguya TVIS data showeda continuous decline in Na through the magneto-tail, whereas LADEE UVS data show Na peaksnear full Moon. These differences are perhaps adifference in vantage points as suggested to re-concile the Kaguya results with the analysis ofground-based observations (9).As the Moon passes into Earth’s magneto-

tail, the total Na is seen by UVS first to decreaseand then increase through full Moon to a max-imum about 30° of lunar phase, after which itbegins to decrease again. Although detailed model-ing is required to understand the cause of thesetrends, some of this variation could be the resultof adsorbed species being released by solar windsputtering as particles spend more time on thesurface than on ballistic trajectories. The absenceof sputtering in the magnetotail should allow forthe Na surface reservoir to increase, perhaps ex-plaining the increase in exospheric concentrationseveral days after the full Moon; then, as the Mooncomes out of the magnetotail, sputtering beginsto release adsorbed particles again and exosphericNa decreases as it is lost to space. This paradigmsuggests then that most of the Na particles donot get lost to the surface on their first bounce,unlike hypothesized in recent models based onground-based observations (7, 15). It was sug-gested that some of the daily variation observedby Kaguya could be explained if there were surficialenhancements of Na in selenographic longitudesaround 90°± 90°, resulting in differences in PSDrates between near-side and far-side regolith (9).Consistent with this idea of a surface dependency,the new LADEE UVS data are suggestive of a dif-

ferent dependence in PSD between mare versushighlands soils (Fig. 3).Potassium shows a much more regular trend

over time, with intensity variations over a luna-tion of approximately a factor of 2 (Fig. 2). Thismore systematic monthly trend for K (comparedto Na) is reminiscent of the extreme variation ofK surface abundance due to the enrichment foundin KREEP (Potassium Rare Earth Elements andPhosphorus) soils (19), which can be as much asa factor of 10. Any variation introduced by thesolar wind appears to be masked by the strongvariation of surface K. In addition to the much

larger variation in K over the course of a lunation,the K dependence on surface composition is moreclearly shown by the location of the peak tangentcolumn densities (Fig. 3). Given the anticipatedinfluence of the magnetotail on sputtering (7, 18),a minimum in column density for both Na andK could be expected to be centered at around alongitude of 0°, with maxima to either side. How-ever, the K peak occurs to the west, centered onthe Oceanus Procellarum and the Mare Imbriumregions, areas of maximum surface K as measuredby the Lunar Prospector mission (20). Sodium ismore symmetric about 0° longitude, with a local

250 15 JANUARY 2016 • VOL 351 ISSUE 6270 sciencemag.org SCIENCE

Fig. 2. The total line-of-site column densities for sodium and potassium. (A and B) Sodium. (C) Po-tassium. The column densities were derived from data taken while the telescope grazing point wasbetween a spacecraft solar longitude of 165° to 180° (near local noon). In (A), the last three observedlunations are shown with green-shaded regions indicating when the Moon was in Earth’s magnetotail.(B) and (C) show the Na and K column densities, respectively, for the entire mission period. In (B), afit to the minima in column concentration (solid blue curve) is shown to highlight the long-term Natrend. Also shown in (B) and (C) are the approximate beginning and end dates (gray-shaded regions)for three meteoroid streams (Leo, Gem, and Qua) and the observed peak (blue dashed lines) and theapproximate dates of the full Moon (red arrows). Average absolute and relative (point-to-point)uncertainties are shown by the red points toward the upper right corner of (B) and (C).

RESEARCH | REPORTSon June 5, 2019

http://science.sciencemag.org/

Dow

nloaded from

minimum at 0°. There is some correlation of Nawith surface albedo (Fig. 3), which suggests apossible dependence on composition (e.g., Mareor Highlands composition); however, albedo isalso a function of surface maturity and roughness,and these factors are expected to influence howadsorbates bind on grains. Additionally, the struc-ture around 0° longitude could also be a reactionto being inside the magnetotail.

Some individual meteoroid streams led to asignificant increase in these exospheric species.Three major streams that occurred during theLADEE mission are indicated in Fig. 2. In theNa data set, although there are peaks in the Nacolumn abundance, it is not obvious that theseare the result of the streams (as opposed to what-ever is causing the monthly variation). Indeed thenoon-time immediate response to streams of an

exosphere populated by previously adsorbed atomsis expected to be rather small, because most streamshave radiants on the lunar nightside (21, 22), withlarger variations in response to the streams ex-pected at around dawn. However, in the K noon-time data there is a clear increase at around thetime of the Geminid (Gem) meteoroid stream (ap-proximately between 4 and 16 December) andalso smaller signatures at around the time of theLeonid (Leo) (6 to 30 November) and Quadrantid(Qua) (1 to 10 January) streams. In addition tothese enhancements at or around individual streams,the overall rise in Na from October 2013 to a peakin December 2013, and then a gradual declineuntil the end of the mission, can be attributedto a cumulative response to meteoroid streams.A similar trend was apparent in Kaguya data (9),but, as with earlier studies (17), it could not becorrelated to any particular process. The UVS datasuggest a strong link between these streams andthe exosphere, lasting far beyond the initial en-counter. The meteoroid impacts expose fresh Naand K, but also supply these species directly tothe exosphere at increased rates.The large instantaneous responses of the

dayside to meteor showers are indirect evidenceof the role of adsorbed particles. If all particleswere lost to the surface in one bounce, it wouldtake more than an order of magnitude enhance-ment of impact vaporization during showersover sporadic impacts for this source to reachthe rate of 2 × 106 Na atoms cm−2 s−1 requiredto explain the Na atmosphere (7). Hence, vaporintroduced by meteoroid impacts can residefor considerable periods on the surface, addingto the overall adsorbed Na and K surface res-ervoir. We performed Monte Carlo simulationsto estimate the total residence time for Na andK in the exosphere and soil. In the simulations,a one-second injection of Na or K was assumedto occur from the nightside hemisphere, withthe duration between hops and loss rates forsputtering and photoionization taken frompublished values (see the supplementary ma-terials). The simulations show that a release ofejecta from a single injection will persist in theexosphere-surface system for much longer thanthe ionization lifetime would suggest (Fig. 4).This is the result of each particle residing inthe soil for approximately an ionization lifetime(i.e., several days) between bounces, combinedwith the many bounces that it has to take beforebeing lost from the exosphere. Residence timesin the lunar environment of 45 to 90 days(mainly on the lunar surface) can be expectedbefore escape to the solar wind, which wouldexplain the long-term smooth increase and de-crease in the Na column density observed asthe result of meteoroid streams (Fig. 2). Shortertimes are predicted if some fraction (but smallerthan 50%) of recycled particles are trapped per-manently between bounces.These observations provide constraints on the

sources and sinks of Na and K in the lunar exo-sphere. The nearly continuous, synoptic nature ofthe observations illustrates the contribution ofimpacts and surface composition. The independent

SCIENCE sciencemag.org 15 JANUARY 2016 • VOL 351 ISSUE 6270 251

Fig. 4. The lifetimes of sodium released by meteoroid impacts, as estimated with a Monte Carlomethod. In these simulations, the effect of a meteoroid stream is modeled by a sudden release of Natest particles, which are tracked until loss by photoionization, sputtering, or deposition into permanentlyshadowed regions (18). The fraction of atoms lost on the first bounce much exceeds gravitationalescape, reflecting losses to sputtering after particles recycle for the first time (A). Prolonged residence ofreleased Na on surficial grains between bounces would continue to affect the exosphere well after themeteoroid stream encounter, with an exponential time decay as long as 90 days (B).

Fig. 3. Column densities for sodium and potassium as a function of selenographic longitude.(A) Sodium. (B) Potassium. The approximate entry into (blue squares) and exit out of (red points)Earth’s magnetotail are shown. Data was acquired at about solar noon. Data acquired during theGeminids stream are indicated with an X or a +. Also shown is the scaled surface albedo averagedbetween –15° and –22° latitude [green line in (A)] and in the relative concentration of K at the latitudeof –20° in the lunar soil as a function of selenographic longitude as derived from Lunar Prospectorobservations [green line in (B)].

RESEARCH | REPORTSon June 5, 2019

http://science.sciencemag.org/

Dow

nloaded from

confirmation by Kaguya and LADEE of the lunarNa trend between November and April providesthe strongest evidence yet for an annual varia-tion of the Na exosphere. This trend is likely thecumulative response of Na to meteoroid streams,whose annual activity peaks fromNovember throughJanuary and then subsides until the summer. Thesubstantial residence time for Na at the surfacesuggested by this interpretation inevitably leadsto the conclusion that Na migrates toward thepoles like other volatiles (e.g., water) in these cyclesof adsorption and redesorption. The K measure-ments show a strong but, contrary to Na, short-lived response to the Geminids meteoroid shower.Outside of the meteoroid streams, K shows aregular variation across a lunation that correlatesstrongly with the abundance of potassium in thelunar bulk soil. Combined, these results and re-cent studies of the Mercurian exosphere (23, 24)indicate a pronounced role for meteoroid impactvaporization and surface exchange in determiningthe composition of surface-bounded exospheres.However, the details of how the exosphere dependson surface composition and responds to meteo-roid streams are not yet understood.

REFERENCES AND NOTES

1. R. C. Elphic et al., Space Sci. Rev. 185, 3–25 (2014).2. A. Colaprete et al., Space Sci. Rev. 185, 63–91 (2014).3. A. E. Potter, T. H. Morgan, Science 241, 675–680 (1988).4. A. S. Stern, Rev. Geophys. 37, 453–491 (1999).5. A. L. Sprague, M. Sarantos, D. M. Hunten, R. E. Hill,

R. W. H. Kozlowski, Can. J. Phys. 90, 725–732 (2012).6. D. Hunten, A. L. Sprague, Adv. Space Res. 19, 1551–1560

(1997).7. M. Sarantos, R. M. Killen, A. S. Sharma, J. A. Slavin, Icarus 205,

364–374 (2010).8. M. Kagitani et al., Earth Planets Space 61, 1025–1029

(2009).9. M. Kagitani et al., Planet. Space Sci. 58, 1660–1664

(2010).10. R. W. H. Kozlowski, A. L. Sprague, D. M. Hunten, Geophys. Res.

Lett. 17, 2253–2256 (1990).11. A. L. Sprague, R. W. Kozlowski, D. M. Hunten, W. K. Wells,

F. A. Grosse, Icarus 96, 27–42 (1992).12. D. M. Hunten et al., Icarus 136, 298–303 (1998).13. S. M. Smith, J. K. Wilson, J. Baumgardner, M. Mendillo,

Geophys. Res. Lett. 26, 1649–1652 (1999).14. S. Verani, C. Barbieri, C. Benn, G. Cremonese, Planet. Space

Sci. 46, 1003–1006 (1998).15. V. Tenishev, M. Rubin, O. J. Tucker, M. R. Combi, M. Sarantos,

Icarus 226, 1538–1549 (2013).16. A. A. Berezhnoy et al., Planet. Space Sci. 96, 90–98

(2014).17. M. Matta et al., Icarus 204, 409–417 (2009).18. Materials and methods are available as supplementary

materials on Science Online19. A. E. Potter, R. M. Killen, T. H. Morgan, J. Geophys. Res. 105

(E6), 15073–15084 (2000).20. B. L. Jolliff, J. J. Gillis, L. A. Haskin, R. L. Korotev,

M. A. Wieczorek, J. Geophys. Res. 105 (E2), 4197–4216(2000).

21. D. J. Lawrence et al., Science 281, 1484–1489 (1998).22. T. J. Stubbs et al., Influence of Meteoroid Streams on the

Lunar Environment: Results from LADEE, 46th Lunarand Planetary Science Conference, held March 16–20, 2015,in The Woodlands, Texas. LPI Contribution No. 1832, p. 2984(2015).

23. M. Horányi et al., Nature 522, 324–326 (2015).24. R. M. Killen, J. Hahn, Icarus 250, 230 (2015).

ACKNOWLEDGMENTS

We thank R. Killen for constructive discussions and the three reviewerswho helped to greatly improve this paper. LADEE UVS wassupported through the NASA Lunar Quest Program. Additional funding

for M.S. was through NASA grants NNX14AG14A, NNX13AP94G,and NNX13AO74G. All LADEEUVS data are available online at the NASAPlanetary Data System (PDS), including all raw and calibratedspectra and derived sodium and potassium line strengths.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/351/6270/249/suppl/DC1Materials and Methods

Supplementary TextFigs. S1 to S5References (25–34)

13 August 2015; accepted 30 November 2015Published online 17 December 201510.1126/science.aad2380

ORGANIC CHEMISTRY

Functionalization of C(sp3)–H bondsusing a transient directing groupFang-Lin Zhang,* Kai Hong,* Tuan-Jie Li,* Hojoon Park, Jin-Quan Yu†

Proximity-driven metalation has been extensively exploited to achieve reactivity andselectivity in carbon–hydrogen (C–H) bond activation. Despite the substantialimprovement in developing more efficient and practical directing groups, theirstoichiometric installation and removal limit efficiency and, often, applicability as well.Here we report the development of an amino acid reagent that reversibly reacts withaldehydes and ketones in situ via imine formation to serve as a transient directinggroup for activation of inert C–H bonds. Arylation of a wide range of aldehydes andketones at the b or g positions proceeds in the presence of a palladium catalystand a catalytic amount of amino acid. The feasibility of achieving enantioselectiveC–H activation reactions using a chiral amino acid as the transient directing groupis also demonstrated.

Precoordination of a metal with functionalgroups in substrates has been extensivelyexploited to control selectivity and promotereactivity in metal-catalyzed or -mediatedreactions (1–5). The same approach has

been successfully implemented in directed C–Hactivation reactions (6–11). However, the cova-lent installation and removal of directing groupsis a major drawback for synthetic applications.First, an additional two steps must be added tothe synthetic sequence. Second, the conditionsfor installation or removal of the directing groupsare sometimes incompatiblewith other functionalgroups present in advanced synthetic intermedi-ates. It is therefore highly desirable to devise afunctionally tolerant reagent that can be reversi-bly linked to the substrate and can serve as adirecting group. Upon C–H activation and sub-sequent functionalization, this reagent woulddissociate from the product and transiently linkto another substrate molecule so that only a cata-lytic quantity of the directing group would beneeded (Fig. 1A). This approach has been success-fully implemented in Rh(I)-catalyzed C(sp2)–Hactivation reactions in a number of pioneeringexamples. Jun et al. reported the use of 2-aminopyridine as a transient directing group for Rh-catalyzed activation of aldehydic C–H bonds (12)(Fig. 1B). Recently, using a related strategy, Moand Dong reported a Rh-catalyzed a-alkylation of

ketones via a vinyl C–H activation step, featuringan enamine intermediate with a pyridine moietyas the transient directing group (13). Bedford et al.developed a Rh-catalyzed ortho-arylation throughreversible in situ transesterification of catalyticamounts of phosphinite ligands with the phenolsubstrate (14). The strategy of using catalyticdirecting groups has also been employed byLightburn et al. (15) and Grünanger and Breit (16)to achieve selectivity inRh-catalyzedhydroformyla-tion reactions.In conjunction with our efforts to develop

Pd-catalyzed C(sp3)–H functionalizations (17, 18),we have extensively investigated the feasibility ofPd(II)-catalyzed C(sp3)–H activation of aldehydesand ketones using a wide range of potentialtransient directing groups, including those pre-viously developed for Rh(I) catalysts. Unfortu-nately, the resultant Pd(II) complexes bound tothe bidentate iminopyridine or iminooxazolineare unreactive toward cleavage of sp3 C–H bondsunder various conditions. The development ofmonoprotected amino acid ligands (19, 20) andthe recent use of amino acids as bidentate di-recting groups in C–H functionalizations of pep-tides (21) led us to speculate that an amino acidcould serve as a suitable transient directing group.We reasoned that the amino acid could be re-versibly tethered to an aldehyde or ketone sub-strate via an imine linkage under appropriateconditions. In a similar manner to that operativein our dipeptide chemistry, the iminemoiety andthe carboxylate could form a bidentate directinggroup to enable subsequent C–H functionaliza-tion (Fig. 1C).

252 15 JANUARY 2016 • VOL 351 ISSUE 6270 sciencemag.org SCIENCE

The Scripps Research Institute, 10550 North Torrey Pines Road,La Jolla, CA 92037, USA.*These authors contributed equally to this work.†Corresponding author. E-mail: yu200@scripps.edu

RESEARCH | REPORTSon June 5, 2019

http://science.sciencemag.org/

Dow

nloaded from

exosphereHow surface composition and meteoroid impacts mediate sodium and potassium in the lunar

A. Colaprete, M. Sarantos, D. H. Wooden, T. J. Stubbs, A. M. Cook and M. Shirley

originally published online December 17, 2015DOI: 10.1126/science.aad2380 (6270), 249-252.351Science 

, this issue p. 249; see also p. 230Scienceto sunlight. There are also increases shortly after the Moon passes through streams of meteoroids.exosphere composition varies by a factor of 2 to 3 over the course of a month, as different parts of the Moon are exposedby using the glow from sodium and potassium atoms as a probe (see the Perspective by Dukes and Hurley). The

have used NASA's LADEE orbiter to investigate how the exosphere varies over time,et al.from the surface. Colaprete Earth's Moon does not have a conventional gaseous atmosphere, but instead an ''exosphere'' of particles ejected

The Moon's time-variable exosphere

ARTICLE TOOLS http://science.sciencemag.org/content/351/6270/249

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2015/12/16/science.aad2380.DC1

CONTENTRELATED

file:/contentpending:yeshttp://science.sciencemag.org/content/sci/351/6270/230.full

REFERENCES

http://science.sciencemag.org/content/351/6270/249#BIBLThis article cites 31 articles, 2 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.Sciencelicensee American Association for the Advancement of Science. No claim to original U.S. Government Works. The title Science, 1200 New York Avenue NW, Washington, DC 20005. 2017 © The Authors, some rights reserved; exclusive

(print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

on June 5, 2019

http://science.sciencemag.org/

Dow

nloaded from