Lunar pickup ions observed by ARTEMIS

13 Pages·2013·0.82 MB·English

by  J. S. Halekas

https://ntrs.nasa.gov/search.jsp?R=20140010168 2018-11-18T06:30:51+00:00Z
JOURNALOFGEOPHYSICALRESEARCH,VOL.117,E06006,doi:10.1029/2012JE004107,2012
Lunar pickup ions observed by ARTEMIS: Spatial and temporal
distribution and constraints on species and source locations
J. S. Halekas,1,2 A. R. Poppe,1,2 G. T. Delory,1,2 M. Sarantos,2,3,4 W. M. Farrell,2,3
V. Angelopoulos,5 and J. P. McFadden1
Received20April2012;revised14May2012;accepted15May2012;published30June2012.
[1] ARTEMIS observes pickup ions around the Moon, at distances of up to 20,000 km
fromthesurface.Theobservedionsformaplumewithanarrowspatialandangularextent,
generally seen in a single energy/angle bin of the ESA instrument. Though ARTEMIS
hasnomassresolutioncapability,wecanutilizetheanalyticallydescribablecharacteristics
ofpickupiontrajectoriestoconstrain thepossibleionmassesthatcanreachthespacecraft
at the observation location in the correct energy/angle bin. We find that most of the
observations are consistent with a mass range of (cid:2)20–45 amu, with a smaller fraction
consistent with higher masses, and very few consistent with masses below (cid:2)15 amu.
With the assumption that the highest fluxes of pickup ions come from near the surface,
the observations favor mass ranges of (cid:2)20–24 and (cid:2)36–40 amu. Although many of the
observations have properties consistent with a surface or near-surface release of ions,
somedonot,suggestingthatatleastsomeoftheobservedionshaveanexosphericsource.
Of all the proposed sources for ions and neutrals about the Moon, the pickup ion flux
measured by ARTEMIS correlates best with the solar wind proton flux, indicating
that sputtering plays a key role in either directly producing ions from the surface,
or producing neutrals that subsequently become ionized.
Citation: Halekas,J.S.,A.R.Poppe,G.T.Delory,M.Sarantos,W.M.Farrell,V.Angelopoulos,andJ.P.McFadden(2012),
LunarpickupionsobservedbyARTEMIS:Spatialandtemporaldistributionandconstraintsonspeciesandsourcelocations,
J.Geophys.Res.,117,E06006,doi:10.1029/2012JE004107.
1. Introduction particle and photon sputtering, micrometeorite impact, ther-
maldesorption,andvolatilereleasefromtheinterior,among
[2] DespitetheseemingsimplicityoftheMoon-solarwind other processes [Stern, 1999]. Therefore, by measuring
interaction,thelunarenvironmenthasaricharrayofdiverse
pickupionsaroundtheMoon,wecangainknowledgeabout
and interesting physical processes, many resulting from the
allofthesefundamentalphysicalprocesses,andtiecharged
directbombardmentofthelunarsurfacebychargedparticles
particledatabacktothelunarsurfaceandexosphere.
andother impactors such asmicrometeorites andsolar pho-
[3] Westillhavearelativelypoorinventoryofthespecies
tons[Halekasetal.,2011].Ionsoflunaroriginrepresentone
that comprise the tenuous lunar exosphere. As summarized
such phenomenon with linkages between the surface, the
in Table 1, we know from a combination of ALSEP surface
plasma, the neutral exosphere, and all of these external dri-
measurements [Hoffman et al., 1973; Hoffman and Hodges,
vers. Lunar ions originate from solar wind sputtering from
1975; Freeman and Benson, 1977] and remote spectroscopy
the surface, and from ionization and dissociation of neutral
[Fastieetal.,1973;FeldmanandMorrison,1991;Flynnand
exosphericgases,whichthemselvesarereleasedbycharged
Stern, 1996; Stern et al., 1997; Potter and Morgan, 1988;
Mendillo et al., 1991] that the most abundant species posi-
1Space Sciences Laboratory, University of California, Berkeley, tivelydetectedsofarintheexosphereareHeandAr.Onthe
California,USA. other hand, the species we know most about, the alkalis Na
2NASA’s Lunar Science Institute, NASA Ames Research Center, andK,havesmallnear-surfacedensities.Thenumberofions
MoffettField,California,USA. produced from each exospheric species depends on their
3HeliophysicsScienceDivision,NASAGoddardSpaceFlightCenter,
neutral density and their scale height (together giving the
Greenbelt,Maryland,USA.
4Goddard Planetary Heliophysics Institute, University of Maryland, column density) as well as the total ionization rate of that
BaltimoreCounty,Baltimore,Maryland,USA. species (Table 1 summarizes photo-ionization rates, which
5Department of Earth and Space Sciences, University of California, usually dominate over charge exchange and electron impact
Los Angeles, California, USA.
ionization). The scale height (or scale heights, if multiple
Correspondingauthor:J.S.Halekas,SpaceSciencesLaboratory, processes operate) depends on particle mass and on the pro-
7GaussWay,UniversityofCalifornia,Berkeley,CA94720,USA. cess-dependentstartingenergydistributionforreleasedgases,
([email protected])
and is not well known for many species. The total neutral
©2012.AmericanGeophysicalUnion.AllRightsReserved. column density for the alkalis Na and K, which have large
0148-0227/12/2012JE004107
E06006 1of13
E06006 HALEKASETAL.:LUNARPICKUPIONSOBSERVEDBYARTEMIS E06006
Table1. ExpectedLunarPickupIonSpeciesa 2009; Wang et al., 2011], and O+, Al+, and Si+ identified
at greater distances by AMPTE and Wind [Cladis et al.,
NeutralDensity PhotoRate
(cm(cid:4)3) Coefficient(s(cid:4)1) SputterYield 1994; Hilchenbach et al., 1991, 1993; Mall et al., 1998].
Many of these species correspond to basic expectations,
Mass1(H+) <17 7.3(cid:5)10(cid:4)8 given the estimated neutral abundances, photo-ionization
Mass2(H+) <9000 5.4(cid:5)10(cid:4)8
Mass4(H2e+) 2000 5.3(cid:5)10(cid:4)8 rates, and sputter yields shown in Table 1, while some do
Mass12(C+) <200 4(cid:5)10(cid:4)6 not. In particular, the significant fluxes of O+ and C+
Mass14(N+) <600 1.9(cid:5)10(cid:4)7 observed at high altitudes by previous missions appear
Mass16(O+) <500 2(cid:5)10(cid:4)7 somewhat unexpected, given their low neutral abundances
Mass16(CH+) <10,000 3.6(cid:5)10(cid:4)7
Mass17(OH4+) <1(cid:5)106 2.4(cid:5)10(cid:4)7 [FeldmanandMorrison,1991]andlowexpectedsputtering
Mass23(Na+) 70 1.6(cid:5)10(cid:4)5 7(cid:5)10(cid:4)6 yields [Elphic et al., 1991]. Some ofthese ions could result
Mass24(Mg+) <6000 9(cid:5)10(cid:4)6 from photo-dissociation of molecular species [Hoffman and
Mass27(Al+) <55 1.5(cid:5)10(cid:4)5 Hodges,1975; Hodges,1976].
Mass28(Si+) <48 6(cid:5)10(cid:4)6
Mass28(N+) 800 3.5(cid:5)10(cid:4)7
Mass29(CO2+) <1000 4(cid:5)10(cid:4)7 2. ARTEMIS Observations of Lunar Pickup Ions
Mass39(K+) 17 2(cid:5)10(cid:4)5 3(cid:5)10(cid:4)6
Mass40(Ca+) <1 2(cid:5)10(cid:4)6 [6] The two-probe ARTEMIS (Acceleration, Reconnec-
Mass40(Ar+) 100,000 3(cid:5)10(cid:4)7 tion, Turbulence, and Electrodynamics of Moon’s Interac-
Mass44(CO+) <1000 6.5(cid:5)10(cid:4)7 tionwiththeSun)mission[Angelopoulos,2011]repurposed
Mass56(Fe+2) <380 3(cid:5)10(cid:4)6 two spacecraft from the original five-probe THEMIS con-
aDaysideneutraldensitiesafterStern[1999]withvaluesfromFeldman stellation on an extended mission to the Moon. The two
and Morrison [1991], Hoffman and Hodges [1975], Hoffman et al. ARTEMIS probes, which entered lunar orbit in June and
[1973], Potter and Morgan [1988], Fastie et al. [1973], Stern et al. July of 2011 in highly elliptical equatorial orbits, provide
[1997],andFlynnandStern[1996].Quietsunphoto-ionizationratesare comprehensive plasma measurements, including three-
fromHuebneretal.[1992],andsputteryieldsarefromElphicetal.[1991].
dimensional ion flux measurements from the ESA instru-
ment [McFadden et al., 2008a] and 3-axis magnetic field
measurements from the MAG instrument [Auster et al.,
scale heights [Potter and Morgan, 1988; Stern and Flynn,
2008]. In mid-September 2011, the ESA instruments were
1995], may in fact compare to that of many other species
put into magnetospheric mode near the Moon [McFadden
more abundant in the lunar regolithy, especially if photo-
et al., 2008a], allowing observation of pickup ions. In
stimulateddesorptionplaysanimportantrole[Sarantosetal.,
magnetospheric mode, the ESA instruments measure an
2010, 2012]. In that case, given their very large photo-ioni-
zationrates,weshouldexpectsubstantialfluxesofNa+andK+ energy range extending up to (cid:2)25 keV with full angular
coverage, with (cid:2)30% energy resolution and 22.5(cid:3) resolu-
ionsaroundtheMoon.
tion. Unfortunately, the ESA instrument has no mass-
[4] Charged particle sputtering from the surface also
resolvingcapabilities,sowemustinferthespeciesofpickup
directlyreleasessomeions,thereforeproducingpickupions
ions fromother characteristics of theobservation.
withouttheintermediatestagesofresidenceintheexosphere
and subsequent ionization. Much of our knowledge of this [7] Figure1showsasimultaneouspickupionobservation
byARTEMISprobesP1andP2,andFigure2showstheorbits
processcomesfromalaboratoryexperimentbyElphicetal.
andfieldgeometryforthisobservation.Thiseventillustrates
[1991].Chargedparticlesputteringoflunarelementsmainly
releasesions withlowionization energies, suchasAl+, Si+, the defining characteristics of all ARTEMIS pickup ion
and Mg+ (see Table 1 for sputter yields from lunar materi- observationsconsideredinthispaper.Givenplasma(e.g.,the
solar wind) with a magnetic field B flowing past the Moon
als). The number of ions produced by sputtering depends
withvelocityU,aconvectionelectricfieldE=(cid:4)U(cid:5)Bexists
directly on the flux of incoming ions, and, to a degree, on
intheMoonframe.PickupionsrespondtoEandB,acceler-
their mass and energy distribution, which can have impor-
atingalongtheelectricfieldEfromtheirproductionpointand
tant implications during major solar events [Killen et al.,
subsequently following cycloidal trajectories in the plane
2012]. The relative contributions to pickup ion fluxes from
containing the electric field E and the solar wind velocity U
surface sputtering and photo-ionization depend on ion spe-
andperpendiculartoB.ForARTEMIStoeasilyobservethese
cies, but could have similar magnitudes in some cases
ions their trajectories must lie close to the nearly equatorial
[Yokota and Saito, 2005].
(XY) plane of the probe orbits; therefore B must have a
[5] Onceborn(ionized),ionsfeeltheeffectofelectricand dominant +/(cid:4)Z-component, and E a correspondingly domi-
magnetic fields in the ambient plasma. Depending on the
nant +/(cid:4)Y-component. Since pickup ion trajectories bend
electric field, these ions either re-impact the surface or fol-
backfromthe y axis,ARTEMIS usuallyobserves these ions
lowcycloidaltrajectoriesextendingdownstreamintheplane
intheirfirst gyroorbit atlarge +/(cid:4)Y and somewhat negative
of the electric field, with the scale and bend-back of the
X coordinates. The pickup ions form a highly collimated
cycloiddependingonionspecies.Thesepickupionsprovide
beam, generally producing a signal in only a single energy-
ausefultooltostudythelunarsurfaceandexosphere[Hartle
angle bin of the ESA instrument. At a few times, ions span
andKillen,2006].IonsobservedneartheMoonspanawide
varietyofspecies,withmass20–28and40–44atomicmass multiple energybins, indicatingeitheranextendedsourceof
pickup ions, fluctuations in the fields during the ESA inte-
units (amu) seen at the surface by the ALSEP SIDE instru-
gration period, or simply an ion energy falling near the
ment [Freeman and Benson, 1977; Vondrak et al., 1974],
H+, He+, C+, O+, Na+, and K+ reported at (cid:2)100 km by boundarybetweentwoenergybins.
2
Kaguya and Chang’E [Yokota et al., 2009; Tanaka et al.,
2of13
E06006 HALEKASETAL.:LUNARPICKUPIONSOBSERVEDBYARTEMIS E06006
atlargelateraldistancesonlyhighermassionscanreachthe
spacecraft from near the surface. A pickup ion follows a
trajectorywithadriftvelocityof(E(cid:5)B)/B2(orequivalently
U , the component of the plasma flow velocity perpen-
perp
diculartothemagneticfield).Pickupiontrajectorieshavea
cycloidal form, scaling with the ion gyroradius
R =Mm U /(qB)=MR (whereMisionmassnumber
M P perp P
andm andR arethemassandperpendiculargyroradiusof
P P
a solar wind proton), and lie in the plane of E and U, with
maximum displacement relative to the drift vector of 2R
M
(or 2MR ) in the direction of E. For the flow speeds of
P
(cid:2)320 km/s and the magnetic fields of (cid:2)7 nT during this
event, R = (cid:2)500 km. Therefore, the lateral distance of
P
4,000–6,000 km from the lunar surface indicates that no
pickupionswithM<4–6canreachthespacecraftfromnear
the surface on this occasion. We will see that the observed
Figure 1. A simultaneous pickup ion observation by both
ARTEMIS probes, showing spacecraft position, magnetic
field, derived convection electric field, and ion differential
energy flux [eV/(cm2 s sr eV)] for P1 and P2 as a function
ofenergyandphiangle(for10–25keVions)intheequato-
rialplane.Dashedblacklinesonenergyspectrashowkinetic
energy gained from electric field for ions born at the sub-
solar point. All vector quantities use SSE (Selenocentric
Solar Ecliptic) coordinates.
[8] ThekineticenergyperchargeW/qgainedbyapickup
ion depends only on the electric field E and the distance
traveled frRom the source along that field. The quantity
DW=q¼ ~E⋅d~risshownontheenergyspectrainFigure1
for a source location at the sub-solar point. The energy per
charge of the observed beam matches that expected for
pickup ions coming from near the Moon, supporting our Figure 2. ARTEMISP1andP2orbits fortheevenshown
identification. Unfortunately, since this quantity does not in Figure 1, in dashed blue and dashed orange respectively.
dependonmass,italonedoesnothelptoidentifythespecies Theblackandsoliddarkorangesectionsshowthelocations
oftheions. ofpickupionsseenbyP1andP2respectively.Black,green,
[9] The characteristic radius of the ion trajectory, how- and red vectors show the solar wind velocity, the interplan-
ever,andthusthedegreetowhichitbendsastheiontravels etary magnetic field, and the convection electric field
outfrom theMoon, depends on theion’smass. As a result, derived fromthese twoquantities.
3of13
E06006 HALEKASETAL.:LUNARPICKUPIONSOBSERVEDBYARTEMIS E06006
pickup ion velocity angles provide a more stringent con- E=(cid:4)U(cid:5)Bfromthesequantities.SincetheESAdoesnot
straint, since they indicate that the observed pickup ions resolve the off-axis components of the solar wind velocity
havenotyetreached thepeakoftheir cycloidalmotion. well inmagnetospheric mode,we usetheknown aberration
[10] Meanwhile,aprotonreflectedfromthesurface[Saito velocity of (cid:2)30 km/s associated with the Earth’s orbit
etal.,2008]orfromamagneticanomaly[Lueetal.,2011], around the Sun, and assume no other off-axis flow compo-
which can have a starting energy as large as that of the nents. We have tested with various assumptions and found
incidentsolarwindprotons,canreachalateraldistanceofat that the small off-axis components expected in the solar
most 4R (for reflection in the opposite direction from the winddonotsignificantlyaffectanyofourresults.Evennear
P
solar wind, and thus the largest possible starting velocity in the edge of the magnetosheath, where the event of
the solar wind frame). Therefore, no reflected solar wind Figures 1–2 took place, off-axis components did not reach
protonscanreachthespacecraftfromnearthesurfaceatthis significant enoughlevels toalter ourresults.
time.Thisisanimportantconclusion,sincereflectedprotons [15] Our analytic trajectory calculation implicitly assumes
candominate overpickup ionsnear theMoon. steadyfieldsoverthetraveltimeofthepickupions.Wetyp-
[11] AtbothP1andP2,anintegrationoverthedistribution ically observe pickup ions with a gyrophase on the order of
function of the pickup ion plume gives a total density esti- (cid:2)60(cid:3), corresponding to a travel time of 1/(6*f ) = 2pMm /
ci P
mate varying from (0.002–.005)*√M cm(cid:4)3 over the course (6qB),whichworksouttoafewsecondsforlightionsandup
of the event, with M the unknown ion atomic mass (see toaminuteforheavierionsundertypicalconditions.Although
McFaddenetal.[2008b]forthecorrectmass-scalingofthe somesmall-scalesolarwindfieldfluctuationsexistonatime
density moment calculation). This density is quite large scale shorter than this, corresponding to some minor pertur-
compared to the peak exospheric pickup ion densities of bation of pickup ion trajectories, large scale fields generally
5(cid:5)10(cid:4)4forHe+and3(cid:5)10(cid:4)3forNa+estimatedbyHartle remainsteadyformuchlongertimeperiodsduringtheevents
et al. [2011], suggesting a contribution by multiple ion wehaveconsidered,allowingustoassumeconstantexternal
species,theexistenceofconditionsmuchmorefavorablefor conditionstoareasonablygoodapproximation.
exospheric pickup ion production than assumed by these [16] TheESAhasanintrinsicenergyresolutionof(cid:2)18%
authors, or a substantial contribution from direct surface and an intrinsic angular resolution of (cid:2)7(cid:3) and 5(cid:3)(cid:4)22.5(cid:3) in
sputtering. phi and theta. However, in magnetospheric mode, the pro-
cessor sums multiple energy/angle bins onboard while cre-
3. Constraining Lunar Pickup Ion Species ating the full 3-d distributions, resulting in a roughly
Gaussian energy response with a FWHM of (cid:2)30%, a
[12] Three challenges face us in using ARTEMIS obser- somewhat double-peaked phi response with total width of
vationstoidentifylunarpickupions.Firstandforemost,the 22.5(cid:3), and a nearly square theta response with a width of
ESA instrument does not measure mass composition 22.5(cid:3).Therefore,ionsmeasuredinagivenenergy/anglebin
directly. Furthermore, the instrument does not have a delta couldinprinciplehavelandedanywhereina22.5(cid:3) (cid:5)22.5(cid:3)
functionresponseinenergyorangle.Finally,theMoondoes angular bin and within an (cid:2)30% FWHM energy window.
not provide a delta-function source. All of these factors
For each possible ion mass, we can identify a bundle of
prevent the direct identification of pickup ion species.
trajectories consistent with these responsefunctions.
However,byutilizingthefactthatpickupionsoriginateator
[17] We show a sample of the resulting trajectories for a
near the lunar surface, with zero energy (to a very good
range of masses for one observation during our example
approximation),wecanpartiallyovercomethesechallenges.
eventinFigure3,forbothP1andP2.Foreachassumedion
[13] As described in Section 2, pickup ions follow easily mass, the range of allowed energy and angle for a single
analytically described cycloidal trajectories, starting with
ESAbincorrespondstoalineofionproductionpointslying
zerovelocityandreachingaknownpeakvelocity(twicethe
in the plane perpendicular to B that passes through the
drift velocity) at the top of the cycloid. Therefore, we will
spacecraft. It may seem surprising that these points form a
observe pickup ions with a precise relationship between
line, rather than a more extended region, but this simply
gyrophaseandkineticenergythatdependsionmass.Wecan
reflects the mathematical relationship between energy and
parameterize a pickup ion trajectory in uniform magnetic
gyrophase for a pickup ion that passes through a given
and electric fields as [X = X + R *(q-sin(q)),
P P0 M observation point, which removes a degree of freedom for
Y = Y + R *(1-cos(q))], where [X ,Y ] are coordinates
P P0 M P P thestartinglocations.Theseproductionpointscorrespondto
aligned with the drift velocity U and the electric field E
perp trajectories that reach the spacecraft in the observed energy
respectively,andqrepresentsthegyrophase.Thepickupion
bin, but with energies corresponding to a range of ESA
energy per charge W/q is simply E*(Y –Y ). Hence, the
P P0 sensitivities,withthepeakenergyresponsecorrespondingto
relationships between the energy of an ion and its arrival
the central part of the production line, as seen in Figure 3.
direction for a given mass and starting location allow us to We find that no ions with mass below (cid:2)20 amu can reach
constrainthespecies ofobservedions.Givenknowledge of
either spacecraft with the correct energy and angle, given
the fields, we can therefore analytically construct possible
theirsmallerbendradius.ForP1,norealpickupionswitha
trajectories for each ion species to determine which ions mass greater than (cid:2)40 amu can reach the spacecraft, since
have trajectories that can reach the spacecraft at the obser-
their trajectories would have had to start below the surface.
vation point and in the correct energy/angle bin, thereby
For P2, on the other hand, inferred trajectories start farther
constrainingthemostlikelymassrangeforobservedions,at
from the surface, and a much wider range of ion species
eachobservation time.
could reach thespacecraft inthecorrect energy/angle bin.
[14] In everything that follows, we use direct ARTEMIS [18] We show the resulting constraints on pickup ion
ESA and MAG measurements of U and B and derive
species and source altitude for this observation time as a
4of13
E06006 HALEKASETAL.:LUNARPICKUPIONSOBSERVEDBYARTEMIS E06006
Figure3. PickupiontrajectoriesthatcouldreachthetwoARTEMISprobes((left)P1;(right)P2)inthe
correctenergy/anglebinandatthecorrectlocationatonetimeduringtheeventofFigures1and2,fora
rangeofionmasses.Foreachmass,thetrajectoryshowncorrespondstothepeakoftheenergyresponse
forthebininwhichpickupionswereseen.Otherpointsindicatethestartinglocationsofothertrajectories
thatcouldreachthespacecraftinthecorrectenergy/anglebin.Blackv-shapesoutlinetheangularresponse
of theESAangle bininwhichpickup ions wereseen, for eachprobe.
functionofionmassinFigure4.Asseenabove,onlyionsin 2009]. Mg, Al, and Si also have respectable sputtering
the mass range of (cid:2)20–40 amu can reach P1 in the correct yieldsofions,andagreewithmoredistantWindobservations
energy/angle bin, whereas any mass > (cid:2)25 amu can reach [Malletal.,1998].TheaverageinferredstartingaltitudeforP2
P2inthecorrectbin,albeitwithratherlargestartingaltitudesfor lies well above the lunar surface, more consistent with exo-
someofthesepossiblemasses.TheP2startinglocationsdonot sphericsources,leadingustoatentativepreferenceforphoto-
overlapwiththoseforP1foranymass,suggestinganextended ionized exospheric Na and K for this event and potentially
sourceofionsand/ormultipleionspeciesatthistime.Wecan providinganexcitinglinktoearth-basedobservations.
already reach some tentative but interesting conclusions from
the mass/altitude constraints for this single observation. The 4. Spatial Distribution of Lunar Pickup
massrange20–40amu,agreeingwiththerangeof23–37amu Ion Observations
inferred indirectly by AMPTE [Hilchenbach et al., 1993],
contains several interesting species (see Table 1). The [19] We collected a data set of ARTEMIS pickup ion
molecularionsN+andCO+havenotbeenpreviouslyrepor- observations during the time range of 9/15/ 2011–2/5/2012
2 by visually searching all measured ion energy spectra for
ted,andhaverelativelylowphoto-ionizationrates,soappear
unlikely candidates. The alkali metals Na+ and K+, which narrow features withenergy perchargewell separated from
the solar wind proton and alpha particle populations, and
makegoodphysicalsensegiventheirhighsputteringyields
picking out the energy/angle bin with the peak pickup ion
and photo-ionization rates, agree with lower-altitude mea-
flux from each corresponding 3-d ion distribution. In prac-
surements fromKaguya[Yokotaetal., 2009;Tanakaetal.,
tice, we can easily pick out such features for any values
5of13
E06006 HALEKASETAL.:LUNARPICKUPIONSOBSERVEDBYARTEMIS E06006
electric field direction) from the lunar surface. This defini-
tively removes any observations contaminated by reflected
protons, but also effectively prevents us from observing
H+ from any location near the Moon, since H+ reaches a
2 2
maximumlateraldistanceof2R (H )=4R fromitssource.
M 2 P
[20] In principle, we can observe ions (for at least some
range of gyrophases) for any ion mass greater than 2.
However,twoeffectsmakeitsomewhatlesslikelyforusto
see very low-mass ions, in particular He+. First, removing
proton contamination erodes our sensitivity to the lowest-
massions,sincewecannotobservethematlowgyrophases
(low gyrophases map to small distances from the Moon,
especially for low-mass ions). In addition, for species that
have a neutral scale height comparable to the size of their
trajectories (usually, the lowest-mass ions), we may not
observeabeamassharplypeakedinenergy,makingitmore
Figure 4. The range of allowed ion masses and starting
difficult to visually identify pickup ions. The lateral scale
altitudes for the time shown in Figure 3, for each of the
size of the source region that produces trajectories observ-
two ARTEMIS probes. Solid lines indicate trajectories
able at a given location is at most a few times the scale
corresponding to the peak energy response for the bin in
heightforagivenspecies(seeFigure3,forexample).Ifthis
whichpickup ionswere seen, ortothehighest responsefor
lateral scale is not small compared to the distance to the
an allowed trajectory, for cases where the peak response
observation location, or equivalently if the energy width of
doesnotcorrespondtoanallowedtrajectory.Dottedvertical
the feature is not small compared to the average observed
lines show the range of starting altitudes for all trajectories
pickup ion energy, we cannot as easily pick out a narrow
with the given mass that could reach the spacecraft in the
pickup ionfeature intheenergy spectrum.
correct energy/angle binatthecorrect location. [21] For He+, which has a scale height of (cid:2)1000 km
[Stern, 1999], the energy width could reach (cid:2)2–4 kV for
above (cid:2)4–5 times the solar wind energy per charge. How- typical conditions (W/q = E*(Y –Y ), and E (cid:2)2 mV/m in
P P0
ever, in the end, we reject many low-energy observations thesolarwind).Asaresult,wewillonlyreliablypickouta
due to possible contamination by reflected protons. To narrow He+ feature in the energy spectra if it has a center
remove reflected proton contamination, we reject observa- energyontheorderof(cid:2)8kVorgreater.He+ionsreachthis
tions that lie less than 4R in lateral distance (along the energy for a portion of their orbits near the peak of their
P
Figure5. ARTEMISorbitsfromSeptember15th,2011throughFebruary5th,2012,forP1inblueand
P2 in green. Orange points indicate times and locations with conditions favorable for observation of
pickup ions by one of the probes (as described in text). Red points indicate times and locations where
one oftheARTEMIS probes actuallyobserved pickup ions.
6of13
E06006 HALEKASETAL.:LUNARPICKUPIONSOBSERVEDBYARTEMIS E06006
Figure6. Locationsandtrajectoriesof100pickupionplumesobservedbybothARTEMISprobesdur-
ingthetimeperiodofFigure5,incoordinatesnormalizedbythegyroradiusofasolarwindprotonR for
P
eachobservation.Coloreddashedcurvesshowarangeofsampletrajectoriesfordifferentionmasses,for
startpointsabovethe(left)terminatorand(right)sub-solarpoint,forelectricfieldspointinginthe+Yand
–Ydirections.
cycloidal motion, under typical conditions. Therefore, on thesecriteria,andshowtheminFigure5asacontrolsample.
average,wewilltendtohavelower(butnonzero)sensitivity The actual observations fill a narrower wedge than the
toHe+thantoheavierions.Asimilareffectcouldalsoreduce locations/times at which we could in principle see ions,
oursensitivitytoO+ifatomicOprovesveryhotandhighly suggestingthatarelativelynarrowrangeofionspeciesmay
dispersed,whichcouldoccurifthemainsourcemechanism contribute totheidentified observations.
forlunarOissputteringormoleculardissociation. [23] We also show the same observations in normalized
[22] After applying all of these constraints, we identify a coordinates, with velocity vectors, along with sample orbits
sampleof100individualpickupionobservations,witheach for several masses for comparison, in Figure 6. As for the
observationconsistingofasinglemeasurementofthethree- event in Figures 1–2, we note that most of the pickup ion
dimensional ion distribution by the ESA instrument. Given observations appear consistent with a narrow massrange of
the typical ESA survey mode cadence of a few minutes, (cid:2)20–40 amu, though for many we cannot rule out higher
some events have only a few individual observations. The masses,giventhebroadangularresponseoftheinstrument.
time period shown in Figures 1–2 has 38 between the two On the other hand, very few observations appear visually
probes (by far the most of any single time period). We dis- consistent with an ion mass below about 20 amu. For the
play the location of these observations in Figure 5. Though lowest few mass numbers, this may result partly from our
wedonotpre-applythefollowingselectioncriteria,wefind selection criteria, as described above. However, we still
that the vast majority of these events occur for times when might expect to observe even ions as light as He+ near the
themagneticfieldlieswithin25degreesofperpendicularto peakof theircycloidal motion.
the orbit plane, the electric field points toward the probe
from the Moon, the flow velocity exceeds 200 km/s, the 5. Inferred Distribution of Lunar Pickup Ion
density exceeds 0.5 cm(cid:4)3, the spacecraft lies in a wedge Species and Source Locations
from slightly forward of the Moon to 45 degrees behind it,
and the predicted pickup ion energy per charge E*Y lies [24] To quantify the ion masses that ARTEMIS could
P haveobservedforthese100pickupionevents,werepeatthe
between the maximum reflected proton energy/charge of 9
same analytical trajectory analysis described in section 3 to
timesthesolarwindvalueandthe25keV/qupperendofthe
findtherangeofmassesthatcouldreachthespacecraftatthe
ESA range. Accordingly, we collate all the times that meet
7of13
E06006 HALEKASETAL.:LUNARPICKUPIONSOBSERVEDBYARTEMIS E06006
case,astypifiedbytheP2example,providesfewconstraints
on mass, with the possible range extending to high masses,
coveringa broad range ofpossible starting altitudes.
[26] Wedividetheobservationsintothesetwogroups,and
firstconsiderthewell-constrainedpopulationthatsatisfythe
constraint of sM/〈M〉 <0.25 (where 〈M〉 and sM are the
mean and standard deviation of the allowed range of mas-
ses). For this population, we make the reasonable assump-
tion that the peak flux for each observation comes solely
fromthemassforwhichthetrajectorywiththepeakenergy
responseoriginatesatthelowestaltitude.Thisshouldprove
trueonaverageforbothsputtered(zerostartingaltitude)and
exosphericsources(higherneutraldensityclosertosurface).
This assumption results in the green mass histogram in
Figure 8 (left), containing only the well-constrained cases.
As an alternative, we construct a similar green mass histo-
graminFigure8(right)throughmuchthesamemethod,but
in this case we equally weight all masses for which the tra-
jectory with the peak energy response originated below
50km(if,foragivenobservation,notrajectoriesoriginated
below 50 km, we take the minimum altitude trajectory as
above). Since the predicted neutral density for many exo-
spheric species does not fall off steeply from the surface to
Figure 7. The solid black histogram shows the fraction of (cid:2)50kmaltitude[Sarantosetal.,2012],wemightexpectto
the 100 individual ARTEMIS pickup ion observations con- have a similar response to all species with starting altitudes
sistentwitheachionmass,takingintoaccountallconstraints in this range. Therefore, assuming we have significant con-
fromtrajectoryreconstructionandinstrumentresponse.The tributions from multiple ion species, Figure 8 (right) may
removal of proton contamination also removes virtually all provideasomewhatmoreunbiasedestimate.Inanycase,the
H+signalsandasignificantfractionofHe+signals,butdoes two methods result in quite similar spectra, indicating that
2
notsignificantly reduce sensitivity tohigher-mass ions. ourresult does notdependtoo stronglyon thealgorithm.
[27] To build up a plausible mass spectrum for all of our
observations, both well-constrained and otherwise, we take
correct location, in the correct energy/angle bin, for each theremainingpoorlyconstrainedobservations,andmakethe
observation. We show the results in cumulative form in assumption that the mass distribution of each individual
Figure 7. No single mass can explain all our observations, poorly constrained observation corresponds to that of the
indicating that ARTEMIS observes multiple ion species. well-constrained cases, over the mass range allowed for the
The largest fraction of our observations, 70–80%, are con- poorlyconstrainedobservation(eventhepoorlyconstrained
sistentwithamassrangeof(cid:2)20–45amu.Thecurveextends caseshavesomeconstraints).Wethendistributeeachpoorly
to relatively high masses at significant (cid:2)50% levels, since constrained observation over its allowed mass range
formanycaseswithhighmassionsatlowergyrophaseswe according to the portion of the well-constrained histogram
can achieve very little mass-discrimination using angular (greencurve)thatcoversthatmassrange.Inotherwords,for
considerations.Ontheotherhand,veryfewofourpickupion each poorly constrained data point, we take the portion of
observationscouldhavesignificantcontributionsfrommas- the green histogram that covers the mass range allowed for
ses below (cid:2)15. As discussed in Section 4, several factors that data point, and use that portion of the distribution to
mayconspiretoreduce(butnoteliminate)oursensitivityto construct a probability distribution for that data point, such
He+,butweshouldhaverelativelygoodsensitivitytomasses that the total of the weighted probability thus calculated
abovethis.Onlyabout35%oftheobservedionscouldhave sums to a total of one. After performing this fractional
massesbelow(cid:2)20amu,indicatingthatO+cannotcomprisea redistribution for each data point, and summing the results,
dominant fraction of the observed ions (unless the neutral wearriveatasecondmasshistogram(bluecurve).Wethen
sourcehashighlynon-thermalcharacteristics).Thiscontrasts sum these two histograms together to form a net histogram
with more distant Wind observations, which found a domi- (red curve).
nantO+component[Malletal.,1998]. [28] Weemphasizethatwehavenotactuallymeasuredthe
[25] We cannot say more with any certainty about the cumulative masshistogramsinFigure 8.Theonly thingwe
pickup ion species, beyond the curve shown in Figure 7. knowwithcertaintyisthedistributionofallowedmassesin
However, by making some reasonable assumptions, we can Figure7.However,bymakingtworeasonableassumptions,
converge on a narrower distribution of masses, which wehavearrivedatamoreconstrainedmasshistogram,with
though not definite, has plausible characteristics. First, we some very plausible properties. This favored spectrum has
notethattheobservationsroughlyseparateintotwogroups, clearpeaksatmassesof(cid:2)20–24amu,andmassesof(cid:2)36–
ofwhichtheP1andP2observationsshowninFigures3and 40 amu, which correspond directly to the masses for the
4providetypeexamples.Thefirstcase,astypifiedbytheP1 alkalis Na+ (23) and K+ (39). Given that these ions have
example, provides a relatively tight constraint on ion mass, predicted densities at altitude comparable to or greater than
and indicates relatively low starting altitudes. The second otherspeciesinthisrange[Sarantosetal.,2012],andphoto-
8of13
E06006 HALEKASETAL.:LUNARPICKUPIONSOBSERVEDBYARTEMIS E06006
Figure 8. Favored cumulative mass histograms for 100 ARTEMIS pickup ion observations, derived
from the assumption the source density should peak at low altitude. The green histograms include all
observations with a good constraint on ion mass (sM/〈M 〉 <0.25). (left) We place each data point at
themasscorrespondingtothetrajectoryofpeakenergyresponsewiththelowestallowedstartingaltitude
forthatobservationtoarriveatthegreenhistogram.(right)Weconstructthegreenhistograminasimilar
fashion,butifmorethanonemasscorrespondstoatrajectoryofpeakenergyresponsewithastartingalti-
tudebelow50km,weevenlyweightallsuchmasses.Thebluehistogramsinbothpanelsusetheproba-
bilitydistributionfromthewell-constrainedpointstoweighttheallowedrangeofmassesforeachofthe
restofthelesswell-constrainedobservations.Theredhistogramsshowtheresultingcumulativemassdis-
tributions. It isnotclear yetif these histograms reflect relative abundance.
ionization rates two orders of magnitude larger than other sputtering of ions from the surface. The inferred source
known species in this range [Huebner et al., 1992], these appears to exhibit a dependence on solar zenith angle as
speciesappearveryplausibleasanimportantcomponentof expectedforeitherdirectlysputteredionsorionizedneutral
lunarpickupionobservations.Ontheotherhand,Mg+,Ar+ exospheric constituents. A secondary spot at low altitudes
and Ca+ also lie in these mass ranges, and we cannot rule neartheterminatorcouldrepresentacontributionfromAr+.
them out, especially if direct ion sputtering dominates over
photo-ionization ofneutrals. 6. Correlation of Pickup Ion Observations
[29] Finally,asanadditionalexercise,wetaketherangeof With External Drivers
possiblesourcelocationscorrespondingtoeachpointinthe
histogramsofFigure8(left),anddistributetheminthesame [30] As we noted in section 4, we do not observe pickup
ions at all the times and locations for which favorable con-
fashion as in the construction of that curve to find a distri-
ditionsexist.Partly, as Figure5 suggests, thismay reflect a
bution of inferred source locations in Figure 9. This source
spatial constraint related to a relatively limited range of
distributionhasonlyasmuchvalidityasthemasshistogram
pickupionmasses.However,itcouldalsoindicatethatsome
in Figure 8, and is subject to all the same assumptions and
solarwindand/orilluminationconditionsleadtoanincreased
caveats. Despite these limitations, we again arrive at a
chanceofobservingpickupions,ortoahigherfluxofpickup
plausible result. The raw occurrence rate (Figure 9, top)
ions.Weconsiderthispossibilityinthisfinalsection.
peaksnearthesurface,andextendsbroadlyabovethesunlit
hemisphere, consistent with photo-ionization of an exo- [31] First, we investigate whether lunar phase may affect
the probability of observing pickup ions, by plotting the
sphericsource.Assumingthattheprobabilityofobservinga
distributionofthephase(thephiangleoftheMooninGSE
given source region scales with the total source volume of
coordinates)foralltheobservationsinFigure10.Wefindan
the cylindrical element represented in the 2-d plot, as it
increased number of observations before and after full
roughly should, Figure 9 (bottom) provides a less biased
Moon,andacompletedropoutaroundfullMoon;however,
map of the pickup ion source occurrence. This inferred
these trends also exist in the distribution of favorable con-
source map could plausibly correspond to photo-ionization
ditionsforpickupionobservationbyARTEMIS.Indeed,the
of an exospheric source with a scale height on the order of
(cid:2)1000 km, with the enhanced contribution from near the chance of observing pickup ions appears statistically uni-
form whennormalized by thenumberofopportunities. The
surfacepossiblyindicatingsmallerscaleheightorthedirect
9of13
E06006 HALEKASETAL.:LUNARPICKUPIONSOBSERVEDBYARTEMIS E06006
our purposes, an event is defined as a separate hour during
which ARTEMIS observed pickup ions. Some events, such
as those shown in Figures 1–2, have many observations,
while othershave only afew.
[33] Figure11showsnosignificantdifferencesinthesolar
wind flux for times when ARTEMIS observes pickup ions.
For Lyman a flux, we see a small trend toward increased
probability of observing pickup ions at times with higher
photon flux, but the trend has little statistical significance
given the small number of events. Apparently, enhanced
solar wind flux or solar photon flux do not significantly
increase the probability that ARTEMIS will observe ions.
This seems somewhat surprising, given that sputtering and
photo-ionization,thetwomainexpectedsourceprocessesfor
pickup ions, depend onthese two inputs. Thismay indicate
thatlunarpickupionsalwaysexistaroundtheMoonatsome
level,andthechanceofARTEMISobservingthemdepends
purelyongeometricalconsiderations.Figure5,whichshows
thatalloftheARTEMISobservationslieinanarrowspatial
region,providessomesupportforthishypothesis.
[34] We can also investigate whether the flux of pickup
ions(whenobserved)dependsoneitheroftheseparameters.
As above, we separate our observations into 29 events, and
calculate the average pickup ion differential energy flux
observedbytheESAinstrumentduringeachevent.Formost
observations, which lie in a single ESA energy bin, differ-
ential energy flux is directly proportional to total number
flux.Weshowtheresultsasafunctionofsolarwindproton
Figure 9. Favored source location corresponding to the andsolar photonflux, usingthesame data setsasabove,in
favored mass distribution of Figure 8 (left), in cylindrical Figure 12. We find that the pickup ion differential energy
coordinates.(top)Rawoccurrence;and(bottom)occurrence fluxcorrelateswithsolarwindflux,andanti-correlateswith
normalized by spatialvolume. photon flux.
[35] The significant correlation between pickup ion flux
increase in observation opportunities before and after full and solar wind flux clearly suggests that charged particle
Moonhasnoclearcause,butrepresentsonlyafewdaysand
may therefore prove completely random. The highest peak
in observations at phase (cid:4)150 corresponds to the 38 obser-
vationsfromP1andP2fromthesingleeventofFigures1–2,
whichoccurredduringaparticularlylongperiodoffavorable
observation conditions. Finally, the dropout around full
Moon corresponds to the period in which the Moon lies
insidetheEarth’smagnetosphere,withnoprevailingflowor
convection electric field strong enough to accelerate any
lunar ions that may exist into an energetic plume that our
eventselectionwouldhaveidentified.
[32] As a final exercise, we investigate whether solar
illumination or solar wind proton flux affect either the
probabilityofseeingpickupions,orthefluxofpickupions
seenbyARTEMIS.Wedonotconsiderathirdelementthat
couldcontrolpickupionfluxes,namelymicrometeoriteflux
[Smith et al., 1999]; however, we have noted no obvious
correlation between pickup ion observations and known
meteor streams. For solar wind proton flux, we use the Figure 10. Distribution of pickup ion observations for the
ARTEMIS ESA measurements. For solar illumination, we time period of September 15th, 2011 through February 5th,
use the TIMED / SEE Lyman a composite [Woods et al., 2012 as a function of lunar phase angle (defined as the phi
2000a, 2000b] as a proxy. In Figure 11, we show the dis- angle of the Moon in GSE coordinates, with +/(cid:4)180 = full
tribution of these two quantities for all data during the time moon, 0 = new moon), as compared to the distribution of
period from 9/15/2011–2/5/2012, for times with favorable favorable conditions (defined in the same way as Figure 5,
conditions for pickup ion observations (as previously asdescribedinthetext).Theredcurveshowsthepercentage
defined), and for 29 pickup ion events. To obtain the most oftimeswithfavorableconditionsduringwhichARTEMIS
unbiasedstatisticaldistribution,weopttoconsidergroupsof observed pickup ions, and the dashed red line shows the
observations that occur sequentially as a single event. For average percentage.
10of13