GEOPHYSICAL RESEARCH LETTERS, VOL. 18, NO. 11, PAGES 2169-2172, NOVEMBER 1991
LUNAR
SURFACE'
SPUTTERING
R.E.
Department of Nuclear
AND SECONDARY
ION
MASS
SPECTROMETRY
Johnson and R. Baragiola
Engineering
and Engineering
Physics University
of
Virginia
Abstract.
We combine laboratory and Apollo
observations to describe the sputtering of the
lunar
surface
year on the moon) every surface species has been
struck and subsequent ions eject chemically-
and the composition of the ejecta
altered
species.
At
high
fluences
sputtering
with special reference to O.
The atmospheric
inventory appears to be dominated by micro-
proceeds stoichiometrically
(Johnson 1990).
Three sputtering mechanismsoccur.
Electronic
meteorite
vaporization
of
lunar
grains.
Sputtering effects are observable in the local
plasma due to ion ejection,
in the extended
atmosphere through energetic neutral ejection,
sputtering (by ions, electrons, or EUV-photons)
results from excitation of electrons to repulsive
states and, for refractory-materials,
applies to
adsorbed species (H20, Na).
Sputtering by
and on grain
momentum-transfer is
fractionation
Ionization
surfaces through the
chemical
of the redeposited sputter-ejecta.
of
the
micrometeorite-vapor
solar
also
contributesto the local plasma.
wind
suggest
ions.
the
0
the dominant process for
[Morgan & Shemansky (1990)
yields
are
due
to
electronic
sputtering' the crosssections,'10-21 cm
2 are
too small.]
Finally,
chemical sputtering
arises
after
long-term
bombardment or when bombarding
with chemically reactive
species (H, O, C).
Introduction
Apollo data indicated
that solar-wind
ions
created amorphous outer layers
on lunar grains
'0.02•m,
changed these grains isotopically
and
The total yield for oxides by low energy (0.110 keV), light ions (H,He) are similar to metals
and peak near the energy-per-nucleon of the solar
chemicallyand the energetic-ionscreated tracks
wind' Ni peaksat 0.016 around1.2 keV for H+;
estimate regolith
that
and
spallation.
These effects were used to
turnover rates.
wind-ions reach the
occurs.
meteorite
surface,
for He+ , Y = 0.2 atoms/ion at 2 keV, close to
Because solar
sputtering
We examine this
here.
bombardment
modifies
Since
lunar
also
micrograins
Samples' Sputtering
-
-
and
The sputtering yield of a species (number of
ejected per incident ion) is roughly proportional
to
its
surface
concentration
and
process neutrals
ions
and
are more readily
chemical-binding
controls
the
oxides'
i.e.
the
formation
of
a
volatile
which is easily sputtered or desorbs thermally.
In a hydrogen glow-discharge [Ishibe & Oyama
inversely
1979]
In this
ejected
Since the solar
5% He++, the contribution is of the
maximum due to surface roughness at
55-85 ø . At
normal incidence
the ejecta
distribution
peaks
toward the
surface
normal,
a behavior
also
drastically
affected by roughness. A decrease in
the effective
yield occurs for a porous-regolith
due to redeposition
Fig.
1.
This can be
estimated
(Hapke 1986'
Johnson 1989)
from
information
on the ejected
species and angles,
and the sticking
probabilities
(Hapke & Cassidy
1978' Kasi & Rabalais 1990).
The sticking
of
the ejecta
results
in a net transport
into the
regolith.
Finally,
the energy distribution
of
collisionally
sputtered atoms peaks at
Ui/2.
Emission of H20 (chemical-sputtering)
has been
observed in long-term H-bombardment of glasses
and SIMS
proportional to its binding energy, Ui.
SiO2 seen in Fig.(1).
same order as for protons.
These yields
increase
with
increasing
angle
of
incidence
having
a
(impact-glasses
and agglutinates,
Taylor 1982)
and produces a vapor we compare it to sputtering.
Sputtering
by space plasmas and detection
of
ejected
ions can determine the composition of
surfaces by analogy with the laboratory technique
SIMS(Secondary
Ion
Mass
Spectrometry),
as
suggested for Saturn's moons (Johnson & Sittler
1990)
and for
the lunar
surface
(Stern
1991'
Elphic et al.,
this issue).
After describing the
sputtering process we consider its application to
the lunar surface and its importance relative
to
micrometeorite produced vapor, Fig. 1.
Laboratory
for
wind has
this
yield
decreases
with
dose from
0.01
H20/H to -0.003, typical of physical sputtering,
than
as the layer
e]ecta
physical
and
is reduced (depleted in O).
chemical
sputtering
lead
Both
to
composition (e.g. Johnson 1990).
The yield
dependsonly weakly on mass leading to slight
preferential ejection of the lighter species
preferential e•ection of oxygen[Thomas& Hofmann
1985] leaving microscopic metallic regions
(Walters et al 1989), a process possibly
In the laboratory, this dependenceis partially
due to projectile-induced surface structures.
Only a fraction of the ion•s energy goes into
sputtering, most of it
producin• chemical
alteration.
After a fluence of 101•ions/cm2(<l
molecular species form' for Na2S neutral S, S2,
Na, Na2S, NaS, Na2 were found after a large
fluence, in order of decreasing signal (Chrisey
et al 1988). Since S also comesfrom cracking of
(Kerridge& Kaplan1978). For a molecular-solid
the yield is fluence (exposure-time)dependent.
Copyright
1991by theAmerican
Geophysic•
Union.
occurringon the Moon(Taylor & Epstein 1973).
Since sputtering is a non-equilibrium process,
molecules in
the mass spectrometer, S2 was
estimated
to be the
Na20 in
a silicate
dominant
sulfur
product.
For
02 should be the dominant
speciesfor long term exposure,with the surface
slowly
Papernumber91GL02095
replenished
in
diffusion from the bulk.
0094-8534/91/91GL-02095503.00
O
increase the sputtering yield
2169
by
beam-enhanced
Surface-charging can
[Bach et al 1974]
2170
Johnsonet al.'
Lunar Surface'
Sputtering andSecondaryIon MassSpectrometry
effective
temperature,
~11,000K (see also Elphic
et al this issue)].
Ion fractions
can drop when
the surface oxide is reduced by sputtering
due to
neutralization
from 'metal-islands'
(Baragiola et
al 1984).
The ejected-ions
are mostly atomic
with average energies
ten eV.
Molecular-ions
.
have low kinetic
energies but the small
are clearly
resolvable
[Elphic et al. this
for
H20 absorbed on a silicate,
yields
issue;
SiOH
+ is
observed, Nagai & Shimizu 1984].
Negative
ions
are also produced, the main one being O'.
Since
ion yields
are affected
by surface-charging,
usually
positive,
positive
ion
emission
is
favored [Baragtola
et al 1990], the case on the
sunlit
face
of
the
moon.
Lunar Sample Sputtering
and the Atmosphere
Effective
yields obtained from the analysis of
lunar
samples
indicates
that
the amount of
material
ejected from a grain is of the order of
the laboratory
sputtering
yield,
allowing
for
microroughness and averaging over incident angle
~O.1-0.2A/yr,
therefore,
the grain is altered by
Fig. 1. Diagramof sputteringof regolith: much this amount. This ejecta impactsneighborswith
of vapor is directed inward and competes
with
sticking probability dependingon the species,
micrometeorite
vaporization.
producinga fractionated-depostt. The direct
ejecta is of interest to the atmosphere, ~O.01O.03A/yr consistent with estimates of McDonnell
and can alter the near-surface compositionby
field-enhanced diffusion of oxygen and ionic
impurities (alkalis in glasses, Pederson1982)
A fraction of the sputtered flux is ions, the
(1977).
Someof this returns gravitationally.
In addition, micrometorite-vapor deposition
reduces the net grain erosion.
Zinner et al
(1977) and Kerridge (1990) place limits of
in
suggesting
basis for SIMSanalysis. Thereis a large spread
the data
for
the ion fractions
from oxides.
~O.002-0.003A/yr,well below the values above,
that
micrometeorite
erosion
is
the
The values for Mg, A1, Si and Ti oxides lie
dominantvapor-productionprocess obscuring the
For impurities the fraction dependsstrongly on
their ionization potential implyinglarge values
The loss of oxygen controls sputter aging of
oxides, related to the relative importance
of O,
between 0.01 and 0.1 for keV Ar ion bombardment.
for alkalis, [• ci exp(-Ii/kT), I i theenergy
of speciesi with concentration
ci; T is an
10
'
sputter
effect.
02, O' or O+ in theejecta. Low-energy
O' may
be
returned to the positively chargedsurface
inhibiting
sputtering
(Hecht et al
1989) or
ballistic
O impacting
the nightside
may be
converted
to
O'
and swept away (Morgan &
Shemansky 1990).
However, directly
ejected
O
atoms
are
predominantly
energetic
escape,
(Ui/2
contrary
>
leV)
to the
of Morgan and Shemansky(1990).
and
suggestion
Since 02 ejection
involves
lower energies
it
can be in thermalequilibrium
with the regolith.
The efficiency
of
adsorption
of returning
O is uncertain
(Hapke
1986; Sprague, A.L. private
communication) but it
will
likely
react
with
the
oxygen-depleted
10 t
surface.
Table
1 Composition
of Grain
lO 0
Stoichiometry(1)
10'1
10';
10'•
10 0
ENERGY
10 •
10 •
(k e.V)
Fig. 2.
Sputtering yields'
equivalent SiO2 per
ion vs ion energy (Betz and Wehner 1983).
Decomposition(2)
Atmospheric
Composition(3)
0
Si
A1
0.60
0.17
0.11
Na
K
Fe
1.0
0.025
0.025
Na 1.0
K 0.25
Fe 0.1
Ca 1.0
Ca
0.062
02
0.001
Fe
0.020
SiO
0.001
Mg
0.037
Mg 0.0006
Mg 4.0
Na
0.0033
Mn
0.0004
A1
4.0
K
0.00035
0
?
0
?
(1) Morgan & Shemansky (1990).
(1971),
amount
of
interactions.
(3)
(1986)
formulation
ballistic
lifetimes,
first
column.
O
uncertain
Si
6.0
(2)De Maria et al
due
to
wall
Relative
to Na, using Hapke
of
regoltth
sticking,
and ionization
loss
with
Johnson et
al.:
Lunar Surface:
Sputtering
Because of the dependence of sputtering
and
and Secondary Ion Mass Spectrometry
determined by the local
fields.
2171
Ions are also
sticking
onUi, the
ejecta
bedominated
a
volatile
species
even
if its can
concentration
isby
low
formed
by photoionization,
the('10-2-10
probability
for
a ballistic
neutral is but
small
- ,
with the
volatiles
depending
sputtered
bombardment enhancing
to
the
surface.
the diffusion
Therefore,
of
the
on the species).
neutrals
are not
Therefore,
ions from
important
[Funsten et
effective
concentration of species changes down
to the depth of penetration of the ions ~100A.
If other regolith alteration/turn-over
processes
al.
this
issue].
Because of
its
larger
contribution to the neutral vapor, ionization of
the meteorite-produced vapor may be an important
are slow, the surface concentration
of the most
volatile
species (Na, K, Fe) will
decrease until
sputtering
becomes proportional
to
the
bulk
component of the local plasma.
This vapor also
permeates
the regolith,
so there
are similar
uncertainties
in determining
its contribution
to
atomic concentration
(Johnson 1990).
The time
for
attaining
stoichiometric
sputtering
determined by the fluence and the yield of
is
the
the atmosphere.
with the
source is
Assuminga net source ~0.1A/yr,
ionization
comparable
probabilities
above, this
to the directly
sputtered
hardest to-sputter species' -104 years on the
ions.
Moon.
affected differently
by the fields,
sputtered-ions
the
This occurs in competition with micro-
meteorite
agitation
and vaporization
(Morgan et al 1989).
Regolith
of
agitation
grains
leads to
Since ions of different energies are
and
distinguished
ionized
(Manka et al this
the directlyvapor
issue).
can
be
Using the
exposureages for lunar grains of -104 - 105
mass spectra observedby a lunar orbiter with
years, comparable to or longer than the time for
sputter
aging.
However, if
micrometeorite
laboratory
data,
concentration
of
vaporization
stoichiometric
sputtering
is the dominant source of vapor then
sputtering
is not achieved.
In
Table 1, the decomposition-products
from a grain
heated
to 1500K are compared to stoichiometricloss.
Although
a non-equilibrium
process,
sputtering
of neutrals
is often characterized
by
a map can be made of the
surface species produced by
or meteorite
impact.
Conclusions
an energy per atom of this order.
Therefore,
the
neutral
ejecta
from a fresh
grain
should be
similar
to the decomposition
data.
Since
the regolith
is an extension
of the
atmosphere,
sticking/reaction
coefficients
control
the atmospheric composition
(Hapke 1986;
Hunten et al 1988).
That is, species that weakly
adsorb to the surfaces
are also easily
ejected
[by ions,
electrons,
photons]
and may desorb
thermally (McGrath et al. 1986).
Therefore, even
for a material
with low Na concentration,
Na can
That solar-wind
ions impact lunar grains
is
certain
[implantation
in and amorphous layers on
grains,
Taylor 1982].
Therefore,
sputtering
of
lunar grains must occur.
That micrometeorites
impact grains
[impact-glasses
and agglutinates,
Taylor 1982] and thereby produces a vapor is also
certain
(Morgan et al
1989).
Micrometeorite
impact also mixes the regolith
and the vapor
redeposited
appears to be at least comparable to
sputter
erosion (Kerridge
& Kaplan 1978) and is,
probably,
the dominant source of lunar vapor.
The sputter-produced
vapor contains a larger
proportion
of atomic species and ions and we give
dominate the vapor (Hapke & Cassidy 1978).
In
Table 1 atmospheric-constituents
are also given
relative
to Na calculated including adsorption in
the regolith.
Although Fe sputters easily,
it is
a minor
component because
it
sticks
well
[possibly accounting for Fe deposits on grains,
Hapke 1986].
Reactive
ion implantation
can
affect
atmospheric composition'
implanted C can
produce CO, although there is no evidence for
loss of implanted C (Kerridge & Kaplan 1978).
the structure of the neutral component.
In sputtering
of an oxide loss of oxygen plays
a special role.
Therefore the low value observed
for
atmospheric
atomic oxygen is
at
first
puzzling
(Morgan & Shemansky 1990).
Whereas
photo-sputtering
may apply to certain
species,
for
oxygen momentum transfer
and chemical
sputtering
are the dominant solar-wind-ejection
processes.
Oxygen is ejected as O, 02 or in an
ionized
form, being sensitive
to the surface
Implanted H can produce H2 and H20 (Hodges this
chemistry
issue)
oxygen
and
Therefore,
can
replace
Na
in
the
lattice.
the atmospheric concentration of H(H2)
is much smaller than expected.
The atmosphere has two spatial
components;
a
and charge-state.
is
ejected
escape fraction
Further,
energetically
and returning
the irradiation-altered
The total
sputtering
surface.
yields
with
atomic
a
large
0 can react with
are
significant
small component of energetic neutrals with a
larger partially
or fully thermally accommodated
component having
a Maxwelljan-like
energy
spectrum, as predicted by McGrath et al (1986).
~0.1-0.2A/yr.
A large fraction of this sputtered
flux sticks to neighboring grains producing an
effective yield -0.01-0.03A/yr.
A small fraction
also returns to the surface gravitationally.
There is also a very small flux of backscattered,
Therefore, in discussing solar-wind sputtering,
neut=alized protons(H) with much larger energies.
the yield
The directly
needs to be stated.
sputtered
species
can have large
escape fractions dependingon Ui
relevant
to a particular
For instance,
measurement
for chemical-
[-90% for O,
fractionation of deposits the relevant yield is
Johnson 1990]. Ignoring escape for species with
low Ui, the sputtered neutral atmosphere decays
slowly for altitudes
much less than a lunar
~0.1-0.2A/yr,
whereas the effective erosion rate
of a grain is much smaller. Whendescribing the
the mass, g gravity].
0.03A/yr.
radius[n(z)=(N/2Hc)(l+Z/Hc)-3/2,
Hc=(Ui/Mig),Mi
N,
produced
The corresponding column,
bM an
incident
N=<Y>•[•(2Ui/Mi
)1/2]
ion
flux
(Johnson1990).
•
is
This
component has a large scale height 2Hc/3 , whereas
the accommodated component has a much smaller
scale height
determined
by surface
temperature,
z (kT/Mig), n(z)=n(o)exp[-z/H].
The
sputtered
ions
follow
H
direct-sputter
contribution
to
fraction
of
The directly
ejected
this
comparable in
and are
lunar
ions are
a
source
strength to ionized meteorite-produced
vapor.
Comparing the time for sputter depletion to the
regolith-turnover
rate a variety of sputter-ages
are
possible
and
are
found.
micrometeorite-produced lunar
trajectories
the
atmosphere
the appropriateyield is -0.01-
grains
(Kerridge
1990)
However,
if
vapor coats the
stoichiometric
sputtering
2172
Johnson et al.'
is not attained
Therefore
trace
the atmosphere.
Lunar Surface'
over most of the
volatile
species
lunar
will
Sputtering
surface.
dominate
and Secondary Ion Mass Spectrometry
Kerridge,
J.F.,
sputtering
of
Planetary
Sci.
Houston,
Acknowledgment
The authors thank A. Sprague
and J. Kerridge for comments. Work supported by
NASA Geology/Geophysics
Program; NASA Magnetospheric-Physics
Program;
NSF Astronomy Division.
I.
Kitzmann,
H.
Schroeder,
Sputtering-
Yields and Energy-Lossesof Ar+ Ions, 5 to 30
keV in SiO2 Rad. Effects21,31,
1974.
Baragiola,
R.A., J. Ferron, G. Zampieri, Effect
of Oxygen on the Secondary Ion Emission from
A1. Nucl.
Instr.
Meth. B2, 614,
Baragiola,
R.A.,
T.E.
Electron-Stimulated
1984.
Madey, A.M. Lanzillotto
Desorption of Multiply-
Charged Ions from SiO2 Phys. Rev. B. 41,9541,
1990.
Betz
G.
and
multicomponent
Particle
G.K.
Wehner,
Sputtering
materials.
In:
Sputtering
Bombardment
II
ed.
R.
of
by
Behrisch,
G.,
G.
Balducci,
M.
Guido,
V.
1990.
J.R.
relationship
lunar
Lunar
to
surface.
Planet.
rate
Kaplan,
isotopic
Sputtering:
Its
fractionation
on the
Proc.Lunar.
Planet. Sci.Conf
9,
Inst.,
Houston 1687-1709,
1978.
defined.
Planet.
Piacente,
Proc.
Inst.
Lunar
Houston,
Sci
Conf.
3835-3857,
8 Lunar
&
1977.
McGrath, M.A., R.E. Johnson & L.J. Lanzerotti,
Sputtering
of Sodium on the planet Mercury
Nature 323, 694-696, 1986.
Morgan, T.H., D.E. Shemansky, Limits to the lunar
atmosphere J. Geophys. Res. 96 1351, 1990.
Morgan, T.H., H.A. Zook, A.E. Potter,
Production
of sodium vapor from exposed regolith
in the
inner
solar
system
Proc.
19th
Lunar
and
Planet.
Sci.
Houston,
Springer,
Berlin,
pp 11-90, 1983.
Chrisey,
D.B.,
R.E. Johnson, J.W. Boring,
J.A.
Phipps, Ejection
of Na from Sodium Sulfide by
sputtering
of Io.
Icarus 75 233, 1988.
DeMaria
301,
J.F. ,
McDonnell,
J.A.M.,
Accretionary
particle
studies
on Apollo
12054-58:
In-situ
lunar
surface
microparticle
rate
and solar wind sputtering
References
Bach H.,
Kerridge,
Re-evaluation
of
solar-wind
the lunar
surface.
Lunar &
Conf. 21 Lunar Planet.
Inst.,
Nagai,
Conf.,
297-304,
S.,
Y.
Shimzu,
impact effects
Mats 128/129,
Pederson,
L.R.,
on
glass
Lunar
and
Planet.
Inst.,
1989.
and keV
ion
on SiO2, A1203, MgO J.
Electron
Nucl.
605-608,
1984.
Are neutral
sodium atoms produced
surfaces
by
election
bombardment
Mass
spectrometric
investigation
of
the
vaporization
processof
Apollo
12
lunar
samples.
Proc. 2nd Lunar Sci. Conf. (Lunar &
Stern,
A. "Lunar Atmosphere Report"
to Lunar
Explor. Science Working Group (unpublished)
Planet. INST. Houston),1367-1380,1971.
Taylor, H.P.,
Hapke, B.,
regoliths
On the sputter
alterations
of
of the outer solar system bodies.
Icarus 56 426-438, 1986.
Hapke, B., W.A. Cassidy, Is
smooth
of
Res.
Hecht,
as
a billiard
sputtering
of
the moon really
ball?
the
Remarks
lunar
surface
Lett • 297-300,
1978.
M.H.,
O.J.
Orient,
A.
Silicon
Vasquez, Oxidation
of
Beam App. Phys Lett 54 422-423,
Hunten, D.M., T. Morgan, and D.
as
on models
Geophys.
Chutjean,
R.P.
with a 5eV O-
1989.
Shemansky, The
C.R. Chapman
& M.S. Matthews, (Univ. of
Arizona Press, Tucson) pp 562-612, 1988.
Ishibe, Y., H. Oyama, Reduction of Metal Oxide
Layer with HydrogenAtomsfrom a Hot Filament,
J. Nucl. Mater. 85/86, 1191, 1979.
Johnson,R.E., Sputtering of a planetary regolith
Icarus 78, 206-210, 1989.
with
119,
L307-L313,
1982.
S. Epstein, O18/O1•, Si30/Si28
Houston, 1973.
Taylor,
S.R.
Perspective.
Planetary
Lunar
Science:
Planet.
Inst.,
Lunar
Tucson,
1982.
J.H.,
S. Hofmann,
Ion-Bombardment
Induced
Changes
in
Sillcoin-Dioxide
Surface
Composition
by
x-Ray
Photoelectron
Spectroscopy
J. Vac. Sci.
Technol.
AJ, 1921,
1985.
Walters, M. D., J.W. Boring, R.E. Johnson, W.A.
Jesser. Crystallization of
SiO by He+
bombardment. Radia Effs. 106, 189-201, 1989.
Zinner, E. R.M. Walker J. Channat and J.C. Dran,
Ion probe surface concentration measurements
of Mg and Fe and microcraters in crystals from
lunar rocks and soil Lunar.Sci.Conf.5th, Lunar
Planet. Inst., Houston, 3859-3883, 1977.
Energetic
Charged-particle
Atmospheres
and Surfaces
(Springer-Verlag, Berlin), 1990.
Johnson, R.E. and E.C. Sittler, Sputter-produced
plasma as a measure of satellite
surface
composition' The Cassini mission G.R.L. 17
1629-1632,
1990.
Kasi,
S.R.,
Rabalais,
J.W.,
The Chemical Physics
of Low-Energy Ion Beam-Surface InteractionsThe Panorama of Phenomena Involved
Rad. Effects
Sci
enrichment studies of lunar samples Lunar
Planet. Sci. Conf. 4th, Lunar Planet. Inst.,
Thomas,
Mercury atmosphere, in Mercury, eds. FaVilas,
Johnson,
R.E.,
Interactions
Surf.
Def. Sol.112,119,1990.
R. E. Johnsonand R. A. Baragiola, Department
of Nuclear Engineering and Engineering Physics,
University of Virginia, Charlottesville, Virginia
22903.
(Received
June 10,
1991;
accepted August 2, 1991.)