GEOPHYSICAL RESEARCH LETTERS, VOL. ???, XXXX, DOI:10.1029/,
X-rays from Solar Wind Charge Exchange at Mars:
A Comparison of Simulations and Observations
H. Gunell and M. Holmstr¨ om
Swedish Institute of Space Physics, Kiruna, Sweden
E. Kallio and P. Janhunen
Finnish Meteorological Institute, Helsinki, Finland
Max-Planck-Institut f¨ ur extraterrestrische Physik, Garching, Germany
A hybrid simulation of the solar wind-Mars interaction
and a test particle simulation of heavy ion trajectories near
Mars are used to compute the contribution from solar wind
charge exchange processes to the X-ray emission from Mars.
It is found that the X-ray halo observed by the Chandra X-
ray observatory can be explained by emissions from heavy,
highly charged, ions in the solar wind undergoing charge
exchange collisions in the upper atmosphere of Mars.
There are several possible mechanisms for X-ray genera-
tion in the atmosphere and exosphere of Mars. Three of the
proposed mechanisms are fluorescence and scattering of so-
lar X-rays; electron bremsstrahlung and line emissions; and
solar wind charge-exchange.
Fluorescence by solar X-rays occur when photons are ab-
sorbed by atmospheric neutrals and then re-emitted isotrop-
ically. Solar X-rays can also be elastically scattered by at-
mospheric neutrals. A detailed, quantitative, analysis of the
intensity of the X-rays produced by these two mechanisms
has been done by Cravens and Maurellis 
Bremsstrahlung and line emissions result when neutrals
that are photo-ionized interact with the solar wind plasma
and excite waves, that in turn can accelerate electrons to
keV energies. When such electrons collide with neutrals,
X-rays can be produced by bremsstrahlung and K-shell ra-
diation. This mechanism has been proposed for cometary
X-rays [Bingham et al., 1997; Shapiro et al., 1999], but could
also take place at Mars.
The third mechanism is solar wind charge exchange
(SWCX). Wherever the solar wind meets a neutral atmo-
sphere, X-rays are emitted by a charge exchange process
between the neutrals and heavy solar wind ions. A small
fraction of the solar wind consists of heavy, multiply charged
ions such as O6+, C6+and Ne8+. Charge-exchange between
such an ion and a neutral atom can leave the ion in an ex-
cited state. When the captured electron then transits to
a lower energy state, within the L- and K-shells, X-rays
may be emitted. This source of X-rays was first proposed
by Cravens  as an explanation of X-rays observed from
comets, and probably this process is the dominant source of
the soft X-ray emissions from comets [Cravens, 2002]. It was
then suggested by Cravens  and Krasnopolsky 
Copyright 2004 by the American Geophysical Union.
that this should be a source of X-ray emissions also at Mars.
Computer simulations of the intensities and morphology of
these emissions was presented by Holmstr¨ om et al.  for
In recent years there have been observations of planetary
X-ray emissions. X-rays from Venus [Dennerl et al., 2002]
and Mars [Dennerl, 2002] were discovered using the Chan-
dra X-ray observatory.
When Mars was observed in 2001, Dennerl  found
that the X-ray emission from a disk the size of Mars was
dominated by fluorescent scattering of solar X-rays. A faint
X-ray halo surrounding the disk was also detected. The X-
ray spectrum of the halo differed from that of the disk, and
cannot be explained by fluorescence. The fluorescence peak
at 0.65 keV, that is seen in the emissions from the disk, is
absent in the X-ray emissions from the halo, c.f., figure 4 of
[Dennerl, 2002]. Fluorescence is an efficient process only at
low altitudes. Dennerl  suggested that the halo could
be caused by the SWCX process. In this work we perform
a computer simulation to enquire whether charge exchange
processes can explain the Martian X-ray halo.
In the future, imaging of SWCX could provide informa-
tion on exospheric densities, solar wind composition and so-
lar wind conditions. Since the imaging is global, we can
get instantanous information on the conditions in the whole
near Mars environment. To extract the information from
images of the X-ray flux, detailed modeling of the produc-
tion process will be needed, as discussed in [Holmstr¨ om and
2. Simulation Model
Here we present the method by which we have produced
simulated images of SWCX at Mars that corresponds to the
observation by Chandra in 2001.
The calculations were performed in three steps. First the
solar wind parameters were estimated from data obtained
by the WIND spacecraft. Since Mars was near opposition
the plasma that was sampled by WIND near the earth on
July 2, 2001 arrived at Mars two days later during the X-
ray observation. The data was scaled with the distance from
the sun, and the average parameter values over the period
of the observation were used as input parameters for a hy-
brid simulation. These parameters are: vsw = 330 km/s;
nsw = 4.40 × 106m−3; Tp,sw = 4.56 × 104K; and?B =
(1.13,−3.63,0)nT = 3.80nT · (cos(72.7◦),−sin(72.7◦),0).
The parameters are estimated in the following way. The
arrival time at Mars of a particular plasma element as a func-
tion of the time it passed by the earth was calculated from
the vxdata. This gives an interval, starting at 00:57:49 (UT)
and ending 18:49:49 (UT) on July 2, over which the plasma
X - 2
GUNELL ET AL: CHARGE EXCHANGE X-RAYS AT MARS
parameters are averaged. The plasma that passed the earth
during this interval reach Mars during the observation two
days later. We use a Mars-centred coordinate system with
the sun in the positive x-direction, a northernly z-axis that
is perpendicular to the ecliptic, and a y-direction that closes
the right-handed system. The solar wind speed vsw and the
proton temperature Tp,sw are assumed to be constants, and
are simply the average values of vx and Tp. The density
nsw and the magnitude of the magnetic field |B| are scaled
by the squared ratio of the two planets respective distances
from the sun: nsw = np/1.4462and Bsw = B/1.4462. The
z-component of the magnetic field is smaller than the other
two for most of the observation, and therefore Bz = 0 is
assumed. We have assumed that the the direction of the
magnetic field is given by a a Parker-spiral, which means
that By/Bx is proportional to the distance from the sun.
The second step was running a hybrid simulation of the
interaction between the solar wind and Mars [Kallio and
Janhunen, 2001, 2002] to obtain the electric and magnetic
fields around Mars at the time of the observation. The hy-
brid code was run using 1.7 × 106particles, with a fully
absorbing obstacle boundary at Robst= 3600 km. The sim-
ulation box was a Mars-centred 6Robst × 6Robst × 6Robst
cube. A grid with three different cell sizes, namely 0.05Robst,
0.1Robst, and 0.2Robst, was used. The smallest cells were
used close to the planet on the dayside.
As a third step a test particle simulation was run, calcu-
lating the trajectories of heavy solar wind ions in the elec-
tric and magnetic fields that were obtained from the hybrid
simulation, and for each time step of the test particle sim-
ulation saving the X-ray emission density on a grid. The
trajectories were compued by integrating the Lorentz force
using Boris’ CYLRAD algorithm (see for example Hockney
and Eastwood ) One hundred thousand trajectories
were calculated for each of the ion species O7+, C6+, O6+,
O8+, Mg10+, Mg9+, Si9+, N6+, C5+, Ne8+, Fe9+, S9+, Si8+,
Fe11+, and Mg8+. These species were selected from table 1
of Schwadron and Cravens  for being the ion species
that generate the highest luminosity, and together compris-
ing 91.6 % of the total luminosity of the species of in that
table. The test particles were injected in the simulation box
of the hybrid simulation at its upstream edge. The initial
velocity distribution was assumed to be a Maxwellian with
temperature Tp,sw that is centred around the drift speed
vsw, i.e., we assume that all species have the same tempera-
ture and that that temperature is equal to the temperature
of the solar wind protons.
The cross sections for charge exchange (σ), the relative
abundance of each ion species in the solar wind (a), and
the energy released (En→n0) were taken from table 1 of
Schwadron and Cravens . The cross sections reported
by Schwadron and Cravens  are for charge exchange
collisions between heavy ions and water molecules, but since
the ionisation potential for O, H, and H2 are close to that
of water using the cross sections for water is a reasonable
approximation [Wegmann et al., 1998]. An excited ion may
go through several intermediate states before reaching the
ground state, and thus release its energy to several pho-
tons. We use a simplified model, assuming that the transi-
tion to the ground state occurs in either one or two steps
[Schwadron and Cravens, 2000]. Neutral exosphere densities
for atomic oxygen and hydrogen and molecular hydrogen
for solar maximum conditions were used [Kallio et al., 1997;
Krasnopolsky and Gladstone, 1996]. The densities are mod-
elled by a Chamberlain exosphere [Chamberlain and Hunten,
1987], which is a spherically symmetric model. The exobase
altitude is 170 km, and the exobase densities are nH =
3.1×1010m−3for atomic hydrogen, nH2= 4.3×1011m−3for
molecular hydrogen, and nO = 3.3×1014m−3for atomic oxy-
gen. The exobase conditions (temperature and density) has
large temporal and spatial variation [Keating et al., 1998],
resulting in large variations of exospheric densities.
Each test particle represents a large number of solar wind
ions, and this number decreases as the test particle moves
through the simulation area, where some of the real parti-
cles it represents are lost in charge exchange collisions. The
density ni represented by a test particle is
ni(s) = nswae−σ?s
where nswis the solar wind proton density, s the path length
along the trajectory, Ni(s) the density of the neutral exo-
sphere, and a is the relative abundance of the particular ion
species in the solar wind, i.e., a = ni(0)/nsw. The contri-
bution from one test particle to the density of the emitted
X-ray power is then
Pi(s) = −dni
For each test particle at each time step this is accumulated
on a grid, in a way that is analogous with the way charge is
assigned to a grid in a particle in cell simulation. The total
emitted X-ray power density is then found by summation of
the contributions from all ion species.
offset along solar direction [’’]
offset perp. to solar direction [’’]
Figure 1. A simulated X-ray image of Mars at a phase
angle of 18.2◦, corresponding to the situation at the time
of the observation. The inner white circle, with radius
10.2 arc seconds, marks the geometric size of Mars, and
the outer white circle, with radius 30 arc seconds marks
the extent of the X-ray halo according to Dennerl .
The grey-scale shows the X-ray radiance in Wm−2sr−1.
GUNELL ET AL: CHARGE EXCHANGE X-RAYS AT MARS
X - 3
Ions can, in principle, undergo several charge exchange
collisions until they have lost all their charge, emitting pho-
tons in each step. This means that O7+ions after a charge
exchange collisions are a source of O6+ions. We have ne-
glected this source term after examining the exponential
by precipitating on the atmosphere or by leaving the outer
boundary of the simulation box, is a measure of what frac-
tion of the original density that still remains. If the value of
the exponential still is close to unity when the particles leave
the system the traversed region of the atmosphere can be
considered collisionally thin and the additional source term
is negligible.The average value of this factor computed
using ten thousand test particles of the most important
species, i.e., O7+and C6+, is 0.85 and 0.92 respectively. The
worst case would be the species with the highest cross sec-
tion, which in our case is Fe11+for which σ = 1.5×10−18m2.
For Fe11+the average value of the exponential is 0.82 for par-
ticles leaving the system. We thus find it justified to neglect
this small ion source.
0Ni(s?)ds?) of Eq. (1). The value of the
exponential when a test particle leaves the system, either
background noise level
projected distance from Mars’ centre [’’]
Spatial distribution [counts sr−1s−1]
Figure 2. Radial distribution of the observed X-ray in-
tensity in units of sr−1s−1. The thin solid curve shows the
results of the simulation. The horizontal thick lines are a
histogram of the observed X-ray photons. The horizontal
thin lines show the average X-ray intensity in the same
intervals as the histogram of the observation. The error
bars mark a 90% confidence interval of the simulated re-
sults, assuming a Poisson distribution for the count rate.
The vertical dashed lines mark the size of Mars and the
extent of the X-ray halo according to Dennerl , i.e.,
these lines correspond to the circles of Fig. 1.
The flux that can be observed by an observer at earth
is found by integration along lines of sight.
observer is very far away we use a parallel projection.
p(ξ,ζ) = 1/(4π)?P(x,y,z)ds is the directional X-ray en-
ξ corresponds to the offset along the solar direction and ζ
to the offset perpendicular to the solar direction. Fig. 1
shows the directional X-ray energy flux from Mars. The in-
ner white circle, with radius 10.2 arc seconds, marks the ge-
ometric size of Mars, and the outer white circle, with radius
30 arc seconds marks the extent of the X-ray halo according
to Dennerl . The grey-scale shows the X-ray radiance
in Wm−2sr−1. In Fig. 1 we have included only photons with
energies above 200 eV. The photon energy is modelled, fol-
lowing Schwadron and Cravens , under the assumption
that an ion in its excited state (quantum number n) emits
its energy, En→n0, in either one or two steps, and that all
transitions have an equal probability given by 1/(n − n0).
The quantity shown in Fig. 1 is thus
ergy flux measured in watts per square metre and steradian.
n − n0ds
where S is the cross section of the simulation box, Eph is
the photon energy, and E = 200 eV is the minimum energy
of the photons included in Fig. 1. For E = 0, that is, if all
photons are included, pE(ξ,ζ) = p(ξ,ζ).
To be able to compare with observations the energy de-
pendent effective detector area A(Eph) has to be taken into
account. The observed directional flux expressed as a count
n − n0ds
The radial distribution of the X-ray power is shown in
Fig. 2, where the thin solid curve displays the quantity
where ρ is the radial and ϕ is the azimuthal polar coordi-
nate of Fig. 1. The thick horizontal lines in Fig. 2 are a
histogram of the observed X-ray photons. To facilitate a
comparison between the simulation and the observation av-
erages of the simulated curve, obtained from Eq. (3), over
the same radial bins as those used for the observed photons,
are shown as thin horizontal lines. The error bars show the
limits of a 90% confidence interval of what would have been
observed from our simulation, assuming a Poisson distribu-
tion for the count rate. The probability of observing a count
rate smaller than the lower limit is 5% and that of observing
a count rate higher than the upper limit of the error bar is
The luminosity of the halo reported by Dennerl  in
the energy range E = 0.5−1.2 keV is 0.5±0.2 MW. In this
energy range the present simulation yields a luminosity of
1.8 MW. This figure is obtained by integrating all emissions
within the simulation box.
We have reported simulations of X-ray emissions from
Mars caused by the solar wind charge exchange process.
Previous observations of Martian X-rays show a high count
X - 4
GUNELL ET AL: CHARGE EXCHANGE X-RAYS AT MARS
rate of photons from a disk of the size of the planet and
lower count rates from an X-ray halo surrounding it. The
emissions from the disk were interpreted as fluorescent scat-
tering of solar X-rays. The photon flux calculated here is
substantially lower than that observed from the disk (Fig.
2), which is what should be expected since we only calculate
the charge exchange contribution. For the halo the calcu-
lated count rates are higher than those that were observed
by a factor between one and three.
There are a few uncertainties of the model that could
contribute to this discrepancy. First, the density of the neu-
tral exosphere of Mars changes with the solar cycle, and the
estimate we used for solar maximum conditions might be
incorrect for the particular time of observation. Secondly,
the estimate of the solar wind parameters might be inaccu-
rate, since it was based on measurements near Earth and
relies on the solar wind conditions being unchanged over
an angle of 6◦. Thirdly, the composition of the solar wind,
i.e. the abundances of the heavy ions, may, on the day of
the observation, have differed from the estimates we used.
Fourthly the fields obtained from the hybrid simulation may
be inaccurate due to the limited spatial resolution. The hy-
brid model reproduces the basic plasma and magnetic field
regions around Mars that were measured by the Phobos-2
mission, for example the bow shock, magnetotail, and how
the magnetic field is piled up against the planet [Kallio and
Janhunen, 2001, 2002]. Fifthly there could be errors in the
values that we have used for the charge exchange cross sec-
tions. The classical over-barrier cross sections this we have
used are generally over-estimates, and that can contribute
to the discrepancy between the between the measured and
simulated halo emissions.
In addition to the overall discrepancy it can be seen in
Fig. 2 that the discrepancy between the simulations and the
observations is the largest for the second bin of the halo.
This can indicate that the radial dependence of the neu-
tral exosphere model deviates from the actual exospheric
density on the day of the observations. In this way X-ray
observations can provide a means of studying the radial de-
pendence of the exosphere density. In a forthcoming paper
[Gunell et al., 2004] we will investigate what influence fluc-
tuations in the different parameters have on the observed
These simulations show that the contribution from the
solar wind charge exchange process to the X-ray emissions
from the halo is large enough to explain the observed X-ray
We are grateful to Ronald P. Lepping (NASA/Goddard
SFC) for the WIND MFI data, and to K.W. Ogilvie
(NASA/GSFC), A.J. Lazarus (MIT), and J.C. Kasper
(MIT) for the WIND SWE data. This work was supported
by the Swedish National Space Board.
Bingham, R. J., M. Dawson, V. D. Shapiro, D. A. Mendis, and
B. J. Kellet, Generation of X-rays from comet C/Hyakutake
1996 B2, Science, 275, 49–51, 1997.
Chamberlain, J. W., D. M. Hunten, Theory of planetary atmo-
spheres, 2nd Edition. Academic Press, inc., San Diego, Cali-
Cravens, T. E., Comet Hyakutake x-ray source: Charge transfer
and solar wind heavy ions, Geophys. Res. Lett., 24, 105–108,
Cravens, T. E., X-ray emission from comets and planets, Adv.
Space Res., 26(10), 1443–1451, 2000.
Cravens, T. E., X-ray emission from comets, Science, 296, 1042–
Cravens, T. E., and A. N. Maurellis, X-ray emission from scatter-
ing and fluorescence of solar x-rays at venus and mars, Geo-
phys. Res. Lett., 28(15), 3043–3046, 2001.
Dennerl, K., Discovery of X-rays from Mars with Chandra, As-
tronomy & Astrophysics, 394, 1119–1128, doi:10.1051/0004-
Dennerl, K., V. Burwitz, J. Englhauser, C. Lisse, and S. Wolk,
Discovery of X-rays from Venus with Chandra, Astronomy &
Astrophysics, 386, 319–330, doi:10.1051/0004-6361:20020097,
Gunell, H., M. Holmstr¨ om, E. Kallio, P. Janhunen, and K. Den-
nerl, X-rays from solar wind charge exchange at Mars, Adv.
Space Res., in preparation, 2004.
Hockney, R. W., and J. W. Eastwood, Computer Simulation Us-
ing Particles, IOP Publishing Ltd, Bristol, 1988.
Holmstr¨ om, M., and E. Kallio, The solar wind interaction with
Venus and Mars: Energetic neutral atom and X-ray imaging,
Adv. Space Res., 33(2), 187–193, 2004.
Holmstr¨ om, M., S. Barabash, and E. Kallio, X-ray imaging of
the solar wind–Mars interaction, Geophys. Res. Lett., 28(7),
Kallio, E., and P. Janhunen, Atmospheric effects of proton pre-
cipitation in the Martian atmosphere and its connection to the
Mars-solar wind interaction, J. Geophys. Res., 106, 5617–5634,
Kallio, E., and P. Janhunen, Ion escape from Mars in a quasi-
neutral hybrid model, J. Geophys. Res., 107(A3), 1035, doi:
Kallio, E., J. G. Luhmann, and S. Barabash, Charge exchange
near Mars: The solar wind absorption and energetic neutral
atom production, J. Geophys. Res., 102, 22,183–22,197, 1997.
Keating, G. M., et al., The structure of the upper atmosphere of
Mars: In situ accelerometer measurements from Mars Global
Surveyor, Science, 279(5357), 1672–1676, 1998.
Krasnopolsky, V., On the deuterium abundance on Mars and
some related problems, Icarus, 148, 597–602, 2000.
Krasnopolsky, V. A., and G. R. Gladstone, Helium on Mars:
EUVE and PHOBOS data and implications for Mars’ evo-
lution, J. Geophys. Res., 101(A7), 15,765–15,772, 1996.
Schwadron, N. A., and T. E. Cravens, Implications of solar wind
composition for cometary X-rays, The Astrophysical Journal,
544, 558–566, 2000.
Shapiro, V. D., R. Bingham, J. M. Dawson, Z. Dobe, B. Kellet,
and A. Mendis, Energetic electrons produced by the lower hy-
brid waves in the cometary environment and soft X-ray emis-
sion: Bremsstrahlung and K-shell radiation, J. Geophys. Res.,
104, 2537, 1999.
Wegmann, R., H. U. Schmidt, C. M. Lisse, K. Dennerl, and J. En-
glhauser, X-rays from comets generated by energetic solar wind
particles, Planetary and Space Science, 46(5), 603–612, 1998.
H. Gunell, Swedish Institute of Space Physics, Box 812, SE-981
28 Kiruna, Sweden (firstname.lastname@example.org)
M. Holmstr¨ om, Swedish Institute of Space Physics, Box 812,
SE-981 28 Kiruna, Sweden (email@example.com)
E. Kallio, Finnish Meteorological Institute, Space Research,
P.O. Box 503, FIN-00101 Helsinki, Finland (firstname.lastname@example.org)
P. Janhunen, Finnish Meteorological Institute, Space Re-
search, P.O. Box 503, FIN-00101 Helsinki, Finland (email@example.com)
K. Dennerl, Max-Planck-Institut f¨ ur extraterrestrische Physik,
Giessenbachstraße, 85748 Garching, Germany (firstname.lastname@example.org)