Rosetta Langmuir probe performance - DiVA portal680862/FULLTEXT01.pdf1.3.1 Debye shielding and Debye...
Transcript of Rosetta Langmuir probe performance - DiVA portal680862/FULLTEXT01.pdf1.3.1 Debye shielding and Debye...
Rosetta Langmuir probe
performance
Fredrik Johansson
IRF Uppsala
Uppsala University
A thesis submitted for the degree of
MSc
2013 June
Abstract
Several Langmuir probe voltage sweeps by a model of the ESA spacecraft
Rosetta was simulated in a plasma with solar wind parameters using the
ESA open source software SPIS 5. The simulations were carried out to in-
vestigate the features of the spacecraft’s environment in the solar wind and
the effect of photoemission from the spacecraft surface on the measurements
made by the Langmuir probes on board Rosetta. We report a best fit to an
existing probe sweep result in the solar wind near the Earth at 1 AU from
9 Nov 2009 for a 4 million particle simulation in SPIS of an 8 V positively
charged spacecraft with the following parameters: Tph = 2 eV, Te = 12 eV,
Ti = 5 eV, ne = 5 cm−3. We also report that the spacecraft is shielding the
Langmuir probes on Rosetta from plasma electrons, and particularly low
energy electrons. In one instance, this blocking is shown to lead to an over-
estimation of solar wind electron temperature by 12% and underestimate
the plasma density by 24% by the Langmuir Probe for a +10 V charged
spacecraft in ne= 5 cm−3, Te = 12 eV solar wind. Two models used in lit-
erature on photoemission, one for isotropical emission from a plane and the
other for radial emission from a point, was used and compared. We report
a clear preference to the approximation of a Maxwellian energy distribution
of photoelectrons emitted radially from a point source model with our sim-
ulation result on the Langmuir Probe aboard Rosetta. We also report the
solar aspect angle dependence on the plasma potential and plasma density
result, which are in overall agreement with Rosetta measurements from the
second Earth fly-by.
Popularvetenskaplig sammanfattning
Ett av de kraftfullaste verktygen for att simulera hur en rymdfarkost paverkas
och paverkar sin omgivning i rymden, SPIS 5, har nyligen utokats med
mojligheten att simulera vetenskapliga experiment precis sasom de gors i
verkligheten. Rymden mellan solsystemets planeter ar inte tom utan fyllt
av plasma, det vill saga, elektriskt laddade partiklar. Nar en rymdfarkost
fardas genom rymden sa paverkas den av plasma och solljus och kan bade
ge ifran sig och absorbera partiklar. Detta forandrar plasmat nara rymd-
farkosten vilket ocksa paverkar de resultat vi far fran de vetenskapliga in-
strumenten ombord.
I denna rapport undersoker vi rymdfarkosten och kometjagaren Rosetta,
som ar 2014 ska observera och landa pa kometen Churuymov-Gerasimenko.
Ombord finns tva stycken svenskutvecklade Langmuirprober som mater
densiteter och laddningar pa plasmat for att undersoka vad kometen bestar
av, och hur en komet beter sig i den starka stralningen fran solen. Nagot
liknande har aldrig gjorts forut och eftersom kometer ar lika gamla som sol-
systemet sa hoppas man bland annat fa reda pa vad solsystemet skapades av
nar det foddes. For att kunna forsta de varden Langmuirproben uppmater
pa sin plats pa rymdfarkosten under sin resa, sa maste vi jamfora plasmat
innan rymdfarkosten paverkat det och efterat, och vad Langmuirproben
faktiskt uppmater.
Vi rapporterar har resultatet av simuleringar av en solbelyst modell av
rymdfarkosten Rosetta i ett plasma pa vagen till kometen, och vilka effekter
resultatet har pa vara matningar fran Langmuirproberna. Detta kan sedan
anvandas for att uppskatta vad ett ostort plasma har for egenskaper bara
med hjalp av vara matningar fran Langmuirproberna pa Rosetta. Med
hjalp av SPIS 5 sa kan vi ocksa visualisera resultatet och rymdfarkostens
omgivning pa ett vackert och lattillgangligt sett.
Acknowledgements
I would like to thank Anders Eriksson for invaluable and enthusiastic guid-
ance during the course of this master thesis work. I would also like to thank
Thomas Nilsson for teaching me the never-changing and problem-free life
that is making simulations in SPIS.
Contents
1 Background 1
1.1 Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 The Rosetta mission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Plasmas and the Solar wind . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3.1 Debye shielding and Debye length . . . . . . . . . . . . . . . . . 3
1.3.2 Particle motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.3 Photoelectric effect . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Langmuir probes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Theoretical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.5.1 Probe theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.5.2 Photoelectron current . . . . . . . . . . . . . . . . . . . . . . . . 8
1.5.3 Photoelectron cloud . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.5.4 Magnetic effects . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.5.4.1 v × B . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.5.4.2 Larmor radius . . . . . . . . . . . . . . . . . . . . . . . 11
2 Simulations 13
2.1 SPIS - Spacecraft Plasma Interaction System . . . . . . . . . . . . . . . 13
2.1.1 SPIS 5 and theory . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.1.2 SPIS materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2 Geometrical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.1 Mesh resolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
v
CONTENTS
3 Validation 19
3.1 Densities and potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.2 Previous work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4 Theoretical model performance 23
4.1 Plasma electron current . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.2 Photoelectron current in wake . . . . . . . . . . . . . . . . . . . . . . . . 24
4.3 Photoelectron current to a sunlit Langmuir probe . . . . . . . . . . . . . 26
4.4 Problematic zones for applying the model fit . . . . . . . . . . . . . . . 28
4.5 Electrostatic potential geometry . . . . . . . . . . . . . . . . . . . . . . 29
4.6 Second derivate of Langmuir probe sweep . . . . . . . . . . . . . . . . . 29
5 Comparison with probe sweep on Rosetta 35
6 Solar aspect angle dependence 39
6.1 Plasma potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
6.2 Density profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
7 Conclusions 47
7.1 Theoretical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
7.2 Rosetta Langmuir probe sweep . . . . . . . . . . . . . . . . . . . . . . . 48
7.3 SAA dependence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
7.4 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
References 51
vi
1
Background
———————————————————————-
1.1 Problem Definition
Rosetta is a spacecraft from the European Space Agency (ESA), on its way to the
comet 67P/Churyumov-Gerasimenko for a first of its kind in situ investigation, as well
as the most detailed environment investigation of a comet ever attempted. Comets
are believed to be the primitive building blocks of the planets in the Solar system
and seeded the Earth with water and possibly even life[12]. By studying a comet’s
dust and gas, its structure and abundances, the Rosetta Mission will help further the
understanding of the evolution of the Solar system and the role of comets therein.
The purpose of this report is to investigate the plasma environment around the
spacecraft and the effect of photoelectron emission from the Rosetta orbiter on the
measurements made on one of the instruments on board, the Langmuir probe instru-
ment. This is done by simulating the spacecraft environment and the data measured
by the Langmuir probe instrument, which is used to investigate the properties of the
plasma in the solar wind and surrounding the plasma, and compare it to real data from
the S/C.
1.2 The Rosetta mission
The Rosetta spacecraft consists of a 2.8×2.1×2.0 meter orbiter with two 32 m2 solar
panels and carries a smaller spacecraft, the lander, which will land on the comet.
1
1. BACKGROUND
Launched in March 2004, the Rosetta spacecraft is currently undergoing a ten year long
journey to comet 67P/Churyumov-Gerasimenko, having already completed three Earth
fly-bys and one Mars fly-by. The purpose of the fly-bys were twofold: providing the S/C
with the gravitational boost needed to reach its destination at correct angle of approach
as well as creating an opportune moment to test its instruments and investigate the
planets’ surroundings. The Rosetta mission is now less than a year from reaching the
same orbit and position as the target comet at 4 AU, and the start of its mission. The
duration of the mission is scheduled for at least a year from January 2014 until the
comet reaches its perihelion (the point closest to the Sun) at 1.2 AU[13].
Rosetta carries a multitude of instruments including two Langmuir probes designed,
manufactured and operated by the Swedish Institute of Space Physics (IRF) in Uppsala,
Sweden.
1.3 Plasmas and the Solar wind
As Rosetta travels to its destination, it is traversing through a medium dominated by
plasma (a collection of ions and electrons) ejected from the Sun, called the Solar wind.
The parameters of the plasma are depending on the condition of the Sun at the time
of its emission.
The fourth state of matter, plasma, as defined by Francis F. Chen, is a:
”quasineutral gas of charged and neutral particles, which exhibits collec-
tive behaviour” [1].
• For an ordinary neutral gas, the force of gravity between individual particles is neg-
ligible and the motion is usually controlled by collisions between individual particles.
This can also be true for a plasma, but as most of the particles in a plasma are charged,
the dominant motion driver is instead the total electric and magnetic field force on the
plasma. The term ”Collective behaviour” is reflecting on the movement of the charged
particles and the currents and fields thereby induced in the plasma, which influences
other plasma particles at a distance and thus drives the motion of the plasma as a
whole. If the plasma is sufficiently tenuous, we can speak of a collisionless plasma, and
disregard collisions completely for calculating the motion of the particles. This is the
case for all plasmas discussed in this work.
2
1.3 Plasmas and the Solar wind
Quasineutrality is a feature of the plasma of how the macroscopic picture of a plasma
is overall neutral, but the microscopic picture of individual plasma particles is far from
neutral. The scale length of this effect and other implications of quasineutrality deserve
a further explanation.
1.3.1 Debye shielding and Debye length
Debye shielding [1] is an innate ability of the plasma to shield out local potentials
within. If a body such as a probe, is put into a plasma, it will attract ions or electrons
depending on the relative potential difference. In doing so, it creates a small volume
sheath around the body with opposite charge of that body, which reduces the potential
of the body as it is felt by particles outside this volume. This phenomenon is known as
Debye shielding and the characteristic thickness of the sheath is known as the Debye
length.
This is also proven mathematically[1] by considering an electrostatic potential of a
single point particle of charge q at a distance r:
V (r) =q
4πε0r, (1.1)
where constants such as ε0,kB, etc. have their usual meaning in the body of this report.
It can then further be shown[1] that the effective potential seen when in a plasma will
be reduced to
V (r) =q
4πε0re− rλD , (1.2)
where λD is the Debye length, given by
λD =
√ε0kBTenee2
, (1.3)
where Te is the temperature of the electrons and ne is the density of electrons.
For mechanisms working at much larger distances than the characteristic scale
length λD, the potential is effectively shielded out. At such distances, the position
and motion of individual charges are therefore not important, and only collective ef-
fects of all particles need be considered. This is the main regime of consideration in
plasma physics.
3
1. BACKGROUND
1.3.2 Particle motion
The solar wind plasma is a very tenuous plasma, and thus assumed to be collisionless.
However it can still be assumed to follow a Maxwell-Boltzmann distribution of the
particle velocities, where the particles have a thermal motion with a mean thermal
velocity vth, given by:
vth,i =
√kBTimi
, (1.4)
where T and m is the temperature and mass of particle species i.
The bulk motion of the solar wind is assumed to be a uniform non-varying plasma
flow speed, vSW , taken to be the typical value of 400 km/s, flowing radially outward
from the Sun.
1.3.3 Photoelectric effect
To complete the picture of the spacecraft-plasma environment, we must also take the
photoelectric effect into account. When electrons in a molecule are hit by photons,
the molecule may absorb photons of specific energy intervals to excite and possibly
emit electrons. For metal surfaces exposed to photons, such as the spacecraft body in
sunlight, this effect has a measurable result in terms of overall charge of the surface,
and also creates a photoelectron cloud around the spacecraft body, moving with the
spacecraft.
It is also important for comets like comet 67P/Churyumov-Gerasimenko, where it is
responsible for ionizing the comet plasma tails, and to a smaller extent, the immediate
surroundings of the comet, called the comet coma.
The photoelectric effect, and the photoelectrons emitted by the effect is thus very
much depending on the solar spectrum and flux, the emission and absorption properties
of the material of the surface subjected to photons, such as the photoemission saturation
current, jf0, and the angle of incidence.
1.4 Langmuir probes
The Langmuir probe was invented by Nobel laureate Irving Langmuir (1881 − 1957). It
is used to determine various properties of a plasma such as temperatures and densities
4
1.5 Theoretical model
of different particle species. The underlying principle is that of an electrode immersed
in a plasma with a variable voltage being applied to it from a power supply, attracting
positive ions at a negative voltage and attracting electrons e.t.c. at a positive voltage.
The current flowing to the probe is then measured and will depend on the parameters
of the plasma measured.
The Langmuir probe instrument (LAP) on Rosetta consists of two sensors mounted
on two booms and associated electronics inside the spacecraft body, and was developed
by the Swedish Institute of Space Physics (IRF) in Uppsala, Sweden. It is part of the
Rosetta Plasma Consortium (RPC) on board Rosetta. The booms are of lengths 2.24
and 1.62 m and the probes themselves are 50 mm diameter titanium spheres covered
by titanium nitride (TiN) mounted on 15 cm stubs, see fig. 1.1[5][3].
Figure 1.1: Langmuir probe - Identical copy of Langmuir probe on Rosetta, image
courtesy of A. Eriksson
By investigating the plasma density, electron temperature and flow speed as well
as the time and space variations of these parameters, the LAP will give us an un-
precedented view of cometary outgassing and plasma environment, and will study the
evolution and activity of the comet 67P/Churyumov-Gerasimenko as it approaches its
perihelion. Some of the quantities measured and the operational ranges is listed in
fig. 1.2. The LAP can also be used in conjunction with the other RPC instruments on
board the Rosetta spacecraft to investigate phenomena none of them could do on its
own, such as magnetohydrodynamical waves [4].
1.5 Theoretical model
1.5.1 Probe theory
To quantify the probe current and separate it into its different particle contributions,
we use a theory first developed by Mott-Smith and Langmuir[10] called Orbital motion
5
1. BACKGROUND
Figure 1.2: Table of parameters accessible to LAP - The LAP has several different
modes of operation that collectively are capable of producing all parameter and their
ranges, but not individually or at the same time and every plasma [4].
limited theory (OML). It regards the plasma as a particle distribution moving in a
vacuum field from the probe, and obtains trajectories solely based on energy and angular
momentum conservation. This approach can only be adopted when the radius of the
probe is much smaller than the Debye length. For larger probe radii, a separate theory
called sheath limited theory (SL) need to be used, but for the plasmas considered and
the instruments used for this report, we can safely assume
rprobe � λD, (1.5)
where typical values of the Debye length in the Solar wind is ∼ 10 m and the probe
radius is 25 mm.
As a starting point, OML examines the current, I, to a body of zero net charge with
particle motion dominated by the particles thermal velocity. The particles are also
assumed to follow a Maxwellian energy distribution, which gives the thermal current
Ith, given by[17]:
Ith =∑i
Apqini
√kBTi2πmi
∀ particle species i, (1.6)
where Ap is the surface area of the probe. For convenience, temperatures will hereafter
be measured in units of eV, where 1 eV = 1.16× 104 K.
This is the random current to a neutral probe, but during a Langmuir probe voltage
sweep, the current varies as a function of the probe potential. This is complicated
further by that the potential applied to the probe, the bias voltage, is not the absolute
6
1.5 Theoretical model
voltage of the probe itself. Instead, the voltage of the spacecraft is often non-zero and
not known before the measurement. i.e:
Vp = Vb + VS/C , (1.7)
where Vp is the probe potential, VS/C is the potential of the spacecraft and Vb is the
bias voltage applied to the probe. Why the potential of the spacecraft is not zero and
the implications of this is discussed further in section 2.1.
Electron current
Now, assuming all particles coming from a zero potential at infinity, and energy and
angular momentum conservation, it can be shown [2] that the OML current for electrons
follows:
Ie =
Ie0(
1 +VpTe
)for Vp ≥ 0
Ie0eVpTe for Vp < 0,
(1.8)
where Ie0 is the thermal current Ith for electrons, given by
Ie0 = Apene
√kBTe2πme
(1.9)
From the equation, we can see that the plasma electron current, dominated by
the thermal motion of electrons, increases linearly for a positive probe and decays
exponentially for a negative probe. A positive current is defined as the flow of a
negatively charged particle to the node, by convention.
Ion current
Ions in the solar wind, being heavier than the electrons, are not as influenced by
their thermal energy, but their motion is instead dominated by their drift velocity,
vd. Engwall (2006) [2] shows that the drift velocity current contribution to a charged
probe is instead
Iion =
{Iion0
(1− 2qionVp
mvd
)for vd ≥ vmin
0 for vd < vmin,(1.10)
7
1. BACKGROUND
where vmin is the minimum velocity, given by
vmin =
{√2qVpm for qVp > 0
0 otherwise,(1.11)
and
Iion0 = −Aramqionnionvd, (1.12)
where Aram is the cross-sectional area of the probe. Comparing eqs. (1.8) and (1.10),
we see that at attracting potential, the drift energy 12mv
2d has replaced the thermal
energy Te, and the drift current Aramqionnionvd has replaced Ie0 but with the area
normal to the ion flow replacing the total surface area. For a repelling potential,
eq. (1.10) is actually consistent with eq. (1.8) in the limit T → 0.
Also, as the solar wind is assumed to have a zero net charge, we can write
ne ≈ nion (1.13)
1.5.2 Photoelectron current
As discussed in section 1.3.3, the probe and the spacecraft body will emit photoelectrons
when hit by the sunlight, and this emission depends on the properties of the material
subjected to it and the potential of the probe. As the probe potential increases, fewer
electrons are emitted from the probe and more electrons are being reabsorbed.
How this photoemission behaves on the body of the Langmuir probe in space is not
easily modelled since it depends on the material of the probe, its geometry as well as
the electric field surrounding the probe. Since the electric field is highly dependent on
the plasma-spacecraft interaction, no obvious solution is known.
In this study we adopt two models suggested by J.R.L Grard[7] for two cases of
photoemission, both proven to be successful from previous studies[7][17].
Model 1
Both models consider the electrons as being emitted at certain energies somewhere in-
side the material, and through collisions within the material, exits the material with a
Maxwellian energy distribution. This was also proven to be consistent with experimen-
tal results for a wide range of materials[7]. There are however two important cases to
8
1.5 Theoretical model
consider. First, Rejean J.L. Grard et al[7] suggests the case of a Maxwell distribution
of electrons emitted isotropically from a plane surface:
If =
−If0e−VpTf for Vp ≥ 0
−If0 for Vp < 0(1.14)
where If0 is the photo emission saturation current, jf0 of the probe surface.
Here the current magnitude decreases to zero as the potential of the probe increases,
and more emitted electrons are being reabsorbed. The current reaches its maximum
negative value when the probe is highly negative and all emitted electrons are being
repelled towards infinity. This model features a very sharp transition from positive to
negative probe voltage and an example can be seen in fig. 4.4.
Model 2
Rejean J.L. Grard et al [7] also discusses a second model for the probe photoemission
current, which introduces a second linear term in the positive probe potential case. This
model describes emission from a point source, which may more accurately describe the
photoemission from the spherical probe:
If =
−If0(
1 +VpTf
)e− VpTf for Vp ≥ 0
−If0 for Vp < 0(1.15)
This model is smoother in the transition region between positive and negative probe
potentials (see fig. 4.4), but there is no obvious conclusion on which of the photoemis-
sion models is preferred in nature. The simulation software SPIS simulates isotropical
photoelectron emission from all surfaces, so naıvely, eq. (1.14) would be preferred. How-
ever the question is rather if the curvature of the Langmuir probe is large enough, and
the sphere small enough, to be approximated as a point source emission. Which would
mean that an isotropical distribution from the sphere is equal to a radial distribution
from a point. Therefore we will use both models for comparison to real data and the
simulation results until a preference is determined.
1.5.3 Photoelectron cloud
Photoelectrons are of course also emitted from the spacecraft body, booms and solar
panels, and are the source of a photoelectron cloud surrounding the spacecraft and the
9
1. BACKGROUND
Langmuir probes. If the photoelectron cloud is large enough, these electrons will be
absorbed by the probe as well. If we assume that the cloud is fairly homogeneous, we
can assume a similar current as eq. (1.8), reasonably as a function of probe bias voltage
instead:
IS =
IS0(
1 + VbTph
)for Vb ≥ 0
IS0eVbTph for Vb < 0,
(1.16)
where Tph is the temperature of the emitted photoelectron in eV and IS0 is given by:
IS0 = Apenph
√kBTph2πme
, (1.17)
where nph is the number density of photoelectrons.
This model has been proven to work well in previous studies[17].
1.5.4 Magnetic effects
So far we have disregarded some effects that complicate the motion of a charged particle
in a plasma, namely, magnetic effects. In this section we will go through two of the
most important magnetic effects to motivate why we can ignore any magnetic effects
in the model.
1.5.4.1 v × B
The interplanetary magnetic field at 1 AU is very small, about 10 nT[17], but will induce
an electric field for a moving charge in the plasma surrounding Rosetta according to:
E = v ×B (1.18)
where the solar wind velocity is about 400 - 500 km/s and the spacecraft velocity is
negligible in comparison.
The maximum induced electric field will then be for a velocity perpendicular to the
magnetic field, such that:
E = 400× 103 × 10−8 = 4× 10−3 V/m. (1.19)
Since the Langmuir probes are located on booms of one to two meters in length, the
maximum induced potential will be on the order of 0.01 V, which is so small that we
can consider it negligible.
10
1.5 Theoretical model
1.5.4.2 Larmor radius
The equations in section 1.5.1 are valid for unmagnetized plasmas, however the solar
wind is also magnetized. As a charged particle move through a magnetized plasma,
it will have a gyrating motion, and may limit the Langmuir probe sample to a small
column of particles moving perpendicular to the magnetic field if the gyro radius is
sufficiently small.
The Larmor radius, or gyroradius, is given by:
rL =mv⊥qB
, (1.20)
which for an electron of temperature 12 eV, moving in the interplanetary magnetic field
of 10 nT becomes ≈ 200 m. This is very much larger than the probe radii of 25 mm,
so we can safely disregard magnetic effects when applying our models.
11
1. BACKGROUND
12
2
Simulations
2.1 SPIS - Spacecraft Plasma Interaction System
Spacecraft charging, where a spacecraft travelling in various plasma environments is
positively or negatively charged due to interactions with the plasma, poses a real threat
for commercial and scientific spacecraft alike. Local potential differences on a space-
craft can lead to sparks that can disable entire subsystems on the S/C. One notable
example of that is Japanese satellite ADEOS, which was lost following severe space-
craft charging[11]. The problem of regulating the voltage over all the spacecraft surfaces
needs to be accounted for in the design of the spacecraft, and potential hazards iden-
tified. The only reasonable solution is then to model the design and test it through
simulations.
Also, the plasma-spacecraft-induced environment around the spacecraft perturbs
any plasma measurements by scientific instruments on-board the spacecraft, a problem
that cannot always be designed away cost-effectively and thus needs to be modelled
accurately to understand the measurements.
To combat these problems for a spacecraft of any shape, size and material and
various plasma environments, ESA started the initiative of a new simulation software
in 2002 that later became the open source software SPIS, Spacecraft Plasma Interaction
System. It has many features and capabilities (too many to list here) but essentially, it is
a Particle-In-Cell (PIC) routine that solves the Gauss’ law for electric fields,∇2φ = − ρε0,
for particles and secondary particles of many different particle distributions [9].
SPIS is built using a modular approach, so that the underlying core, the GUI or any
13
2. SIMULATIONS
open source plug-in can be updated and inserted with greater ease. Although funded
mostly by ESA, SPIS relies on an open-source license (GPL) to enable a community-
based development, and high compatibility with many systems and file formats [15].
2.1.1 SPIS 5 and theory
The study presented was made using the latest SPIS version at the a non-released
development version, SPIS 5.0.0 [16]. This latest version supports the simulation of
a particle detector experiment, such as a full Langmuir probe Sweep inside the sim-
ulation and handles the previously computationally very demanding particle detector
simulation a rather ingenious way.
SPIS is normally simulating the plasma-spacecraft interaction by solving the Poisson
equation for each macro-particle to see how the plasma around a spacecraft evolves with
time. The differential equation is solved using a Runge-Kutta step method, which is
one of the most stable and a reasonably efficient method to solve differential equations
and can be fully parallelized.
New in SPIS 5 is the ability to perform a separate particle detection experiment
at any stage of the plasma simulation, called backtracking. Liouville’s theorem for a
Hamiltonian dynamical system, states that:
” The distribution function is constant along any trajectory in phase
space.” [18].
Equivalently, we can say that the existence of a conserved current in a system implies
that the system is invariant under time translation. We can thereby freeze the system
at any point and emit test particles in the negative time direction from a particle
detector. The test particles are weighted by the value of the distribution function of
the tracked species at the surface it finally encounters, either on the spacecraft or on the
simulation boundary, to simulate particles of different species and emitted at different
energy levels.
The test particles final destination determines what particle type is detected on
the particle detector. An electron whose origin in the positive time direction is the
boundary of the spacecraft model is of course a plasma electron type, and an electron
emitted from the spacecraft is a photoelectron type, etc.
14
2.1 SPIS - Spacecraft Plasma Interaction System
This backtracking method enables SPIS to dedicate processor time to only the
particles that the particle detector detects, without losing information of the plasma
surrounding the particle detector. The implication of this is enormous as the random
noise in a particle detector result is proportional to√n, where n is the number of
particles detected. To achieve a reliable result, a particle detector experiment normally
prefers the simulation volumes to be small, particle detection areas to be large and
particle densities to be high. Any deviation from the ideal set-up will result in large
errors for short simulations, or large computational time costs for high accuracy mea-
surements. Backtracking enables SPIS to simulate the real world scenario, with tenuous
plasma in large simulation volumes. It is also enabling the use of small detection areas
and reaching a sufficient accuracy within a reasonable simulation time with the use of
quite modest modern computers.
2.1.2 SPIS materials
SPIS has a large collection of pre-defined materials, and the possibility to define your
own. Any spacecraft volume is treated as an empty space where no particles are
allowed with infinitely thin surfaces surrounding it. These surfaces can then be defined
to mimic the properties of any material of a certain thickness and can both absorb and
emit particles. This is no small a task for a spacecraft, which consists of many parts of
all kinds of materials and thicknesses. For simplicity, the entire spacecraft, solar panels
and Langmuir probes in this model is treated as Indium Tin Oxide, ITO,[6] which the
solar arrays are coated with, and whose properties are similar to the material of the
probe. While the details of the actual surfaces differ, the most important property, i.e.
that they are conductive, is the same.
One current limitation of SPIS is that the mean photoelectron temperature is de-
fined globally for each surface of the simulation. In reality, material photoemission
properties are depending on the quantum structure of the material, and thus rather
unique for each material, including but not limited to the mean photoelectron tem-
perature [7]. As this limitation is inherent to SPIS, the only additional approximation
caused by assuming all s/c surfaces are of the same material is that the photosaturation
current is set equal everywhere.
Also, we do not simulate all types of ions present in the solar wind, but only
the most abundant, H+. Regardless, we expect only a negligible contribution to the
15
2. SIMULATIONS
detected currents from any ions for the voltage ranges used in our simulations. This
is because the ions are highly energetic and scarcely affected by the relatively small
voltage ranges simulated on the S/C and the Langmuir probe.
2.2 Geometrical model
Some simplifications are made to save computational time in the making of the model.
We modelled Rosetta, see fig. 2.1 in a prolate spheroid simulation volume, which is an
ellipse with major axis b = 30 m and minor axis a = 15 m rotated around its major
axis. Rosetta itself was modelled as a 2.8 × 2.1 × 2.0 meter cuboid of Indium Titanium
Oxide (ITO) with two 32 m2 ITO solar arrays and two ITO cylindrical booms with two
probes corresponding to the Langmuir probes and booms of the same shape and size.
The model was originally created by Alexander Sjogren [14] in 2009, but was reworked
and improved upon to accommodate the new SPIS version, including the addition of
detailed models of the Langmuir probes and improved models of the probe booms on
Rosetta.
Figure 2.1: Model mesh - The 2D meshed model of Rosetta with Langmuir probes
and booms as used in SPIS. Top right: The entire simulation box, containing Rosetta and
support rendering boxes.
16
2.2 Geometrical model
2.2.1 Mesh resolution
The model mesh represents volumes by tetrahedrons and surfaces by triangles of dif-
ferent sizes to accommodate a varying resolution in the volume to make the simulation
more time efficient. The resolution is most refined near the probes, followed by the
booms, spacecraft body, solar panels and simulation box borders in decreasing order of
resolution.
One obvious source of error would be the resolution of the spherical Langmuir
probe. As a circle has infinite amount of edges, the perfect sphere surface would have
to be represented by an infinite amount of triangles, otherwise resulting in errors in the
volume and surface area of the probe, see fig. 2.2. But since the backtracking algorithm
in SPIS defines each surface of a langmuir probe as a separate particle detector, the
resolution of the probe was found to be one of the largest computational time driver,
and as such, needed to be limited.
The resolution chosen represented the probe as a surface of 124 triangles with a
surface area corresponding to 92.4% and volume of 87.2% of the actual probe. The
longest simulation time with this resolution was 24 hours.
17
2. SIMULATIONS
Figure 2.2: Probes at different resolutions - Four spherical probes meshed at a
resolution of 30 to 2624 triangles, from bottom right to top left. The probe used in the
simulations reported is pictured bottom left, consisting of 124 triangles with volume and
surface area of 87% and 92% of the actual Langmuir probe, respectively. Note that even
at the highest resolution, the probe is not quite spherical, and one would require a more
complicated model to achieve a higher volume ratio than 96.7%
18
3
Validation
3.1 Densities and potential
Many simulations were performed to test stability and correctness of the simulation
parameters. An overview of the important features of the plasma-spacecraft interaction
is given in figs. 3.1 and 3.2.
Figure 3.1: Photoelectron and ion densities visualisation - Photoelectron density
increasing from to blue to yellow in the xz-plane, ion density increases from blue to red on
the xy-plane and the Rosetta spacecraft depicted as white. The Sun is in the positive x
direction. Simulation parameters: 4 Million particle simulation, VS/C = 10 V, Te = 12 eV,
Tion = 5 eV, Tf = 2 eV, ne = 5 cm−3 solar wind at v = 400 km/s. Simulation name:
EFSRE8V1.2eV-220513-1.
19
3. VALIDATION
Figure 3.2: Electrostatic potential overview - Equipotential volume shells are shown
increasing in potential from blue to red for a +10 V charged Rosetta spacecraft. The Sun
is in the positive x direction. Simulation parameters: 4 Million particle simulation, Te =
12 eV, Tion = 5 eV, Tf = 2 eV, ne = 5 cm−3 solar wind at v = 400 km/s. Simulation
name: optimeradtest-100413-3.
We clearly detect an ion wake behind the spacecraft, shown in dark blue, due to
the ion flow velocity is much greater than the thermal velocity in nominal solar wind,
as expected. We also detect a photoelectron cloud distribution surrounding the S/C
and extending in the sunward direction. The highest densities of photoelectrons are
located in the immediate vicinity of sunlit surfaces, with a rapidly decreasing density
in the negative x direction from the S/C.
The electrostatic potential from the spacecraft decreases in magnitude rapidly with
distance from the spacecraft and reaches 1 V at approximately 10 meters from the
spacecraft in the +y and -y directions. Clearly, the solar array is the largest influence the
surrounding plasma, both in terms of photoemission and generating the photoelectron
cloud, but also in terms of dominating the electrostatic potential geometry.
3.2 Previous work
To validate the new version of SPIS, SPIS 5, a reference simulation of the same exact
parameters as Alexander Sjogren SPIS 3.7RC05 version was made, and some results
20
3.2 Previous work
are shown in fig. 3.3 below.
Figure 3.3: Photoelectron density(left column) and ion density(right column) of refer-
ence simulation in the new version of SPIS (top) and Sjogrens SPIS 3.7RC05 simulation
(bottom)[14]
In Sjogren’s geometrical model, the booms were cuboid and the Langmuir probe
itself missing, whereas in this version we have added detailed models of the probe and
used cylindrical booms. The new SPIS version also allows for much faster performance
due to parallelization and the full utilisation of multiple processors. Comparing overall
features of ion, electron and photoelectron distribution, as well as potentials at various
points we find no discrepancy more than was expected from simulation uniqueness,
where every simulation is different due to the randomness in meshing the model volume
and sourcing the plasma particles.
21
3. VALIDATION
22
4
Theoretical model performance
Simulations of the Langmuir probe sweeps were made, where the current detected on
the probe is measured at a stepwise varying potential and subsequently compared with
the theoretical model to test our understanding of the problem in the following sections.
The parameters chosen for all simulations (unless otherwise specified) were the same
as Sjogren’s reference simulation [14] of a Rosetta spacecraft in a nominal solar wind,
and detailed in Appendix A.
4.1 Plasma electron current
SPIS generates separate particle detection simulations for each particle population, and
can thus separate each contribution with ease. To test our model we separated and
simulated only the plasma electron current, from a Langmuir probe sweep and compared
with the expected result, see fig. 4.1. We also adjusted the ne and Te parameters to
the theoretical expression to obtain a good fit to the simulated sweep
The plasma electrons are repelled by the probe when the probe potential is negative
and attracted linearly otherwise. All three electron sweeps reveal identical results, with
best fits obtained for Te = 13.5(± 0.1) eV and ne = 3.83(± 0.01) cm−3. This is however
a lower current than we expected from the theoretical model and suggests that the probe
is somewhat shielded from electrons, and especially electrons with less energy, which
gives an overestimation of the plasma electron temperature and underestimation of
the plasma density. The plasma electron density is also confirmed to be lower near
23
4. THEORETICAL MODEL PERFORMANCE
Figure 4.1: Plasma electron sweep - Multiple electron sweeps with different potential
steps and VS/C = 10, 8 and 7 V (Black, red and green circles). 4 Million particle simu-
lations, Te = 12 eV, Tion = 5 eV, Tf = 2 eV, ne = 5 cm−3 solar wind at v = 400 km/s.
Theoretical result for simulation parameters(red line) and model fit fromeq. (1.8) (blue
line).
all spacecraft surfaces by investigation of the 3D-plots of the plasma, see fig. 4.2, thus
suggesting that the model is valid, and that the discrepancy has a physical origin.
4.2 Photoelectron current in wake
To further test and separate our understanding of the photoelectron current to the
Langmuir probe, a simulation was carried of a Langmuir probe sweep while in the wake
and shadowed by the spacecraft. As there is no photoemission on objects in shadow, the
only photoelectron current detected would be the absorption of photoelectrons from the
photoelectron cloud, as detailed in eq. (1.16). The result is then plotted and compared
to the theoretical result of different scenarios, see fig. 4.3.
The result was far from expected, so the simulation was redone with different poten-
24
4.2 Photoelectron current in wake
Figure 4.2: - Electron density 3D-plot of a nominal solar wind parameter simulation
EFSRE8V1.2eV-220513-1, with the Sun is in the +x direction.. The Rosetta Spacecraft
and Langmuir probe 1 visible in gray
tial steps to assure that the unexpected result didn’t arise from numerical divergence
errors in the code or incorrect circuit relaxation timescales on the particle trajectories.
Also other solar aspect angles in the shadow of the S/C was investigated and found to
yield similar results for both probes.
Quickly apparent is that the knee of the function does not occur around Vb = 0, as
expected by the model in eq. (1.16), but otherwise follows the same shape. This model
has proven useful in reality for other scenarios on Rosetta[17], but leads in this scenario
to an unrecoverable overestimation of either Tph or nph.
The photoelectrons, largely occupying a space of non-zero potential originating
from the potential of the spacecraft, will instead be attracted or repelled by the probe
depending on the relative potential of the probe to its surroundings. Similarly, pho-
toelectrons emitted by the probe is ejected into a plasma with a non-zero potential,
Vplasma, and will not be reabsorbed if the probe is more negatively charged than its
immediate surroundings. Therefore, the photoelectron current should instead be a
function of Vp − Vplasma. To accommodate this, we substitute this for Vp in eqs. (1.14)
and (1.15) and for Vb in eq. (1.16) and test it for a sweep on a sunlit probe.
25
4. THEORETICAL MODEL PERFORMANCE
Figure 4.3: Photoelectron sweep in wake - Two simulations(LP2SAA-40-270513-
1 and LP2SAA-40-270513-2) of langmuir probe sweeps with different potential steps is
plotted in blue diamonds of a nominal solar wind and Tph = 2 eV. The theoretical model
result of different scenarios is overlayed in blue, green and purple (for repelling potentials)
and a fitted red line for attracting potentials. The expected nph from 3D plots is on the
order of 10 cm−3, and the mean photoelectron temperature, Tph is 2 eV.
4.3 Photoelectron current to a sunlit Langmuir probe
In this scenario, the probe is sunlit at 180 ◦ SAA while conducting the Langmuir probe
sweep. The result is then compared with the competing models from section 1.5.1,
combined with the photoelectron emission current discussed in section 4.2 to determine
which model describes our results best.
As all plasma and photoelectron parameters are input parameters to the simulation
and hence are known, there are only two free parameters for fitting the simulated probe
sweep to the theoretical expressions: the photoelectron cloud density at probe position,
nph, and the plasma potential at probe position, Vplasma. We adjust these manually
until the best fit is obtained, with the results as shown in table 4.1.
When the absolute potential of the probe is highly negative, all emitted photo-
electrons are emitted from the probe to infinity but as the potential increases, some
26
4.3 Photoelectron current to a sunlit Langmuir probe
-‐1.00E-‐07
-‐5.00E-‐08
0.00E+00
5.00E-‐08
1.00E-‐07
1.50E-‐07
-‐10 -‐8 -‐6 -‐4 -‐2 0 2 4 6 8 10
I vs Vb
Backtrack 120313
Model 1
Model 2
Figure 4.4: Langmuir probe sweep and theoretical models - Current vs bias volt-
age. SPIS 8.3 million particle simulation probe sweep (blue diamonds) on a sunlit Langmuir
probe with a +10 V charged spacecraft at 1 AU, in Te = 12 eV, Tion = 5 eV, Tph = 2 eV,
ne = 5 cm−3 solar wind at v = 400 km/s. The Model 1 (red line) and Model 2 fit (green
line) includes the photoelectron cloud current from the eq. (1.16) and are both offset by
some potential, Vplasma, as discussed in section 4.2.
Model 2, Equation (1.15) Model 1 Equation (1.14)
If0 (A) 6.0× 10−8 6.0× 10−8
Vfloat(V) 6.4 ± 0.1 7 ± 0.1
nph (cm−3) 14.13 ± 0.3 14.5 ± 0.3
Te (eV) 12 12
Tf (eV) 2 2
Tion (eV) 5 5
Table 4.1: Model comparison - parameter results of model fits to probe sweep from
fig. 4.4.
of the emitted photoelectrons are reabsorbed by the probe as described in eqs. (1.14)
and (1.15). Also, as the potential of the probe is positive w.r.t. the photoelectron cloud,
we observe a current linearly increasing with the potential, offset by some potential,
27
4. THEORETICAL MODEL PERFORMANCE
Vplasma, as we predicted.
The results in table 4.1 show that the models are consistent with regard to the num-
ber density of the photoelectrons, but yields significantly different results for Vplasma.
Combined with the result from fig. 4.4, we find that model 2 describes the simulation
result confidently and was then chosen to yield the experimental results of the parame-
ters Vplasma and nph in the body of this report. The final photoelectron current model,
called Model 3, then becomes:
Iph = IS + If , (4.1)
where
If =
−If0(
1 +V†Tf
)e−V†Tf for V† ≥ 0
−If0 for V† < 0,(4.2)
where
V† = Vp − Vplasma, (4.3)
and
IS =
IS0(
1 +V†Tph
)for V† ≥ 0
IS0eV†Tph for V† < 0.
(4.4)
Note that by using the model for photoemission from a point, we by no means
suggest that the probe is a point and that all photoelectrons are emitted radially in
reality (or in SPIS). We merely suggest that it seems that the probe is sufficiently small
so that we can approximate it to a point in the model.
4.4 Problematic zones for applying the model fit
The model works perfectly well for a wide range of solar aspect angles, but at certain
scenarios we find fitting the model not as straightforward, as seen in figure fig. 4.5.
The problem arises only when the probe is sunlit, in the wake and the boom is
fully or partially in the shadow of the spacecraft. At a SAA of -70◦, Langmuir probe
2 (LP2) is fully sunlit but in the ion wake and most of the boom is in the shadow of
the spacecraft. Only by erroneously using a too high photoelectron temperature do we
reach a satisfying fit, and this also yields a slight overestimate of the plasma potential
and photoelectron density.
28
4.5 Electrostatic potential geometry
4.5 Electrostatic potential geometry
The OML theoretical model for electrons and photoelectrons assumes particles coming
from a zero potential source at infinity, approaching a probe with a monotonic electro-
static potential field geometry of a plane or a point. However, as seen in fig. 4.6, the
electrostatic potential is far from spherically symmetric, and the probe is fully inside
the electrostatic potential generated by the spacecraft.
Only very close to the probe does the electrostatic potential field look spherical,
and approaches a cylinder with a spherical top as the distance to the probe increases
until the field is fully dominated by the spacecraft potential. This geometry is not
straightforward to model analytically for a probe with varying potential, but the model
works very well to yield precise results for most scenarios. When the boom is shadowed
by the spacecraft, the probe loses a large and easily accessed source of photoelectrons,
and this is exactly when the model fitting fails and yields very imprecise results, and
we need to acquire a plasma potential measurement by other means.
4.6 Second derivate of Langmuir probe sweep
Since we expect a sudden sharp increase of the gradient in the CV function around
the floating potential, we expect a peak in the second derivative of the CV function
somewhere near the floating potential. This means that the second derivative should
show a maximum around the floating potential. As differentiating a signal always
increases noise, fitting to an exact theoretical expression of the probe current can be
problematic. However, we should be able to recover the peak location, i.e. the floating
potential, by fitting to some generic function with a localized peak, e.g. a Gaussian
curve. The second leapfrog derivative was then plotted in fig. 4.7 and fitted with a
Gaussian curve.
Here we find that the first peak of the second derivative can clearly be used to
accurately deduce the potential at the probe position, and is remarkably consistent
with the model fit result of 6.4 (±0.1) V. We have no other analytical support for using
a Gaussian fit routine to the second derivative, merely other than it works very well.
This Gaussian fit routine will not work for a probe in shadow, as the electron
and photoabsorption current does not follow the same exponential function as the
photoemission current. However, it provides a means for getting a secondary estimate
29
4. THEORETICAL MODEL PERFORMANCE
whenever our theoretical model fails to produce precise results, such as the regions
discussed in the previous section
30
4.6 Second derivate of Langmuir probe sweep
(a)
(b)
(c)
Figure 4.5: CV sweep and model fits, current (A) vs Vb (V) - In the right column
we have Model 3, as detailed by eq. (4.1) fitted to the simulation result . The model is then
separated into photoelectron emission and photoelectron cloud absorption parts, where the
photoelectron emission part is subtracted from the simulation result and plotted against
the photoelectron cloud absorption model (upper left) or vice versa (lower left) on each
row.
31
4. THEORETICAL MODEL PERFORMANCE
(a)
(b)
(c)
Figure 4.6: Electrostatic potential geometry surrounding the Langmuir probe
- Equipotential shells in the plasma for a +10 V charged S/C for a +10 V probe (a), +8V
charged S/C for a +5.2 V probe (b) and -1 V probe(c).
32
4.6 Second derivate of Langmuir probe sweep
-‐1.00E-‐08
-‐5.00E-‐09
0.00E+00
5.00E-‐09
1.00E-‐08
1.50E-‐08
2.00E-‐08
2.50E-‐08
-‐10 -‐9 -‐8 -‐7 -‐6 -‐5 -‐4 -‐3 -‐2 -‐1 0 1 2 3 4 5
Leapfrog deriva-ve of Probe Sweep
Gaussian
Second Deriva>ve Deriva>ve
Figure 4.7: Current voltage derivatives - First and second current leapfrog derivative
(red and blue diamonds) of a simulation Langmuir probe sweep and Gaussian fit (black
line) with µ = -3.6 and σ = 0.9, corresponding to a plasma potential of 6.4 V at the probe
position
33
4. THEORETICAL MODEL PERFORMANCE
34
5
Comparison with probe sweep on
Rosetta
The final test of SPIS is of course the ability to produce simulation results that are con-
sistent with actual Rosetta measurements. We therefore analyse a Rosetta spacecraft
Langmuir probe 1 (LP1) measurement from 11/9 2009, where Rosetta is at approxi-
mately 1 AU, with SAA -10◦ in the Solar wind, by fitting the Rosetta results with the
theoretical model. Thereafter we simulate iteratively until we find the best fit to the
Rosetta result and plot the results in fig. 5.1.
The simulation result is adjusted in the photoelectron emission dominated region by
a factor of 1.04, accounting for the unknown instantaneous Solar UV spectrum, roughly
corresponding to a 4% increase in flux at the time of the Rosetta measurement. The
Solar UV spectrum is not static, as it is in our simulation, and we find a 4% offset in
the photosaturation current to be very reasonable.
We find the overall fit and shape of the result to be largely a success, with only minor
discrepancies at certain regions. The CV sweep is dominated by the photoelectron
current, and we only find a small contribution to the current from the plasma electrons,
and a negligible contribution from the ions, as expected.
The average photoelectron temperature, Tph governs the shape of the knee in the
photoemission current by the electrons emitted by the probe, as well as the slope of
the photoelectron cloud absorption. Unfortunately, SPIS does not yet allow for setting
the average photoelectron temperature, Tph, to be a material property and thus unique
for each material. Instead it is treated as a global constant throughout all simulation
35
5. COMPARISON WITH PROBE SWEEP ON ROSETTA
Figure 5.1: Probe sweep in SPIS and on Rosetta - Current vs Voltage Langmuir
probe sweep. Rosetta RPCLAP091109015SRDS18NS LP1 sweep at 1 AU (red circles) with
unknown plasma parameters and spacecraft potential. Three SPIS simulations for each
particle contribution, plasma electron, ion and photoelectron (purple, solid blue and green
diamonds respectively) and the total current (solid black) on LP1. The total simulation
result current in the photoelectron emission dominated region (from -10 V to 0 V) is
slightly adjusted by a factor of 1.04 to account for the unknown instantaneous Solar UV
spectrum at the time of the Rosetta measurements. Simulation parameters: 4 million
particle simulations for a +8 V charged spacecraft at 1 AU, in Te=12 eV, Tion=5 eV, Tph=
1.2 eV, ne = 5 cm−3 solar wind at v = 400 km/s.
surfaces. In reality, photoelectrons emitted by the Langmuir probe coated with TiN be-
haves differently than photoelectrons emitted from materials from the spacecraft body,
booms and solar arrays, including Tph as well as emission and absorption properties[7].
A lower Tph will make the knee sharper, but will also steepen the photoelectron
cloud absorption curve, as shown by eq. (4.1). This is also visible in our results, where
the knee is not as sharp as the Rosetta result and we cannot simulate with a lower Tph
without diverging wildly in the attractive potential region. Nevertheless, the overall fit
is quite good.
Another source of error is the probe resolution, where the surface area is only 92%
36
of the Rosetta probe. The influence of this error on the result is expected to generate
a very small offset in the photosaturation current, much in the same way as of the
unknown instantaneous Solar UV flux error source, thus already accounted for in our
photosaturation current factor of 1.04.
37
5. COMPARISON WITH PROBE SWEEP ON ROSETTA
38
6
Solar aspect angle dependence
Another important aspect is to model the solar aspect angle dependence of the Lang-
muir probe sweep results. As Rosetta rotates around its axis, LP1 and LP2 will occupy
regions of different photoelectron densities and potentials, which will have a large ef-
fect on the probe measurements. As each Langmuir probe is mounted on booms of
different lengths which are connected at dissimilar angles to the spacecraft surface, we
expect the dependence to behave differently for each probe. A schematic on the probe
configuration at a selection of important solar aspect angles is shown in fig. 6.1.
6.1 Plasma potential
We analysed the dependence by making Langmuir probe sweep simulations of pho-
toelectrons of each angle for both probes, and by fitting the resulting photoelectron
current to our model according to eq. (4.1) to obtain values for nph and Vplasma. For
each fit, a worst case fit was made by eye to estimate the error margin of each parameter.
For Langmuir probes in a few selected regions close (. 30◦) to the wake, with
the boom partially or fully shadowed, Model 3 fails to give precise results, as seen in
section 4.4. Therefore, a second estimate for the potential was made, using the Gaussian
fit to the secondary derivative of the current, detailed in section 4.6. However, we still
preserve the error margins from the worst fit routine.
The Vplasma result, as seen in fig. 6.2 was compared to a previous SPIS 3.7 simulation
result from Sjogren [14] in fig. 6.3 with identical plasma parameters. In this version,
the Langmuir probe booms are cuboid and the Langmuir probe themselves are missing.
39
6. SOLAR ASPECT ANGLE DEPENDENCE
Figure 6.1: Schematic of solar wind wake for LAP - Assuming radial flow from the
Sun, for various values of the solar aspect angles. The green box is the spacecraft body,
seen here from the +y direction in the s/c coordinate system, with the +x and +z faces
marked in the top left panel. The red thin rectangle indicates the solar panels, and the
black arrows show the probe positions on the booms. The sunlight and solar wind flow
direction are shown by a yellow arrow. Nominal wake edges are indicated by the blue lines.
Image and text courtesy of A. Eriksson.
Therefore, the plasma potential is measured at some region in space corresponding to
where the Langmuir probes should be, and plotted against solar aspect angle.
40
6.1 Plasma potential
Figure 6.2: SPIS 5 LP sweep resultt, Vplasma vs SAA - Plasma potential mea-
surements from Model 3 fits to 42 Langmuir probe sweeps at different SAA, each point
representing a unique 7 million particle simulation in SPIS 5.
Figure 6.3: SPIS 3.7 Result, Vplasma vs SAA. - Result from Sjogren [14], with an
older model, lacking Langmuir probes, and older version of the Langmuir probe booms. The
measurements are taken as manual measurement in a region in space where the Langmuir
Probe center should be and plotted against SAA angle.
41
6. SOLAR ASPECT ANGLE DEPENDENCE
We find large discrepancies in both the shape and amplitude of the SAA dependency
to the plasma potential result Even though, from fig. 3.3, we found the plasma density
profiles to be in agreement.
The wake effect appears to be much more prominent in our results, and we recover
the expected potential drop of about 0.5 V behind the solar array in the LP1 measure-
ment (around 60◦). In the LP2 measurement (around -45◦), we expect a larger potential
drop in the wake behind the spacecraft of about 1.5V, but we find the actual drop to
be 2-3 V. Note however that we have large uncertainties in these measurements in the
wake from the model, and they are not normally evaluated for the Langmuir probes
on Rosetta. This wake effect in Sjogren’s results however, is barely distinguishable and
much less than we would expect.
The SPIS 3.7 simulation result has failed to predict the solar aspect angle depen-
dence in the solar wind [8]. To challenge Sjogrens result, we need to compare our result
to real world measurements, and a comparison can be seen in fig. 6.4. The Earth 2 and
Lutetita fly-by measurements were selected because Rosetta is believed to have been
in a similar tenuous plasma as in our measurements.
The results in fig. 6.4 show an overall agreement in the shape of the potentials, and
a decreasing potential difference from SAA = -10◦ reaching zero at around 40◦. This
is an improvement to previous simulation results, where the potential difference goes
from 0 V to 0.5 V from 0◦ to 40◦[14][8]. However, the VPS values seem exaggerated by
a factor of 4. This is further discussed in chapter 7.
42
6.1 Plasma potential
(a) VPS measurements (top row) from Rosetta Langmuir probes at the Earth 2 fly-by (left)
and Lutetia fly-by (right), and the corresponding difference in potential between probe(bottom
row), from Johlander (2012)[8].
(b) VPS vs SAA. Result of SPIS simulation.VPS = Vplasma - VSC . VSC =
+10 V, Tph = 2 eV in nominal solar wind.
(c) The VPS difference between probes in the Rosetta simulations for VSC
= +10 V, Tph = 2 eV in nominal solar wind.
Figure 6.4: Comparison of potential measurements on Rosetta measurements in space
(a) and simulation results (b)(c)
43
6. SOLAR ASPECT ANGLE DEPENDENCE
6.2 Density profile
Another very important parameter is the angular density profile surrounding the space-
craft for each probe, seen in fig. 6.5. The nph result is calculated from the current of
the probe sweep from the model fitting, and is also adjusted for the surface area error
from the resolution of the probe. For the Earth 2 fly-by, where Rosetta was in tenuous
plasma in Earth’s magnetosphere, the Rosetta surrounding is believed to be dominated
by photoelectrons [8], and should as such be a good comparison to our simulation result.
In fig. 6.5 we find a very good agreement for the Earth 2 fly-by, and we accurately
detect the size of the density drop to the wake (-10◦ for LP2, and 40◦ for LP1) from
SAA 0◦ to be a factor of 3 to 4. Also, we find an excellent fit to the shape of the density
profile for both Langmuir probes.
The results are again off by a factor between 3 and 5 for each density profile, and the
most probable cause of this is of course the somewhat unrealistic Tph in our simulation,
as well as the potential of the spacecraft, VSC . Even though we have adjusted the result
to account for the smaller probe we have in SPIS, due to the limited resolution of the
sphere, we cannot disregard this as another small source of error.
44
6.2 Density profile
(a) (b)
(c)
Figure 6.5: Rosetta log density measurements from Johlander 2012[8], assuming Tph =
1 eV for LP1 (a) and LP2 (b). Measured log photoelectron density vs SAA from Rosetta
simulations for VSC = +10 V, Tph = 2 eV in nominal solar wind (c). The error margins
for the simulation result is often too small to be visible.
45
6. SOLAR ASPECT ANGLE DEPENDENCE
46
7
Conclusions
7.1 Theoretical model
We find that the photoelectron current model described by (eq. (4.1)) accurately de-
scribes simulation and Rosetta results for a probe inside the electrostatic potential of a
positively charged spacecraft. In regions where the Langmuir probe normally operates
on Rosetta (not in wake, or in partial shadow of spacecraft), the confidence of the re-
sults reported are high and consistently produces good fits to simulation results. The
model is also independently confirmed by the first peak of the secondary derivative of
the current to the probe.
We find a clear preference in the simulation result to a theoretical model with radial
photoemission from a point[7]. This suggests that in plasmas such as the solar wind, the
SPIS Rosetta Langmuir probe is sufficiently small for the emission to be approximated
as from a point source. The theoretical model of isotropical photoemission from a
plane consistently fails to produce a good fit to the isotropical photoemission from the
Langmuir probe in SPIS.
When the probe and probe boom is in partial shadow of the spacecraft, applying
the model is problematic, as the flow of electrons is not fully understood. Reasonably,
this is because the geometry of the solution becomes very different from the ideal OML
case of particles originating from a zero potential at infinity being absorbed by a probe
with a monotonic repelling or attracting potential. To further the understanding of
this region, no better tool can be suggested than further modelling in software such as
SPIS 5.
47
7. CONCLUSIONS
7.2 Rosetta Langmuir probe sweep
The best SPIS simulation fit for a Rosetta LP1 sweep from 9/11 2009 was found to
be a +8 V charged spacecraft in the solar wind of ne = 5 cm−3, Tion = 5 eV, Te =
12 eV, Tph = 1.2 eV, vSW = 400 km/s at 1 AU. The fit was applied with a scaling
of the saturation current by a factor of 1.04, with overall great success. Some minor
discrepancies near the knee and at attracting potentials are thought to be chiefly due
to an erroneous mean photoelectron temperature of the photoemitted electrons from
the probe, and needs further investigation.
Because the simulations modelled the S/C surface and solar arrays as being of the
same material as the Langmuir probes, it is plausible that the photoelectrons emitted
from these surfaces are not accurately described. The photoelectrons emitted from the
surfaces, which amount to the photoelectron cloud, are actually from a wide array of
materials with different properties in photon absorption, photoelectron emission and
average photoelectron temperature. This yields large uncertainties in the tail end of the
probe sweep, as the positively charged probe is absorbing particles from the surrounding
photoelectron cloud.
7.3 SAA dependence
The SAA dependence of Vplasma in the the new simulation model and software config-
uration differs largely from previous results by Sjogren[14], but is found to be a more
accurate description of the actual Rosetta spacecraft Langmuir probe sweep results
[8]. The discrepancy to the previous simulations are both in applying a more realis-
tic model, as well as how the measurement was made. We believe that the Sjogren
simulation result should be regarded as unrealistic as it seems to exaggerate the pho-
toelectron density contribution to the plasma potential, as well as underestimating the
wake effect.
When comparing densities and potentials to measurements at the second Earth fly-
by ([8]), we find the SAA dependency to be largely a success in predicting the shape
of the profile, including excellent predictions of the photoelectron density drop in the
wake of the spacecraft. However the absolute values of the measured Langmuir probe
potential difference differs by a factor of 4. For the density, we find our results off by
a factor between 3-5 for both probes. The main source for these errors is believed to
48
7.4 Future work
be the high and unrealistic average photoelectron temperature, as well as, of course,
the choice of plasma parameters. Other sources of errors included the potential of the
spacecraft in our simulation, and to a minor extent, the probe resolution.
7.4 Future work
To achieve a closer fit to Rosetta results, we need a more realistic electron temperature
profile and suggest simulating with the correct material properties of each surface in
the model. If the average photoelectron temperature is still not a material property in
future versions of SPIS, we suggest a compromise by defining a Maxwellian photoelec-
tron distribution with two or more peaks at different temperatures. SPIS is apparently
already capable of doing this, but the author has yet to figure out exactly how that
can be implemented.
Improving accuracy and confidence in SPIS results can be done by improving the
probe resolution, and reworking the model of the Langmuir probe sphere, to achieve a
better spherical approximation. To optimize efficiency without increasing the sampling
error with an increased probe resolution of some factor, we suggest reducing the number
of particles backtracked by each surface by the same factor.
Investigation of the solar aspect angle dependence can also be improved by simu-
lating with a more realistic photoelectron temperature and material properties.
The new model for photoelectron absorption from a cloud can possibly explain
some, or all, of the so called ”leakage current” which has been under debate[8]. By
trying to reproduce a measurement where this phenomenon is present, we might find
support for this hypothesis.
We finally suggest simulating different plasma environments, possibly even the
comet coma, which would be a very useful reference for analysis during Rosetta’s ap-
proach to the comet 67P/Churyumov-Gerasimenko and subsequent orbit insertion.
49
7. CONCLUSIONS
50
References
[1] Francis F. Chen. Introduction to Plasma Physics and Controlled Fusion,
Volume 1: Plasma Physics. 1975. 2, 3
[2] E. Engwall. Cold magnetospheric plasma flows: Properties and inter-
action with spacecraft. Licentiate Thesis, Department of Astronomy and Space
Physics, Uppsala University, March 2006. 7
[3] A. Eriksson. RPC-LAP: The Rosetta Langmuir Probe Instrument. Space
Science Reviews, 128(DOI: 10.1007/s11214-006-9003-3):729–744, 2007. 5
[4] A. Eriksson. RPC-LAP: The Langmuir probe instrument of the Rosetta
plasma consortium. Technical report, Swedish Institute of Space Physics, 2008.
5, 6
[5] ESA. Rosetta mission. Available from: http://www.esa.int/Our_
Activities/Space_Science/Rosetta_overview [cited June 2013]. 5
[6] D’Accolti et Al. The Solar Array Photovoltaic Assembly for the
ROSETTA Orbiter and Lander Spacecraft’s. Space Power, Proceedings of
the Sixth European Conference held 6-10 May, 2002 in Porto, Portugal. Edited by
A. Wilson. European Space Agency, ESA SP-502, 2002.,, page p.445, 2002. 15
[7] Rejean J.L. Grard. Properties of the Satellite Photoelectron Sheath
Derived from Photoemission Laboratory Measurements. Journal of Geo-
physical Research, 78(16):2885–2906, January 1973. 8, 9, 15, 36, 47
[8] A. Johlander. Photoemission on the Rosetta spacecraft. Bachelor thesis
in physics, Uppsala University, November 2012. 42, 43, 44, 45, 48, 49
51
REFERENCES
[9] J.-C. Mateo-Velez, P. Sarrailh, B. Thiebault, J. Forest, A. Hilgers,
J.-F. Roussel, G. Dufour, B. Riviere, V. Genot, S. Guillemant,
A. Eriksson, C. Cully, and D. Rodgers. SPIS Science: modelling space-
craft cleanliness for low-energy plasma measurement. 12th Spacecraft
Charging Technology Conf. Kitakyushu, Japan, May 2012. 13
[10] H. Mott-Smith and I. Langmuir. The theory of collectors in gaseous
discharges. Physical Review, 28:727–763, 1926. 5
[11] M. Nakamura. Space plasma environment at the adeos-ii anomaly.
Proceedings of the 9th International Spacecraft Charging Technology Conference
(SCTC-9), JAXA-SP-05-001E, 2002. 13
[12] J Oro, T Mills, and A Lazcano. Comets and life in the universe. Ad-
vances in Space Research, 15(3):81 – 90, 1995. Available from: http://www.
sciencedirect.com/science/article/pii/S0273117799800672. 1
[13] A. Sjogren. Modelling of Rosetta Langmuir Probe Measurements. Oc-
tober 2009. 2
[14] A. Sjogren. Simulation of Potential Measurements Around a Pho-
toemitting Spacecraft in a Flowing Plasma. IEEE Transactions on Plasma
Science, 40(4):1257, April 2012. 16, 21, 23, 39, 41, 42, 48
[15] SPINE. SPINE community Web site, June 2013. Available from: www.spine.
org. 14
[16] Benoıt Thiebault. SPIS-GEO User Manual. Artenum, Paris, esa-spisgeo-d7-
sum-2012-08-002 edition, August 2012. 14
[17] C. Weyde. Plasma Parameters from the Rosetta LAP instrument.
November 2006. 6, 8, 10, 25
[18] Wikipedia. Liouvilles theorem, 2013. Available from: http://en.wikipedia.
org/wiki/Liouvilles_theorem_(Hamiltonian) [cited June 2013]. 14
52