MUPUS – A THERMAL AND MECHANICAL PROPERTIES PROBE
FOR THE ROSETTA LANDER PHILAE
TILMAN SPOHN1,2,∗, KARSTEN SEIFERLIN3, AXEL HAGERMANN4,J¨
KNOLLENBERG2, ANDREW J. BALL4, MAREK BANASZKIEWICZ5, JOHANNES
BENKHOFF2,7, STANISLAW GADOMSKI5, WOJCIECH GREGORCZYK8, JERZY
GRYGORCZUK5, MAREK HLOND5,G¨
UNTER KARGL6, EKKEHARD K ¨
NORBERT K ¨
OMLE6, JACEK KRASOWSKI5, WOJCIECH MARCZEWSKI5
and JOHN C. ZARNECKI4
ur Planetologie, Westf¨
alische Wilhelms Universit¨
ur Planetenforschung, Deutsches Zentrum f¨
ur Luft- und Raumfahrt, Berlin, Germany
3Physikalisches Insitut, Universit¨
at Bern, Bern, Switzerland
4Planetary and Space Science Research Institute, CEPSAR, The Open University, Milton Keynes, UK
5Space Research Centre, Warsaw, Poland
ur Weltraumforschung, ¨
Osterreichische Akademie der Wissenschaften, Graz, Austria
7European Space Technology Centre, ESA, Noordwijk, The Netherlands
8Telecommunication Institute, PIT, Warsaw, Poland
(∗Author for correspondence: E-mail: Tilman.Spohn@dlr.de)
(Received 28 February 2006; Accepted in ﬁnal form 11 October 2006)
Abstract. MUPUS, the multi purpose sensor package onboard the Rosetta lander PHILAE, will mea-
sure the energy balance and the physical parameters in the near-surface layers – up to about 30 cm
depth- of the nucleus of Rosetta’s target comet Churyumov-Gerasimenko. Moreover it will monitor
changes in these parameters over time as the comet approaches the sun. Among the parameters studied
are the density, the porosity, cohesion, the thermal diffusivity and conductivity, and temperature. The
data should increase our knowledge of how comets work, and how the coma gases form. The data may
also be used to constrain the microstructure of the nucleus material. Changes with time of physical
properties will reveal timescales and possibly the nature of processes that modify the material close
to the surface. Thereby, the data will indicate how pristine cometary matter sampled and analysed by
other experiments on PHILAE really is.
Keywords: rosetta, comets, surface, heat ﬂow
1. Introduction and Scientiﬁc Goals
Rosetta was sucessfully launched on 2 March 2004 and is expected to start its
rendevous with Comet Churyumov-Gerasimenko in May 2014. The Rosetta Lander
PHILAE will be the ﬁrst spacecraft to make a soft landing on a comet nucleus.
Scientiﬁc observations are to be carried out for a minimum of one week, but might
continue for several months as the comet approaches perihelion. Rosetta’s target
comet Churyumov-Gerasimenko has a period of ∼6.6 years. its nucleus, with an
estimated size of 3×5 km, is expected to have a rotation period of approx 12 hours.
Space Science Reviews (2007) 128: 339–362
DOI: 10.1007/s11214-006-9081-2 C
340 T. SPOHN ET AL.
The physics of comets involves the production of the coma by the sublimation of
ices at or close to the surface of the nucleus. The rates of production of coma gases
depend on the energy balance at the surface and in a boundary layer underneath
the surface into which heat is transferred by conduction and vapour transport. The
surface energy balance is
RiHi=S−εσ T4−(q+qv) (1)
where Riare the sublimation rates of the coma gases (species i), Hiis the enthalpy
of sublimation of species i(approx. 2.8×106Jkg−1for water ice), Sis the insolation
corrected for the surface albedo, εis the surface emissivity (slightly less than 1.0
for a ‘dark’ comet), σthe Stefan-Boltzmann constant, Tis temperature and qand
qvare the conductive heat ﬂux and the heat ﬂow associated with the vapour ﬂow
into or out of the interior, respectively. Temperatures at a dust-covered surface can
exceed the temperature of a sublimating water ice surface (∼200 K) by more than
100 K near 1AU. The vapour ﬂux into the interior is driven by the gradient in
vapour density that forms in response to the temperature gradient during cometary
day time. The gas ﬂow is likely to be in the Knudsen regime. During night time the
ﬂow may be reversed.
The energy balance in the porous interior of the nucleus reads
where ρand care the density and speciﬁc heat of the nucleus ice, ρi,ci, and ui
are the densities, speciﬁc heats (at constant volume) and ﬂow velocity of vapour
species iin the interior. Since the vapour pressures are functions of temperature,
the ratio between the two heat transport terms on the right-hand-side (RHS) of
(2) depends on the temperature, the enthalpy of sublimation and on the thermal
diffusivity and permeability of the ice. The thermal properties of the ice are not
precisely known. Depending on pore volume and structure in the ice, thermal con-
ductivity could be 2 or more orders of magnitude lower than solid ice, i.e. 568/T W
m−1K−1(Klinger, 1981), so that the vapour could contribute considerably to the to-
tal energy transport. Smoluchowski (1982) was the ﬁrst to point to the importance
of heat transfer via the vapour phase. It has been demonstrated experimentally
through the KOSI (comet simulation) experiments (e.g., Gr¨un et al., 1991) that
heat transfer via vapour was important at temperatures above about 200 K for
water ice (Spohn and Benkhoff, 1990; Steiner, 1990; Benkhoff and Spohn, 1991;
Espinasse, 1991; Steiner and K¨omle, 1991; Steiner et al., 1991; Benkhoff et al.
1995). Spohn and Benkhoff (1990) have outlined a porous medium heat and mass
transfer theory to describe the effects. Benkhoff and Huebner (1995) and Hueb-
ner and Benkhoff (1997,1999) expanded the model and applied it to cometary
MUPUS (’Multi-Purpose Sensors for Surface and Sub-Surface Science’) origi-
nated from the proposal to the then RoLand comet lander by Spohn et al. (1995),
MUPUS – A THERMAL AND MECHANICAL PROPERTIES PROBE FOR THE ROSETTA 341
with a related experiment (’SuSI’) also having been proposed for the (ultimately
cancelled) Champollion lander (Benkhoff et al., 1995). The central part of MUPUS
is a thermal probe that, after insertion into the regolith of the comet nucleus up to a
depth of 32 cm, aims to measure both the temperature proﬁle underneath the surface
and the thermal diffusivity and conductivity proﬁles. The thermal diffusivity and
conductivity will be derived from the time rates of change of sensor temperatures
during active heating cycles. In addition, MUPUS will measure the surface temper-
ature and the mechanical strength of the nucleus material. The temperatures and
the thermal transport parameters will be measured regularly over the life time of
the lander. The mechanical strength of the material will be derived from the energy
spent per unit distance penetrated during the hammered insertion. Proﬁles of pene-
tration resistance will also be derived from MUPUS accelerometry measurements
in each of the lander’s two harpoon anchors. The anchors also contain MUPUS
In doing these measurements, MUPUS can contribute to an assessment of the
energy balance of the comet nucleus and the physical properties of its material.
Since the thermal conductivity and diffusivity are strong functions of the structure
of the porous ice and the degree to which it has been sintered (Seiferlin et al.,
1995), MUPUS can constrain the microstructure and the degree to which the comet
material has been thermally altered. MUPUS will thus predict at what depth pris-
tine material can be expected. This is important for characterising the context of
the samples extracted by the drill, as well as for our understanding of the thermo-
physics of cometary activity. The scientiﬁc objectives of MUPUS are summarised as
−To understand the properties and layering of the near-surface matter as these
evolve with time as the comet nucleus spins and approaches the Sun.
−To understand the energy balance at the surface and its variation with time
−To understand the mass ﬂow at the surface and its evolution with time.
−To provide ground truth for thermal mapping from the Orbiter, and to support
other instruments on the Rosetta Lander (e.g. SESAME-CASSE).”
2. Instrument Description
The MUPUS package consists of three major parts, the penetrator MUPUS PEN
with ist subsystems, the radiometer MUPUS TM, and the anchor sensors MUPUS
ANC. Their positions on the lander are coloured red in Figure 1. The MUPUS main
electronics are integrated into the Common Electronics Box on the lander, together
with the main electronics of other experiments and lander subsystems.
342 T. SPOHN ET AL.
Figure 1. MUPUS instrument package conﬁguration overview. The penetrator is placed on the sur-
face away from the lander by a deployment arm (not shown). Both anchors of the lander harbour
a temperature sensor (ANC-T) and an accelerometer (ANC-M). The 4-channel IR sensor TM is
mounted near the top of the lander housing. The penetrator is equipped with a depth sensor (PEN-M),
thermal sensors that measure the temperature proﬁle (TP) and a thermal diffusivity / conductivity
proﬁle (THC). The hammering device and the front-end electronics are mounted in the housing (thick
cylinder) on top of the penetrator tube (thin cylinder).
2.1. THE PENETRATOR MUPUS PEN
The main part and the most complex instrument in the MUPUS experiment suite
is the thermal probe MUPUS PEN (Figure 2). A ﬁbre compound tube with a
metal tip will be inserted into the ground about one meter away from the lander
by a deployment device and a hammer mechanism (Seiferlin et al., 2002). The
hammer mechanism is accommodated together with heated front end electronics
in a gold plated cylinder housing at the top of the penetrator and will remain above
the surface. The total length of the probe is restricted by the vertical height of the
lander on the balcony of which MUPUS PEN was required to ﬁt. The hammer and
electronics housing is about 10 cm high, leaving 33 cm for the tube and 3 cm for
the tip. The length of the tube is several times the thermal skin depth for rotation
rates of the nucleus between 10 and 100 h and thermal conductivities less than that
of compact ice. We consider a skin depth of only a few cm as most likely.
Until its descope in mid-2001, MUPUS PEN also incorporated a gamma ray den-
sitometer (Ball et al., 2001). It was designed to measure the attenuation of gamma
rays emitted by a 137Cs source in the tip of the penetrator. During penetration the
MUPUS – A THERMAL AND MECHANICAL PROPERTIES PROBE FOR THE ROSETTA 343
Figure 2. MUPUS PEN: The right image shows the MUPUS PEN, mounted on the lander (launch
and cruise phase conﬁguration). A stepper motor (lower centre, between the two black support tubes)
will uncoil two metal proﬁles (brass-coloured) from their stored position on two spools (left and
right of the stepper motor) and thus deploy the penetrator to a distance of about 1 m from the lander.
A hammer (in the gold-plated cylinder on top) will insert the probe (thin, brownish tube) into the
regolith. The tube carries 16 thermal sensors that can be used to measure the temperature. These
sensors can also be used to heat the surrounding comet regolith (2 such sensors are shown in the
left panel) for a measurement of the thermal conductivity. A densitometer photon source and two
densitometer detectors (top, next to the hammer housing) were planned to be integrated (left image).
The densitometer had to be descoped because of the lack of funding.
344 T. SPOHN ET AL.
changing ﬂux of transmitted gamma rays would have been measured by two semi-
conductor detectors mounted on opposite sides of the front-end electronics. The
sensor was descoped because of technical problems arising ultimately from fund-
ing difﬁculties in the early stages of the project. Unfortunately, the vertical proﬁle
of bulk density will now have to be inferred by other means. For example, a bulk
density proﬁle might be retrievable via microstructural models of the sub-surface
material, constrained by other physical properties measurements as well as compo-
sitional information. Such measurements include: mechanical properties measure-
ments made by the MUPUS thermal probe during hammering, anchor deployment
and sampling drill operation; thermal diffusivity measurements made by the MU-
PUS thermal probe; permittivity measurements made by the SESAME-PP exper-
iment; acoustic wave propagation measurements made by the SESAME-CASSE
experiment (Seidensticker et al., this volume); surface thermal inertia measure-
ments made nearby by MUPUS-TM, and compositional measurements made by
the Philae geochemical experiments. Larger-scale bulk density can also be assessed
by the orbiter radio science experiment and by the CONSERT radio transmission
tomography experiment (Kofman et al., this issue).
2.1.1. The Thermal Probe
The tube has a radius of 10 mm and a wall thickness of 1 mm. The mantle of
the hollow tube is made of cyanato-ester with ﬁbreglass, which has a thermal
conductivity of 0.5 Wm−1K−1. This material choice provides a good compromise
between the structural requirement of being stiff enough to penetrate the surface,
and the thermal requirement to minimize vertical heat ﬂow along the probe. A more
important factor for this heat pipe effect is, however, not the probe itself but the
sensors and their conductors which are made of Titanium and copper, respectively.
Though they are only a few micrometres thick, their total conductance is about
twice that of the ﬁbre compound tube. The heat pipe effect of the probe can be
illustrated by a comparison of the total conductance of the probe with estimates
of that of the cometary material that occupied the volume within the tube before
insertion. For a thermal conductivity of the replaced porous cometary material of
0.1 Wm−1K−1the probe conducts about 5 times as much heat as the material it
replaced. In addition, the thermal effect of the zone of compacted material around
the inserted probe needs to be taken into account.
The effect of a thermal ’short circuit’ through the probe is minimised by using
a thin, hollow probe of a low-conductivity material, but signiﬁcant enough that it
needs to taken into account in the data analysis. However, remaining perturbations
can be reduced by an inversion-type data evaluation method that was developed by
the MUPUS team (Hagermann, 1999; Hagermann and Spohn, 1999). This method
makes use of the damping and lagging effect which temperature signals from the
surface experience as they penetrate to greater depths. More details about this
method can be found in the appendix.
MUPUS – A THERMAL AND MECHANICAL PROPERTIES PROBE FOR THE ROSETTA 345
Figure 3. A new type of thermal sensor has been developed by the MUPUS team. A 20 μm thin
layer of titanium on a Kapton substrate is used to measure the temperature (electrical resistance is
proportional to the temperature). The same titanium cell can be actively heated in order to perform
thermal diffusivity and conductivity measurements (upper panel). 16 such cells and all required
electrical connections (thin lines) are laser-sputtered on one Kapton sheet (lower panel). The cell
dimensions grow from top (left in image) to bottom. The Kapton sheet is rolled and glued to the inner
tube wall of MUPUS PEN.
2.1.2. The Thermal Sensors
The PEN tube carries a Kapton sheet that is glued to its inner wall onto which the
thermistors are vapour deposited. The technology used for manufacturing the ther-
mal sensors was developed by the MUPUS team for the experiment. It is described
in Gregorczyk et al. (1999).
The Kapton sheet is glued to the inner wall of the tube in order to protect the
thermal sensors from mechanical strain upon insertion. There are sixteen meander-
shaped titanium sensors with temperature-dependent electrical resistance (see Fig-
ure 3). The number of sensors matches the number of independent data channels
of the electronics. The depth intervals covered by these sensors increase from the
top to the tip of the tube, from 1 cm to 4 cm length. This conﬁguration has been
chosen in order to allow a denser coverage of the thermal proﬁle near the surface
where the temperature gradient is expected to be steeper than at greater depth. The
resistance of the titanium is a function of temperature similar to the temperature
dependence of the resistance of PT100 type sensors. PT100 sensors and their use for
temperature measurements are standardized under the IEC 751 norm, its European
counterpart EN 60751 and the German DIN 60751.
The temperature dependence is given by
R(T)=R0(1 −s1·T+s2·T2) (3)
346 T. SPOHN ET AL.
where R0is a reference resistance of 100 Ohm at a reference temperature of 275.73
K, Tis the temperature difference to the reference temperature. For a PT100 and
other platin-based sensors, s1in eq. 3 is 3.9083×10−3K−1, and s2=−5.775×10−7
The linear coefﬁcient s1for other bulk metals is typically in the order of 4×10−3
K−1, while s2is typically two or more orders of magnitude smaller. The relation
holds for a temperature interval of 110 K to 375 K. For the MUPUS-type Titanium
sensors, s1and s2have to be determined individually for each sensor, because tab-
ulated values for bulk metals cannot be applied to thin ﬁlms of vacuum-deposited
metal. s1was found to be about 2×10−3K−1with small variations between individ-
ual sensors, which is about half the value of that for PT100. After vacuum-deposition
on the Kapton sheet, the sensors have been tempered and cured. This procedure
has been repeated after integration of the Kapton sheet into the hollow tube. The
long-term stability of the sensor’s characteristics is unknown. In-ﬂight calibration
will be required.
The 16 sensors thus allow measuring a temperature proﬁle that extends over the
length of the tube (32 cm). Because of the small mass of the sensors the reaction
time of the sensors to changes in temperature is small, typically a few seconds.
The speciﬁc thermal timescale for a sensor being separated from the medium by 1
mm (i.e. the thickness of the probe wall) of a material with a thermal diffusivity of
about 10−6m2s−1is one second. The surface (proportional to heat ﬂux) to volume
(proportional to total heat capacity) ratio of a MUPUS-type sensor is 100 times
better than that for a conventional ceramics-sealed PT100, and reacts to changes in
temperature faster by about the same factor. The sensitivity of the bare sensors is
slightly worse than that of PT100 standard sensors because of the difference in s1
in Equation (3). The effective resolution of the ﬂight instrument is limited by the
performance of the 16 bit AD converter to about 12 meaningful bits, covering about
200 K, which corresponds to about 0.05 K (see also Marczewski et al., 2004). The
temperatures are measured by applying a constant current of 20 mA and measuring
the voltage drop across the resistors. All PEN sensors are connected to the current
source through one shared conductor and individual sensor wires inside the probe
structure, and two external PEN cables. The number of wires is limited to 18 in order
to minimize undesired heat losses along the cables. Several options are available
for the measurement sequence, but a default operation is deﬁned for long term
operations, consisting of a temperature scan every 20 sec.
2.1.3. Expected Performance
Initial tests of the probe performance were obtained by heating individual sensors
in vacuum with background temperatures between -160 C and -100◦C as well
as under ambient conditions with the PEN probe immersed in different media.
These tests showed that the inﬂuence of the heating on the neighbouring sensors is
moderate and that mainly the immediate neighbours are affected. The temperature
increase in immediately adjacent sensors will depend on the thermal conductivity
MUPUS – A THERMAL AND MECHANICAL PROPERTIES PROBE FOR THE ROSETTA 347
of the nucleus surface layer but is expected to be relatively small. Under worst
conditions in vacuum we measured a temperature rise of the neighbour sensor by
about 50% of the temperature of the heated sensor depending to some extent on the
ambient temperature. In solid ice the increase was 5%, in snow and sand 10–15%
and in a Teﬂon cylinder with a thermal conductivity of 0.25 W/m K about 10%.
Additional tests with a model of the MUPUS penetrator in terrestrial soil showed
that the penetrator was more sensitive to weak energy ﬂuxes than the commonly
used method of heat ﬂux plates (Marczewski et al., 2004). These authors have
reported on a series of test experiments with the MUPUS thermal probe.
In order to demonstrate and illustrate the performance of MUPUS PEN mea-
suring the temperature proﬁle of the uppermost layers of a comet, we studied two
different data sets that are related to cometary thermal evolution, and come from
two quite different sources:
1. In recent years, the International Space Science Institute ISSI, located in Bern,
gathered comet modellers (one of them was J. Benkhoff, co-author of this paper)
in a group and invited this group several times to workshops. One goal of this
4-year program, coordinated by W. Huebner, was to compare individual model
codes, improve them and ﬁnally develop a set of standard thermophysical comet
models. The used model code is a 1-dimensional, multi-component (e.g., water,
CO) ﬁnite difference code which solves a set of coupled mass- and heat diffusion
differential equation, thus including heat transport by the vapour phase. We
selected a model comet with the following key parameters: (a) A spherical
model comet in an orbit of the Rosetta target comet P/Wirtanen, (b) the spin axis
is assumed to be perpendicular to the orbital plane, (c) porosity is simulated by a
pipe network (pore radius 10mm), (d) the nucleus consists of water ice, several
minor volatile components, and dust, (e) molecular ﬂux in the pores, (f) low heat
conduction (Hertz factor 0.001).
The model calculations were carried out as follows: a homogeneously mixed
body at a constant initial temperature (T =20K) and a constant mass den-
sity distribution is considered. Due to heating of the body and sublimation of
the volatile components, the initially homogeneous body differentiates into a
multi-layer body (if it contains more than one volatile component), where the
deepest layer has the original composition. The subsurface temperatures are cal-
culated every 15 Minutes for several orbits. The underlying model is described
in Benkhoff (2002) and Prialnik et al. (2004).
2. From ca. 1987 to 1992, the German Research Foundation (DFG) sponsored a
research program in which several German and International teams cooperated
in oder to simulate comets, (see Gr¨un et al., 1991, for example) or at least some
physical processes with relevance to comets, in a large space simulator – basically
a vacuum chamber equipped with a LN2 cooling facility and an arrangement
of Xenon lamps to simulate solar insolation. In each of the KOSI (“Kometen-
Simulation” =Comet Simulation) experiments, a sample of porous ice-dust
348 T. SPOHN ET AL.
mixtures was ﬁlled into a sample container of typically 30 cm diameter and 15 cm
depth, then cooled down in vacuum and ﬁnally exposed to artiﬁcal sunlight. We
(Benkhoff, Spohn and Seiferlin) were responsible for the thermal measurements
during these experiments. The selected KOSI-9 test was special in one respect:
there were 3 subsequent insolation periods of more or less similar intensity
proﬁle (the ﬁrst one a bit shorter but all with the same intensity proﬁle) and a
total experiment duration of about 70 h, in order to simulate natural day/night
cycles. Differences between the temperature proﬁles for the 3 phases can be
explained by texture modiﬁcation like sintering and recrystallization of water
ice, which would be more effective in warm layers. This KOSI 9 experiment was
described by Seiferlin et al. (1995).
Temperature proﬁles from both sources were processed in the same way to
simulate a measurement done with MUPUS:
1. The temperature proﬁles, as they were available, contained fewer data points in
z-direction than MUPUS would measure. The proﬁles were therefore projected
onto an array of sufﬁcient length, and gaps between existing data points were
ﬁlled by a polynomial ﬁt. In the time domain, an artiﬁcial equidistant data set
with sufﬁcient resolution was generated by interpolating between existing time
steps (i.e. subsequent data points).
2. Because the individual MUPUS sensors are between 10 and 40 mm long, they
record the average temperature along their length. The ﬁtted temperature proﬁle
obtained in step 1 was projected onto the geometry of the MUPUS sensors in
such a way that each sensor was assigned a temperature equivalent to the average
temperature of the stretch covered by the sensor considered. After this step, the
temperature proﬁle consisted of 16 temperature values, just as they would be
recorded by MUPUS.
3. In this step, the quality of the data was degraded and limited to 10 bit resolution,
which is slightly less than provided by the MUPUS ﬂight electronics.
4. Moderate random noise was then added to simulate random errors such as may
be caused by varying thermal contact to the medium, calibration uncertainties,
electromagnetic noise etc.
5. The temperature proﬁle as a function of time was then converted into a colour
coded image. The colour palette contains only 256 colours corresponding to
8 bit resolution. Thus, the resolution of the data as represented by the colour
maps is degraded in comparison with the expected results from MUPUS. The
simulated time step is 5 minutes according to the the nominal measurement cycle
Figure 4 shows a simulated measurement of the comet temperature history as was
proposed by the ISSI working group. Figure 5 shows the KOSI 9 data set. Because
the KOSI 9 record covered only 15 cm depth, the simulated MUPUS measurement
was also truncated at 15 cm. The results suggest that if the ISSI and the KOSI results
MUPUS – A THERMAL AND MECHANICAL PROPERTIES PROBE FOR THE ROSETTA 349
Figure 4. Simulated measurement with MUPUS. Temperatures are color coded in Kelvin. Time is
horizontal, depth is vertical. The time step (i.e. measurement interval and pixel resolution in the
image) is 5 minutes. The maximum depth is 32 cm. Black horizontal lines indicate sensor edges.
The temperature data processed for this image were taken from a numerical model of comet nucleus
temperatures developed in the framework of an ISSIworking group, and were provided by J. Benkhoff.
See text for further description.
are representative of the near surface temperature proﬁles in a cometary nucleus,
the MUPUS probe should be suitable to record it. The total penetration depth and
the number and spacing of sensors seems appropriate and, considering the limited
resources on the Rosetta Lander PHILAE, satisfactory.
2.1.4. Transient Thermal Properties Measurements
The titanium resistor cells may also be heated by applying electrical power to them
of up to 1 W at 12 V. The temperature increase caused by the heating is a function of
the known power and the thermal diffusivity (resp. conductivity) of the material that
surrounds the heated cell(s). In a typical measuring cycle, the heating is applied for a
350 T. SPOHN ET AL.
Figure 5. Simulated measurement with MUPUS. Temperatures are color coded in Kelvin. Time is
horizontal, depth is vertical. The time step (i.e. measurement interval and pixel resolution in the image)
is 5 minutes, total duration is 70 hours. The maximum depth is 15 cm. Black horizontal lines indicate
sensor edges. The temperature data processed for this image were taken from the results of the KOSI
9 experiment (see text for further detail). Three day/night cycles are visible. The discontinuity at the
afternoon of the third day is caused by a data gap in the original data set.
given time interval (5 minutes under standard conditions) and is then interrupted for
a temperature measurement for a few milliseconds. The temperature measurement
thus affects the heating only insigniﬁcantly. After switching off the heating power,
the temperature relaxation can additionally be measured.
The heating power is controlled by the ﬁlling ratio of the pulses, varying from
zero to the maximal available power (approx. 1 W) with 12-bit resolution. Under
standard operational conditions, heating will use up to 1/4th of the maximal avail-
able power (approx. 0.2–0.3 W). Heating will be applied to one or more sensors
simultaneously. For a measuring a conductivity depth proﬁle, heating will be ap-
plied consecutively to sensors with increasing depth. The heating and the associated
temperature measurements can be repeated every hour.
The evaluation of the data to obtain diffusivity and conductivity proﬁles is not
straightforward, but Banaszkiewicz et al. (1997) have derived appropriate mathe-
2.1.5. The Hammer
The difﬁcult task to emplace a sub-surface probe into a medium of unknown hard-
ness (but not harder than ca. 2 MPa) is performed by a mechanical hammer, es-
pecially designed for MUPUS. The hammer works like a mechanical diode. A
conventional 22 μF capacitor is charged to several 100 V. The stored electrical
energy is discharged through a coil that generates a strong magnetic ﬁeld. A small
mass (30 g) is accelerated by the magnetic ﬁeld into the opening inside the coil and
hits the penetrator tube with its maximum speed of ∼8 m/s, thus causing a powerful
hammer stroke onto the tube. The friction exerted on the tip and the tube and some
weak force extended by the deployment device take most of the rebound following
the hammer stroke. Together, hammer stroke and rebound absorption cause a net
forward movement of the penetrator into the comet nucleus regolith. Both the tube
and the hammer mass are connected by weak springs to the hammer housing. These
springs act to bring both back to their starting positions and prepare the motor for
MUPUS – A THERMAL AND MECHANICAL PROPERTIES PROBE FOR THE ROSETTA 351
the next hammer stroke. The energy of the hammer strokes can be adjusted to the
requirements set by the hardness of the nucleus material .
The probe also includes a sensor (PEN-M) to monitor the vertical displacement
of the probe as it is inserted. It starts close to the tip and slides up the probe; the
displacement is sensed electrically in the manner of a potentiometer. The assembly
carrying the displacement sensor also houses an electromagnet to hold the penetrator
and one of the electrodes of the SESAME-PP (Permittivity Probe) experiment
(Seidensticker et al., this issue).
Compared to alternative insertion methods, e.g. a pyro, the mechanism has
−the total energy is only limited by the power supply offered by the hosting
spacecraft, while a pyro has a maximum stored energy deﬁned at design time,
−insertion is done in small steps and may be interrupted to take measurements
once a given depth is reached,
−the advancement of the penetrator is small enough such that the insertion efﬁ-
ciency (depth reached per dissipated electrical energy) can be measured. From
these data a cohesion proﬁle of the penetrated layers can be derived.
Whenever erosion of the cometary surface material re-exposes parts of the pene-
trator, the hammer can be restarted to compensate for the material loss at the top by
inserting the penetrator accordingly. The expected surface erosion may well reach
1 m per comet orbit. This is about 3 times the total length of the penetrator tube.
2.1.6. Front-end electronics
To keep the thermal losses of the lander small and to reduce the harness between
the penetrator and the lander, a front-end electronics package has been included
in the hammer housing. The front-end electronics package communicates with the
electronics on board the lander via digital signals only, using a serial interface. The
cable connection between the lander and the front-end electronics is routed through
the central rotation axis of the lander, below the baseplate. This conﬁguration
ensures that the lander can rotate after PEN deployment without being locked by
a direct, tense cable connection to the balcony, for example. The design of the
front-end electronics proved to be very demanding, since it has to operate in a
low temperature environment, while the small size and mass of the housing make
an effective thermal design difﬁcult. Figure 6 shows the layout of the MUPUS
electronics. A detailed thermal model of the Front-End electronics can be found in
Seweryn et al. (2005).
2.1.7. The Deployment Device
Measuring the nucleus energy balance in the near-surface layers requires a thermal
probe that will be placed far enough away from the lander’s shadow. This require-
ment made it necessary to design and develop a complex mechanical deployment
device for the MUPUS PEN. The deployment device (compare Figures 1, 2 and
352 T. SPOHN ET AL.
Figure 6. Layout of the MUPUS electronics and its integration into the Philae lander. MUPUS’s main
electronics components are the internal electronic box and the PEN front-end electronics.
7) will place the penetrator normal to the ground about1mawayfrom the lan-
der. The value of1mwasdetermined from model calculations that investigated
the subsurface temperature ﬁeld underneath a lander on a slowly spinning comet
nucleus. An example of the results is shown in Figure 8. Because the lander is
intended to rotate after MUPUS PEN deployment, the deployment device must be
separated from the penetrator and retracted after deployment. A cable then provides
communications and power to the penetrator while the lander can rotate freely. The
design solution solves both problems efﬁciently: two spools hold one metal strip
each that is coiled ﬂat in its stored conﬁguration. For deployment, a stepper motor
pulls these approximately one meter long metal proﬁles forward from the spool
MUPUS – A THERMAL AND MECHANICAL PROPERTIES PROBE FOR THE ROSETTA 353
Figure 7. The MUPUS PEN and deployment device during a deployment test.
and the metal proﬁles attain their naturally bent shape. Their C-like cross section
provides sufﬁcient stiffness to hold the penetrator in the low-gravity environment
and supports the PEN driving, which is especially important at the initial insertion
phase. After PEN insertion, the PEN is released from the deployment device and
the metal strips retracted onto the spools.
2.2. THE THERMAL MAPPER
The MUPUS Thermal Mapper TM (see Figure 9) is an infrared radiometer that
consists of a set of 4 IR (thermopile-type) sensors designed to measure the brightness
temperature at the very surface, averaged over its ﬁeld of view. In its normal mode,
MUPUS TM will have the MUPUS penetrator in the center of its ﬁeld of view,
thereby adding an important data point to the temperature proﬁle. Fragile, ﬂuffy
layers such as a dust mantle on the cometary surface with low thermal conductivity
may cause a strong temperature gradient with depth and a signiﬁcant temperature
drop within a few mm below the surface. These ﬁrst few mm cannot be resolved with
the thermal sensors on MUPUS PEN even where the sensors are closely spaced. In
354 T. SPOHN ET AL.
Figure 8. Temperature ﬁeld as a function of radius and depth underneath a circular lander on a slowly
rotating nucleus (20h period) made of porous water ice whose matrix is 20 times less heat conductive
than compact ice. Representing a worst case, this model shows that the temperature proﬁle underneath
or even near the lander is not representative of the undisturbed proﬁle. Tecorresponds to sunrise.
Solid isotherms indicate the deviation in Kelvin from the temperature proﬁle without a lander shadow.
addition to the surface temperature, a direct estimate of the thermal conductivity of
the nucleus surface layer can be derived from the temperature measurements of TM
and the uppermost temperature sensor of the penetrator. Furthermore, the thermal
inertia of the nucleus surface at the landing site can be determined from an analysis
of the TM data.
2.3. ANCHOR SENSORS
The third component of MUPUS is the set of sensors implemented in PHILAE’Stwo
anchors MUPUS ANC (compare Figure 1). The anchors have been described by
Thiel et al. (2001, 2003). Two types of sensors have been implemented:
MUPUS – A THERMAL AND MECHANICAL PROPERTIES PROBE FOR THE ROSETTA 355
Figure 9. MUPUS TM: 4 thermopile-type IR sensors and a small front-end electronics package are
mounted in a small housing ﬁxed to a diagonal strut above the lander balcony. The box is tilted such
that the MUPUS penetrator will be in the ﬁeld of view of the sensors if the lander balcony is pointing
in the right direction. TM will thus provide the temperature at the very surface at the PEN location.
rANC-T, a PT100 temperature sensor to monitor the local temperature at the ﬁnal
resting place of the anchor after it has been shot into the regolith of the comet
nucleus. ANC-T provides an additional location for temperature measurement,
laterally displaced by approximately 1 m from the MUPUS PEN sensors, and
probably also at a greater depth. (It is likely that the deepest in situ measurements
by PHILAE will be those of the anchor sensors.) A sufﬁciently high sampling rate
in the ﬁrst few minutes after deployment for ANC-T may yield constraints on the
thermal diffusivity and conductivity of the cometary material (Paton, 2005).
rANC-M, a miniature shock accelerometer (ISOTRON 2255B-1 from Endevco)
monitoring the acceleration and deceleration of the anchoring projectile while it
is ﬁred into the ground.
The anchors will be accelerated pyrotechnically within a few milliseconds at the
timeimmediatelybeforelandertouchdownto reach a launch speed of approximately
∼100 m s−1. A peak acceleration value of about 90000 m s−2(i.e. ca. 9000 g) is
356 T. SPOHN ET AL.
Figure 10. Example of an anchor test shot. The target sample is a sintered CO2ice crust with softer
material underneath. The shot holes and the anchor cables are visible.
expected. After a very short free ﬂight period the anchor will penetrate the regolith
and be decelerated by the resistance of the nucleus material. The main challenge
concerning the selection of an appropriate accelerometer was the high dynamic
range that it has to cover. On the one hand, ANC-M will have to survive the high
amplitude acceleration phase and on the other hand it will have to be able to resolve a
much smaller amplitude deceleration signal expected when the anchor will penetrate
soft porous ice layers with low cohesion. With the 14-bit resolution of the ﬂight unit
and a sampling frequency of ∼50 kHz an accuracy of about 1.5 g can be achieved.
The chosen sensor type thereby will allow to record the full acceleration phase
(which is important for calibration purposes) while still giving a reasonable signal
for sintered porous ices. Judging from comet analogue samples, a sintered water
ice crust is to be expected for the nucleus of Comet Churyumov-Gerasimenko.
2.3.1. Dynamic Penetrometry Tests with MUPUS ANC
Penetrometry tests have been performed to check the performance of the anchoring
harpoon and to calibrate the dynamic strength measurements. We show results of a
shot into a CO2ice sample in Figures 10 and 11. Two shot holes are visible together
with parts of the anchoring cables in Figures 10. At the impact sites, small craters
with diameters larger than those of the penetration channels are visible. Note that
the projectiles were accelerated with a cold gas system. The maximum acceleration
in this experiment was lower than the acceleration expected for the PHILAE anchors.
In Figure 11 the deceleration proﬁle measured by the shock accelerometer in one
of the anchors is displayed together with a ﬁt calculated with the similarity model
described by K¨omle et al. (2001). To calibrate the data in terms of strength/cohesion,
a quasi-static strength measurement was performed close to the impact site. This
measurement consisted of a slow (40 mm s−1) penetration of a cylindrical rod
MUPUS – A THERMAL AND MECHANICAL PROPERTIES PROBE FOR THE ROSETTA 357
Figure 11. Upper panel: Accelerometer signal recorded from the impact close to the centre of the
target. Lower panel: Strength proﬁle obtained by an independent quasi-static measurement with a
conventional force cell.
with a 60◦tip into the sample, whereby the resistance force was measured by a
conventional load cell that was mounted at the top of the rod. The results of the
quasi-static measurements are shown in the lower panel, together with a simple
ﬁt to obtain a strength proﬁle. The main vertical structure of the layer seen in
the quasi-static test, namely a strong surface crust and a slightly consolidated part
below, is also well represented in the deceleration data. This demonstrates that
the dynamic penetrometry method, together with an appropriate model of the tip
geometry, should be able to detect at least variations and discontinuities in the
vertical strength proﬁle. More detailed discussions of the penetrometry tests have
been discussed by Kargl et al. (2001) and K¨omle et al. (2001).
The following table gives an overview of the required resources. Considering the
complex mechanics that are required to deploy and insert the penetrator, the overall
358 T. SPOHN ET AL.
Figure 12. Direct (“true”) temperature ﬁeld (left, PEN marked with dotted lines and sensor locations
marked by small rectangles) and “evolution” of the inverted temperature ﬁeld with increasing order
k of the solution in normalized coordinates. The dashed vertical line in the rightmost plot indicates
the location of the estimate of the undisturbed proﬁle.
experiment mass is very moderate. The data volume contains 2 Mbit of accelerom-
eter data (sampled during anchor shots at the very beginning) and a very low data
volume for the remaining science data (temperature and thermal properties data).
If needed, the data rate can further be reduced by conﬁguring larger time intervals
between measurement scans. The average power consumption can only be esti-
mated because it is partly determined by the heating power needed to keep the
front-end electronics inside the operating range, and, thereby, dependent on the
actual ambient conditions on the nucleus.
Mass (total) 2.35 kg
ID +PEN 0.65 kg
DD +DS 0.85 kg
Main electronics 0.6 kg
TM 0.12 kg
Harness 0.13 kg
Volume (envelope on balcony) 565 ×160 ×188 mm
Power (total) 2.2 W∗
Main Electronics 1.2 W
PENEL 0.2 W
TM 0.2 W
Data rate 180 kBit/ha
aAverage for nominal long-term operations (measurements every 20 sec)
The combination of MUPUS sensors will provide us with temperatures, thermal
properties and cohesion data for the uppermost 32 cm of a comet nucleus as a
MUPUS – A THERMAL AND MECHANICAL PROPERTIES PROBE FOR THE ROSETTA 359
function of time. Using MUPUS data, we will be able to determine the energy bal-
ance at one location of the nucleus as it approaches the Sun, thereby gaining insight
into the way comets work. A very important aspect of the MUPUS research is to
observe changes of physical properties in situ, and determine the speciﬁc timescales
of processes modifying cometary matter. Thermal and mechanical properties can
be interpreted in terms of microstructure parameters of the ice, such as particles
with variable texture, contact area between individual grains, and a chain-like par-
ticle structure which might be expected from a low-density material as snow (e.g.,
Keller and Spohn, 2002 and references therein). Having in mind that one of the main
goals of the mission is the search for pristine material, thought to be a record of the
formation of the solar system, it is of extraordinary importance to understand how
much the material analysed by COSAC, PTOLEMY and APXS has been modiﬁed
in the geological time record of the nucleus.
The authors, the MUPUS team, appreciate very much the co-operation with the
whole lander team. Markus Thiel, Max-Planck-Institut f¨ur Extraterrestrik, Garch-
ing, is responsible for the design of the lander’s anchors, and was always very
cooperative when the integration of the MUPUS sensors and their tests were con-
cerned. The MUPUS project and MUPUS team members are supported by various
grants, some of which are: Marek Banaszkiewicz: Grant No 2 PO3C.009.12 p/05;
Andrew Ball: PPARC (Particle Physics and Astronomy Research Council), The
Austrian Academy of Sciences, and the Royal Society. German contribution: DLR
grant WE 150 OH 9503 7-ZA. Austrian contribution: FWF Projects P12416 and
The principles of temperature and thermal property measurements are very simple,
but ensuring that the measured temperature is representative of the environment
temperature can pose great difﬁculty. The very existence of any thermal probe
inﬂuences the temperature ﬁeld, and the MUPUS PEN too can have a signiﬁcant
inﬂuence on the sub-surface temperature ﬁeld. If its thermal diffusivity is much
larger than that of the ambient material, heat conduction through the penetrator rod
can result in a distortion of the temperature gradient, resembling a thermal short
circuit. This is a common effect of all heat ﬂow experiments employing penetrators.
Especially in cases where measurement errors play a signiﬁcant role, using
forward modelling to explain the data is a way to pick one model that appears to
be physically realistic, but this model is not necessarily the only one ﬁtting the
360 T. SPOHN ET AL.
data. Forward modelling techniques demonstrate the existence of a solution, but
the problem of uniqueness of the solution remains unsolved.
The calculation of an unperturbed temperature proﬁle from perturbed data with
associated errors is a classical problem of inverse theory. It differs from the clas-
sical forward heat conduction problem in that it does not require boundary con-
ditions, but solves for them. In the case of an inverse heat conduction problem,
we have temperature histories at a number of points inside the volume and try
to calculate the boundary conditions, which involves estimating the temperature
throughout the whole volume (e.g. Stolz, 1960). Hagermann and Spohn (1999)
have successfully developed a numerical inversion scheme to ﬁnd the undisturbed
temperature proﬁle. Their algorithm uses the time-slope dependent extrapolation
of measured temperature histories (Kurpisz, 1991). The algorithm approaches the
undisturbed temperature proﬁle by adding time-dependent extensions to a stationary
In normalised cylindrical coordinates rand zand normalised time t, the temper-
ature response R of the i-th temperature sensor equals the normalised temperature
at the sensor location
With the time derivatives
dtk,k=1,2, ... (A.2)
we can ﬁnd the solution
where the ψ(k)
ij are recursive solutions of
His the linear system describing the thermal system of the penetrator and the
surrounding material. Figure 12 shows how the result of the transient temperature
ﬁeld is constructed from a crude stationary solution. The method proved to be robust
in most realistically conceivable scenarios (Hagermann and Spohn, 1999).
Ball, A. J., Gadomski, S., Banaszkiewicz, M., Spohn, T., Ahrens, T. J., Whyndham, M., et al.: 2001,
Planet. Space Sci. 49(9), 961.
Banaszkiewicz, M., Seiferlin, K., Spohn, T., Kargl, G., and K ¨omle, N.: 1997, Rev Sci. Instrum. 68,
Benkhoff, J. and Spohn, T.: 1991, Geophys. Res. Lett. 18, 261.
MUPUS – A THERMAL AND MECHANICAL PROPERTIES PROBE FOR THE ROSETTA 361
Benkhoff, J., et al.: 1995, SuSI Proposal (Principal Investigator: Dr. Johannes Benkhoff, Southwest
Research Institute, San Antonio). SwRI Proposal No. 15-17876, May 1995.
Benkhoff, J. and Huebner, W. F.: 1995, Icarus 114(2), 348.
Benkhoff, J., Seidensticker, K. J., Seiferlin, K., and Spohn, T.: 1995, Planet. Space Sci. 43, 353.
Benkhoff, J.: 2002, Adv. Space Res. 29(8), 1177.
Espinasse, S., Klinger, J., Ritz, C., and Schmitt, B.,1991, Icarus 92, 350.
Gr¨un, E., Heidrich, R., Hesselbarth, P., Kohl, H., and Benkhoff, J.: 1991, Geophys. Res. Lett. 18, 257.
Gr¨un, E., Bar-Nun, A., Benkhoff, J., Bischoff, A., D¨uren, H., Hellmann, H., et al., 1991, in: Newburn,
Jr., R. L., Neugebauer, M., and Rahe, J. (eds.), Comets in the Post-Halley Era (Kluwer Academic
Publishers, Dordrecht), p. 277–297.
Hagermann, A.: 1996, Measurement of the subsurface temperature proﬁle of a cometary nucleus:
Model calculations for the PEN-TP experiment of the MUPUS package (in German). Diploma
thesis, Institut f¨ur Planetologie, Westf¨alische Wilhelms-Unversit¨at M¨unster.
Hagermann, A.: 1999, Die Ermittlung des oberﬂ¨achennahen Temperaturfeldes aus einer
gest¨orten Gradientenmessung. Zur Inversion planetarer W¨armeﬂußmessungen unter besonderer
Ber¨ucksichtigung des Experiments MUPUS PEN-TP. Ph.D. thesis, Westf¨alische Wilhelms-
Hagermann, A. and Spohn, T.: 1996, Adv. Space Res. 23(7), 1333.
Huebner, W. F. and Benkhoff, J.: 1997, Earth Moon Planets 77(3), 217.
Huebner, W. F. and Benkhoff, J.: 1999, Space Sci. Rev. 90(1/2), 117
Kargl, G., Macher, W., K¨omle, N. I., Thiel, M., Rohe, C., and Ball, A. J.: 2001, Planet. Space Sci.
Keller, T. and Spohn, T.: 2002, Planet. Space Sci. 50, 929.
Klinger, J.: 1981, Icarus 47, 320.
Kofman, W., Herique, A., Goutail, J.-P., Hagfors, T., Nielsen, E., Barriot, J.-P., et al.: The comet
nucleus sounding experiment by radiowave transmission (CONSERT); a short description of the
instrument and of the commissioning stages, this issue.
K¨omle, N. I., Ball, A. J., Kargl, G., Keller, T., Macher, W., Thiel, M., et al.: 2001, Planet. Space Sci.
K¨omle N. I.: 1997, Planet. Space Sci. 45(12), 1515.
Kurpisz, K.: 1991, J. Heat Transfer 113, 280.
Marczewski, W., Schr ¨oer, K., Seiferlin, K., Usowicz, B., Banaszkiewicz, M., Hlond, M., et al.: 2004,
J. Geophys. Res. 109(E7), 1.
Paton, M.: 2005, Penetrometry of NEOs and Other Solar System Bodies. Ph.D. thesis, The Open
Prialnik, D., Benkhoff. J., and Podolak, M.: 2004, in: Festou, M., Keller, H. U., and Weaver, H. A.
(eds.), Comets II (University of Arizona Press, Tucson).
Seidensticker, K. J., M¨ohlmann, K. J. D., Apathy, I., Schmidt, W., Thiel, K., Arnold, W., et al.:Space
Sci. Rev., this issue, doi: 10.1007/s11214-006-9118-6.
Seiferlin, K., Spohn, T., and Benkhoff, J.: 1995, Adv. Space Res. 15(10), 35.
Seiferlin, K., Hagermann, A., Banaszkiewicz, M., and Spohn, T.: 2001, in: K ¨omle, N. I., Kargl, G.,
Ball, A. J., Lorenz, R. D. (eds.), Penetrometry in the Solar System (Verlag der ¨
Akademie der Wissenschaften, Vienna).
Seweryn, K., Banaskiewicz, M., Grunwald, M., Grygorczuk, J., and Spohn, T.: Int. J. Heat Mass
Transf. 48, 3713.
Smoluchowski, R.: J. Geophys. Res. 87, 422.
Spohn et al.: 1995, MUPUS Proposal (Responsible Proposer: Prof. Tilman Spohn, Institut f¨ur Plan-
etologie, Westf¨alische Wilhelms-Universit¨at, M¨unster), June 1995.
Spohn, T. and Benkhoff, J.: 1990, Icarus 87, 358.
Steiner, G.: 1990, Astron. Astrophys. 240, 533.
362 T. SPOHN ET AL.
Steiner, G. and K¨omle, N. I.: 1991, Planet. Space Sci. 39, 507.
Steiner, G., K¨omle, N. I., and K¨uhrt, E.: 1991, in: K¨omle, N. I., Bauer, S. J., and Spohn, T. (eds.),
Theoretical Modelling of Comet Simulation Experiments (Austrian Academy of Sciences Press,
Stolz, G.: 1960, J. Heat Transfer 82, 20.
Thiel, M., St¨ocker, J., Rohe, C., Hillenmaier, O., K¨omle, N. I., and Kargl, G.: 2001, in: K¨omle, N.
I., Kargl, G., Ball, A. J., and Lorenz, R. D. (eds.), Penetrometry in the Solar System (Austrian
Academy of Sciences Press, Vienna), p. 137–149.
Thiel, M., St¨ocker, J. E., Roh´e, C., K¨omle, N. I., Kargl, G., Hillenmaier, O., et al.: 2003, in: Harris,
R.A. (ed.), 10th European Space Mechanisms and Tribology Symposium. ESA SP 524 (ESA
Publications Division, Noordwijk), p. 239–254.
Thiel, M., St¨ocker, J. E., Roh´e, C., K¨omle, N. I., Kargl, G., Hillenmaier, O., et al.: 2003, The ROSETTA
Lander anchoring system. in: 10th European Space Mechanisms and Tribology Symposium. R.A.
Harris (ed.). ESA SP 524, ESA Publications Division, Noordwijk, 239–254.