Cryogenics 48 (2008) 463–468

Contents lists available at ScienceDirect

Cryogenics
journal homepage: www.elsevier.com/locate/cryogenics

Review

Direct measurement of total emissivities at cryogenic temperatures: Application

to satellite coatings
P. Herve a, N. Rambure a, A. Sadou a,*, D. Ramel a, L. Francou a, P. Delouard b, E. Gavila c
a
Université Paris 10, 1 Chemin Desvallières, 92410 Ville d’Avray, France
b
Contraves Space AG, Schaffhauserstrasse, 580 CH-8052 Zürich, Switzerland
c
Alcatel Space, 100 boulevard du Midi, BP99-06322 Cannes La Bocca, France

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents a direct measurement method for optical properties of different materials at cryo-
Received 9 March 2007 genic temperatures from 20 K to 200 K. It has been developed within the framework of the design of
Received in revised form 20 June 2008 Planck program. Planck is a satellite of the European Space Agency (ESA) that will be launched in
Accepted 16 July 2008
2008. The scientific goal of the Planck mission is to make observations of the temperature anisotropy
and polarisation of the Cosmic Microwave Background. The equivalent temperature of the observed radi-
ation is about 3 K and the telescope baffle temperature should not exceed 60 K in order to work properly.
Keywords:
The large Planck telescope is passively cooled by radiating to the Deep Space, so that a good knowledge of
C. Heat transfer
C. Radiant properties
the thermo-optical properties of its coating is of utmost importance for thermal modelling. However, up
D. Infrared detectors to now, few measurements have been done at such low temperatures. We derived a direct measurement
F. Space cryogenics method for the total directional emissivity of various coatings of interest for satellites applications. The
effective spectral range chosen the measurements covers 6–800 lm. We will describe the design of the
measurement apparatus and present results for several coatings.
Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction work in this spectral range as the propagation is strongly attenu-

ated by the numerous absorption bands of atmospheric water. That
Planck satellite that should be launched by ESA in 2007 will is why common infrared applications work in the 1–12 lm spec-
‘‘look back to the dawn of time”, close to the Big Bang, and will ob- tral band which covers most of the radiant energy at temperatures
serve the most ancient radiation in the Universe whose average between 200 K and 3000 K. For satellite applications, however,
temperature is 2.73 K [1]. there are no such transmission gaps and, when studying emission
To perform its mission, the satellite, inserted into the Lagrange- from very cold bodies, it is necessary to determine optical proper-
2 point of the Earth–Sun system, must comply with several techni- ties of coatings in the far infrared.
cal requirements. Thus, not to disturb the measuring process, We will present here the method and the equipment developed
temperature of the satellite telescope must remain stable around by LEEE for this project. The temperature range for experiments is
50 K. This temperature is passively obtained by the radiation of between 10 K and 200 K. The spectral range is between 6 lm and
the telescope toward deep space with the cold areas constantly 800 lm.
shadowed from the sun by the service module. The most emissive
coating should be chosen for the telescope outer elements in order 2. Measurement method
to maximize the radiated flux.
LEEE was asked by CONTRAVES SPACE (Planck telescope struc- To measure emissivity of opaque materials, we can use three
ture responsible), under supervision of ALCATEL SPACE (Planck different methods:
prime contractor), to study the emissivity of different coatings in
the temperature range 40–200 K. (1) by direct measurement of emissivity e (directional or hemi-
In order to make total emissivity measurements for coatings, spherical, spectral or total);
we have to consider that 97% of the radiant energy of bodies (2) by measurement of the reflectivity q, knowing the relation-
around 40 K is located in the far infrared at wavelengths between ship e = 1  q [2];
40 lm and 800 lm. It is of little use for on Earth applications to (3) by calorimetric measurements [3–5].

* Corresponding author. Tel./fax: +33 1 47097013/1645. With a calorimetric measurement, one obtains total hemispher-
E-mail addresses: pherve@u-paris10.fr (P. Herve), asadou@univ-ubs.fr (A. Sadou). ical emissivity. It is a relatively simple method to implement and

0011-2275/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cryogenics.2008.07.007
464 P. Herve et al. / Cryogenics 48 (2008) 463–468

Nomenclature

Cte apparatus function q total reflectivity

L0 radiation total intensity of the blackbody (W m2 sr1) qk spectral reflectivity
Mb measurement on blackbody qT total reflectivity at temperature T
Ms measurement on sample r constant of Stefan–Boltzmann (W/m2 K4)
T temperature (K) Subscripts
e total emissivity J index of iteration
ek spectral emissivity c chopper
eT total emissivity at temperature, T w wall
D temperature difference between two points B blackbody
k wavelength (m)

widely used at ambient temperature but, in previous measurement As we could check (see Section 4) that temperature of optical
campaigns and for temperatures lower than 100 K, we got heat components does not change during the time of measurement
leakages that induced parasitic phenomena of the same order of the term eO rT 4O is the in all the equation. We have four equations
magnitude as those of interest. for nine unknown parameters. To solve this system, several meth-
Indirect measurements could be a good alternative to obtain ods can be used, with their benefits and drawbacks conditioned by
emissivity but, as the coatings to be experimented are thin layers the assumptions we have to make. Combining Eq. (1) to Eq. (4), we
of dielectric materials deposited on metallic substrate, they be- then obtain the following formula.
come semi-transparent in the far infrared. Such phenomenon   ðMs  Ms Þ
B A
induces multiple reflections at the interfaces air-coating and eT B ¼ eT A ðT A =T B Þ4 þ 1  ðT A =T B Þ4
ðMbB  MbA Þ
metal-coating, resulting in a non-validity of the relationship
 ðqT B  qT A Þew ðT w =T B Þ4 ð5Þ
e = 1  q. We then concluded that, in this very situation, direct
emissivity measurement was the right choice. For evaluation of emissivity eT B the first and the last terms must
The main flaw lies in that the emitted radiation of the sample at be known or negligible. Results of measurements that had been
such low temperatures is very faint compared to the background previously conducted on the same Planck samples at the LEMTA
radiation received by the detector. In order to minimize the back- laboratory (Laboratoire d’Énergétique et de Mécanique Théorique
ground radiation, we use a vacuum chamber cooled by liquid nitro- et Appliquée; ENSEM; Nancy, France) showed that qk presents lit-
gen (77 K). In addition, we use a lock-in amplifier with a chopper tle variation with T in the range 40–300 K [6]. Such a result implies
R1
that modulates the component of the signal coming from the inner that qT ew rT 4w ¼ p 0 qk ðTÞew L0k;T w dk remains nearly constant with
part of the vacuum chamber and eliminates the outside back- T, i.e. qT A ew T 4w  qT B ew T 4w , where L0k;T w stands for the luminance of
ground radiation. When the detector sees the sample between a blackbody at the temperature of the vacuum chamber Tw. The
two rotating blades of the chopper, the signal is the sum of the influence of this weak temperature dependence on the measure-
thermal radiation emitted by the sample itself, added to the radia- ment results is discussed further in Section 4.
tion coming from all the parts of the vacuum chamber (with emis- Note: Whatever the sample temperature, the reflective part of
sivity ew  1 and a radiant equivalent temperature Tw = 80 K) that the signal is due to the chamber wall at 80 K. At this temperature,
is reflected by the sample plus the thermal radiation emitted by the wavelength of the maximum of radiated energy is 37 lm and
the optics (with emissivity eo, reflectivity qo and temperature To). about 97% of energy is contained in the spectral band 20–
When a chopper blade (with emissivity ec and temperature Tc) 220 lm. So will be the spectral band of the reflected signal and
hides the sample, the detector receives the sum of the thermal if, as assumed, qðk; TÞ does not vary with T, the variation of the to-
radiation emitted and reflected by the blade, which value can be tal emissivity of samples with temperature is only due to the shift
higher than the useful signal. The lock-in amplifier calculates the in the spectral distribution of the measured radiation energy.
difference between the two above signals. Hereafter we shall write qT A ¼ qT B in Eq. (5). We then obtain the
The method proposed operates by subtracting the signals ob- following formula.
tained for two different sample temperatures TA and TB. Measuring
ðMsB  MsA Þ
the radiation of the sample and of a blackbody for the same two eT B ¼ ðT A =T B Þ4 eT A þ ð1  ðT A =T B Þ4 Þ ð6Þ
temperatures, we get the following relationships: ðMbB  MbA Þ
We shall then present two methods for solving this equation:
(1) Intensity measurement of the sample at temperature T1.
 A ‘‘differential method at the lowest temperature”:
MsA ¼ Cte ðeT A qO rT 4A  ec rT 4c þ qT A qO ew rT 4w  qc ew rT 4w þ eO rT 4O Þ
ð1Þ To eliminate the influence of the first term with unknown emis-
sivity eT A , we can keep for TA the lowest temperature for which we
(2) Intensity measurement of the sample at temperature T2.
 4 result (Tmin = 10 K). As we raise the sam-
have got a measurement
MsB ¼ Cte ðeT B qO rT 4B  ec rT 4c þ qT B qO ew rT 4w  qc ew rT 4w þ eO rT 4O Þ ple temperature, TTmin B
 1 and the Eq. (6) quickly reduces to:

ð2Þ ðMsT B  MsTmin Þ

eT B ¼ ð7Þ
ðMbT B  MbTmin Þ
(3) Intensity measurement of the blackbody at temperature T1
For the temperature TB close to Tmin, the accuracy of the results
MbA ¼ Cte ðqO rT 4A  ec rT 4c  qc ew rT 4w þ eO rT 4O Þ ð3Þ strongly depends on the hypothesis of the constancy of qk when T
varies, but as the temperature TB becomes higher than Tw, this
 4
(4) Intensity measurement of the blackbody at T2 assumption is useless because TTwB  1.
MbB ¼ Cte ðqO r T 4B  ec rT 4c  qc ew r T 4w þ eO rT 4O Þ ð4Þ  An ‘‘iterative differential method”:
 P. Herve et al. / Cryogenics 48 (2008) 463–468 465

In Eq. (6), let’s take for TA the temperature of the sample at the imize the effects of internal reflections. With a ray tracing code, we
immediately preceding measurement step; i.e. TA = TB  DT. As the found that the emissivity of the blackbody exceeds 0.996 when the
difference between TB and TA remains lower than in the variant 1, emissivity of the inner coating exceeds 0.3. We used Velvet paint-
the hypothesis regarding qðk; TÞ should be more easily verified. In ing whose emissivity is clearly better than 0.3 for the lowest tem-
return, as there are relatively low differences between the results perature of interest. So the corresponding error can be neglected in
of two successive measurements steps, the method should be more regard of the other sources of error.
sensitive to noise. A very good thermal conduction is needed to reach such a low
Let then write eT A  eT B . Eq. (5) reduces then to temperature as 10 K and an aluminium alloy AA1050 has been cho-
sen for the cylinder and the rod. A 25 W heating resistance stuck on
ðMsT B  MsT B DT Þ the back of the sample holder is connected to a PID regulator. The
eT B ¼ ð8Þ
ðMbT B  MbT B DT Þ very weak calorific capacity at low temperatures of this alloy
makes easier the temperature regulation. Moreover, all thermal
Writing eT A  eT B is a rough approximation and the result will be
contacts have been optimised by application of a vacuum grease
refined by iterations through equations system Eq. (9).
(Apiezon N).
If TA = TB  DT, Tmin is the lowest temperature for which we
The sample temperature is controlled by a silicon diode (tem-
got measurement results and if we define ‘‘j” as index of iteration,
perature range 4–500 K with a 0.01 K sensitivity) inserted at the
calculation of emissivity is done following equations system Eq.
back of the sample holder and the temperature gradient between
(9):
the blackbody and the samples is checked by a second silicon diode
9 stuck on the back of the former. We found that, whatever the tem-
if j ¼ 1 : >
>
>
> perature, the measured gradient does not exceed 0.1 K.
ðMs Ms
eT B ¼ ðMbTT B MbTTB DDTT Þ
Þ >
>
= The global measurement apparatus is mainly composed of a
B B
ð9Þ vacuum chamber enclosing the optical parts (Fig. 2) and of the
else if 2 6 j 6 T B T
DT
min
: >
>
>
>
ðMsT MsT DT Þ >
detection setup.
>
eðT B Þj ¼ ðT BTBDT Þ4 eðT B DTÞj1 þ ð1  ðT BTBDT Þ4 Þ B
ðMbT MbT
B ;
Þ With a turbo-molecular pump vacuum chamber we get in the
B B DT
chamber a secondary vacuum about 107 mbar. So we avoid con-
The last step gives the best evaluation of emissivities. This var- vective heat transfers and selective absorption of radiations by
iant should normally give better results than variant 1, but the re- atmospheric water.
verse can be true with noisy measurement results. To minimize the level of ambient radiation, the sample holder is
For example to determine emissivity at 50 K for DT = 10 and placed into a double wall enclosure poured with liquid nitrogen. In
Tmin = 10 K : addition and to minimize reflections, the inner wall has been
ðMs50  Ms40 Þ painted with Velvet black coating. So parasitic radiations in the
e501 ¼ for j ¼ 1 and e50j chamber depend only on the wall temperature.
ðMb50  Mb40 Þ
 4  4 ! The global background flux is equivalent to the radiation of a
40 40 ðMs50  Ms40 Þ blackbody at 80 K and we checked that it remains very stable
¼ e40j1 þ 1  for j ¼ 2 to 4:
50 50 ðMb50  Mb40 Þ and homogeneous when the temperature of the sample varies.
The optical part of the system is composed of:
Details of such iterations are reproduced further in Table 1, Sec-
tion 4.
(1) a parabolic mirror disposed in front of the samples, which
collects the sample signal into a parallel beam;
3. Experimental apparatus (2) a flat mirror that reflects the parallel beam out of the main
chamber through the modulator rotating chopper;
To cool the samples, we use a two-stage helium refrigerator. (3) a second parabolic mirror that focalises the beam onto the
The lowest sample temperature that can be obtained with this bolometer detector through a diamond window that has a
refrigerator is 8 K. very flat transmission curve from 6 lm to 800 lm.
The sample holder (Fig. 1) is a cylinder directly linked through a
rod to the second stage of the helium refrigerator. For each mea- All the optical components and their supporting devices are
surement, five different samples can be disposed on the holder. A thermally insulated and they are passively cooled down to 80 K.
blackbody, specially designed for the application [7], with its aper-
ture disposed on the optical path of the measurement setup, is
positioned at the top of the sample holder. The whole system can
Black Body
be rotated and translated along its axis to allow measuring of the
emission of each sample at multiple angles. 4 mm
Global accuracy of our results depends on the precision of the
Temperature
estimation of the reference emissivity value. The geometry of the sensor
cavity of the blackbody has been carefully studied in order to min-
Samples

Heater
Table 1 8 mm
Example of error estimation in ‘‘iterative differential method”

T etarget e1 De1 e2 De2 e3 De3 e4 De4

(K) (%) (%) (%) (%)
10 0.557
20 0.606 0.629 2.27
Rotation
30 0.690 0.711 2.07 0.695 0.45
40 0.757 0.789 3.12 0.764 0.66 0.759 0.14 translation
50 0.810 0.847 3.66 0.823 1.28 0.813 0.27 0.811 0.06
Fig. 1. Schematic diagram of the sample holder.
466 P. Herve et al. / Cryogenics 48 (2008) 463–468

Liquid N2 2.5% 1
Shrouds
0.9
Chopper Bolometer
2.0% 0.8

0.7
Sample holder

Emissivity
Error (%)
& Black body 1.5% 0.6
Vacuum “Iterative differential method”
Chamber 0.5
“Differential method at the lowest
20 K 1.0% temperature” 0.4
Emissivity total
0.3
10 K Coating Samples
0.5% 0.2

77 K 0.1
Refrigerator Thermal shield
Helium 0.0% 0
20 40 60 80 100 120
Axis of rotation
& translation Temperature (Kelvin)

Fig. 4. Comparison of systematic errors for two methods of the calculation

Fig. 2. Schematic diagram of the experimental set-up. ‘‘differential method at the lowest temperature” and ‘‘iterative differential method”.

0.5
B=1.10 -5
6
0.4

4
Error (%)

0.3
Error (%)
2

0.2
0
40 45 50 55 60 65
0.1
-2

0 -4
40 50 60 70 80
Temperature (Kelvin)
Temperature (Kelvin)
Fig. 5. Estimation of global error.
Fig. 3. Systematic error due to a supposed reflectivity variation qT;S¼80 K .

The detector is a bolometer with an effective bandwidth 2– eT and the total reflectivity qT;S¼80 K of the coating at T reflecting
1000 lm. The detector is cooled at 4.2 K by liquid helium and its a signal emitted by a source at 80 K. Thus, we modelize the energy
associated amplifier NEP is 1.17E13 W Hz1/2. radiated by the sample and by the blackbody for each temperature
To measure the total emissivity, the energy must be collected T:
over the whole optical bandwidth. With gold mirrors, the bolome- 
MsT ¼ eT  L T þ qT;S¼80 K  L Tw
ter and diamond window, we get a very good achromaticity all ð10Þ
MbT ¼ L T
over the spectral range of interest.
The signal (from the preamplifier of the bolometer) is sent to and we can study the influence of the various assumptions stated to
the lock-in amplifier with a 0.02 Hz equivalent Noise Bandwidth. determine total emissivity from our measurements.
The difference between the flux received at 40 K and at 20 K is To evaluate the error made by the assumption qT A ¼ qT B , we
about 108 W and, as the global noise of our detection system is introduce the model values of MST and eT A in Eq. (5) to determine
about 1.6  1014 W, the signal/noise ratio in the calculation meth- eT B and we compare with the modelized emissivity at TB. The
od described above is about 105. resulting errors are shown for B = 105 (Fig. 3).
The system is cooled for 12 h before beginning the measure- We estimate the error due to the other assumptions made for
ments to make sure that all the components are well stabilized the two methods when qT A ¼ qT B (Fig. 4). The error is always ‘‘in
at 80 K. excess”. The ‘‘iterative differential method” looks better than the
‘‘differential method at the lowest temperature”, but the reverse
4. Error analysis is true if we introduce noise levels that are of the same order of
magnitude as the radiant energy to be measured. The resulting er-
There are two main categories of errors that can be considered ror is less than 0.5% for temperatures above 40 K and remains neg-
separately in the global error budget: the systematic one, that re- ligible for temperatures higher than 80 K. An example of results for
sults from the assumptions in the calculation method and the ‘‘iterative differential method” is reproduced in Table 1 below.
experimental one, part of which being random. We shall then consider the different sources of experimental er-
To get an evaluation of the systematic error, we use already rors. The main causes of experimental errors are:
mentioned (unpublished) results from LEMTA. We use the results
ek;300 on the spectral band 1–600 lm and we suppose that the var- (1) a variation of the background (temperature stability of the
iation of the spectral emissivity obeys a law ek;T ¼ ek;300  ‘‘vacuum chamber” and the Au optics).
Bð300  TÞ with B comprised between 106 and 5  105 depend- (2) an error on the blackbody (reference) emissivity value:
ing of the coating. From ek;T , we can integrate the total emissivity 60.4%.
 P. Herve et al. / Cryogenics 48 (2008) 463–468 467

1 0.96

0.95 0.94

0.92
0.9
Emissivity

Emissivity
Velvet PUK A 0.9
Velvet PUK B
0.85
Z306 PUK A
Z306 PUK B 0.88
0.8
0.86
Velvet
0.75 PUK B COND (390 ; 128µm)
0.84 Z306
PUK B REF (062 ; 116µm)

0.7 0.82
40 60 80 100 120 140 20 30 40 50 60
Temperature (Kelvin) Angle (°)

Fig. 7. Angular total emissivities obtained by the differential method.

Fig. 6. Run-to-run errors emissivity errors due to the uncertainty of measurement
(the repeatability of the process).

(3) the relative average deviation of the measured electrical sig- 1.15
nal due to the electronic noise of the measuring equipment: Transition
±0.05% of the measured signal at 40 K and practically negli- 1.05
gible above 60 K.
(4) differences in temperature between the samples disposed on 0.95 MH21

the holder and the blackbody. We found that these differ- Signal (mV)
Black
ences were less than 0.1 K. 0.85
Anodisation
Z306

We used thermocouples to check the temperature stability of

0.75
the vacuum chamber and of the optics. We could observe a very
good temperature stability of the pieces of equipment during the
0.65 KT
measurement process. For instance, the temperature measured Honeycomb
Blackbody
for the inner surface of the double wall in the vacuum chamber
0.55
is 80 K with a less than ±0.2% variation when the sample temper- 400 500 600 700
ature is below 60 K. Above 60 K, the influence of the background Time (Sec)
radiation reduces very quickly.
To figure out the uncertainties due to all these perturbations, Fig. 8. Example of signals recorded for several samples at 20 K (KT: Kayser-Threde).
we took the simulated measurement with B = 105 and we added
random perturbations to the corresponding measured intensities.
Then we introduced back in equation system Eq. (10) the disturbed
1
intensity values to derive emissivity by the differential method. In
Fig. 5, we represented the global error. The error on emissivity is 0.9
about 2% for 60 K and still acceptable for lower temperatures.
In Fig. 6 we show the results of measurements realized on two 0.8
different samples of the same coatings, respectively Velvet and
Emissivity

Z306 A and B. The dispersion of results is coherent with the estima- 0.7
KT
tion of global errors. Moreover, looking at the curve aspect gave us Mh21
0.6 Honeycomb
a good confidence concerning the repeatability of the process. BA
Z306
0.5
5. Measurements and results
0.4
In order to avoid multiple reflections between the sample and
0.3
the detection optics, we measured the signals at a minimum angle 40 60 80 100 120 140 160 180 200
of 20° with the normal to the sample surface. For complete results,
Temperature (Kelvin)
we measured the emissivities for several angles between 20° and
60° (Fig. 7). Fig. 9. Emissivities at 20° of different coatings obtained by the differential method
For a 20° angle and a sample temperature stabilized at 20 K, (KT: Kayser-Threde; BA: Black Anodisation).
typical results are shown in Fig. 8. Even for such a low temperature
we observe significant differences between all samples, which
means that the sensitivity of our measuring equipment is good. tained with about +/4% error at 40 K. For temperatures higher
On Fig. 8, for the lowest temperature, we mainly detect the re- than 60 K, the accuracy is much better and the cumulated error
flected signal from the samples. So a higher signal value means a does not exceed 3%. We then classify the coatings following their
lower emissivity value. We can verify that our blackbody is effec- emissivity. We made some repetitivity measurements that confirm
tively less reflective than the coatings. our error calculations (Fig. 5).
The results obtained for the total emissivity of the different As we can dispose five samples at the same time on the holder,
coatings are reproduced. As discussed above, the results are ob- we wanted to check that the results are not influenced by the
468 P. Herve et al. / Cryogenics 48 (2008) 463–468

position of the samples. So, we repeated the experiment after A new improvement could result from cooling the double wall
changing the order of the samples. The results have a lower disper- with liquid helium instead of liquid nitrogen. The color tempera-
sion than that due to repetitivity. We then concluded that the ture of the background could then get down and stabilise around
position of the sample does not influence measurements. 30 K instead of 80 K. We could also dispose the chopper motor
It clearly appears in Fig. 9 that emissivity decreases with temper- and the bolometer inside the chamber.
ature for all coatings and the lowest temperatures part of the curves We are often questioned about direct spectral emissivity mea-
is very similar for all coatings. Obviously, the lower the temperature surements at cryogenic temperatures. We already conducted such
the higher the measurement error, but the smooth appearance of the experiments at as low temperatures as 200 K. With the identified
curves suggests that the measurement error is small enough not to improvements to the experimental setup, we can foresee to study
hide the underlying physical phenomenon. The coatings studied the spectral emissivity of dielectric materials at the same levels of
are thin dielectric films disposed on copper or aluminium substrate. temperatures as those considered here. Such measurements could
At low temperatures, the global radiated power is located in the far then be done using the bolometer in association with a Fourier
infrared where the coatings become transparent. The energy mea- Transform Spectrometer.
sured corresponds then to the radiation of the metallic substrate
through a more and more transparent layer.
References
6. Conclusions and future prospects
[1] Collaudin B, Rando N. Cryogernics in space: a review of the missions and of the
technologies. Cryogenics 2000;40:797–819.
The described work was aimed to obtain the total directional [2] Francou L, Hervé P. An FTIR based instrument for measuring infrared diffuse
emissivity at cryogenic temperatures of various dielectric coatings reflectance. Quantitative infrared thermography (QIRT 1998), Eurotherm
Seminar no. 60, Lodz Pologne; 7–10 September 1998.
disposed on metallic substrate. The results give ESA the capability [3] Mattei S, Especel D. Une méthode de mesure de l’émissivité thermique des
to choose the most efficient coating in order to cool the Planck sa- matériaux opaques à la température ambiante. Revue générale de la thermique
tellite baffle and to calculate the equilibrium temperature close to 1994;33(8):239–58.
[4] Fabron C. Measurement of total hemispheric emissivity at low temperatures –
the telescope. designing a cryogenic test bench. In: Intal conference on environmental
The main flaws of the measurement process were due to the systems; July 2000, Toulouse France, Publication SAE 2000-01-2526.
background radiations inside the vacuum chamber that reflect on [5] Musilova V, Hanzelka P, Kralik T, Srnka A. Low temperature properties of
materials used in cryogenics. Cryogenics 2005;45(8):529–36.
the sample. We constantly tried to reduce the sources of errors
[6] Delouard P, Krähenbühl U, G.Peikert G. Materials characterisation at cryogenic
by improving the experimental setup. Using the ‘‘iterative differen- temperatures for the Planck telescope. In: European conference on spacecraft
tial method” and using a very precise regulation of the chamber structures. materials & mechanical testing 2005, Noordwijk, The Netherlands;
wall, we obtained a drastic reduction of the influence of the most 10–12 May 2005. Publication ESA, SP-581; August 2005.
[7] Te Y, Jeseck P, Camy-Peyret C, Payan S, Briaudeau S, Fanjeaux M. High emissivity
obvious perturbations. However, after each improvement step, black body for radiometric calibration near ambient temperature. Metrologia
new perturbations appeared that were previously hidden. 2003;24–30(40).