Pulsed carbon fiber illuminators for FIR instrument characterization

Infrared Physics & Technology 52 (2009) 159–165

Contents lists available at ScienceDirect

Infrared Physics & Technology journal homepage: www.elsevier.com/locate/infrared

Pulsed carbon fiber illuminators for FIR instrument characterization S. Henrot-Versillé *, R. Cizeron, F. Couchot LAL, Univ Paris-Sud, CNRS/IN2P3, Orsay, France

a r t i c l e

i n f o

Article history: Received 15 November 2007 Available online 7 July 2009 Keywords: CMB Characterization Carbon fiber Planck HFI Submm Bolometer Heat capacity Thermal conductivity

a b s t r a c t This article describes the properties of the carbon fibers that were used during the ground calibration of the High Frequency Instrument of the Planck satellite. It focuses on the properties of this new device used as radiation sources, and on the modelling of its thermal behaviour. Experimental data are presented and successfully compared with the proposed theory. Their small time constant, their stability and their emission spectrum pointing in the submm range make these fibers a very useful tool for characterizing FIR instruments. Ó 2009 Elsevier B.V. All rights reserved.

1. Introduction We have been studying carbon fibers as illuminators for FIR instrument characterization between 2000 and 2005 with the specific goal of measuring the sum of the electrical and optical crosstalk between the different channels of the High Frequency Instrument of the Planck satellite [1,2]. As far as this article is concerned, most of the effort has been put on thermal modelling since a good agreement has been obtained between data and simulations within this framework. The first section gives an overall view of the experimental setup and ends up with a list of requirements for the carbon fibers. The second section details the fiber device and its electronics. While the third and fourth sections give the basic equations used to determine the fibers behaviour and the extraction from experimental data of their main properties, ending up with a self consistent comparison with simulations. 2. Experimental setup 2.1. Planck-HFI The ESA CMB Planck mission will be launched in April 2009. It will produce full sky maps in nine frequency bands ranging from 30 to 1000 GHz, including polarisation maps up to 350 GHz. The full analysis of these maps will provide percent level constraints on cosmological models. The satellite hosts a 1.5 m diameter telescope that sweeps slowly the sky. Two instruments share the focal * Corresponding author. Tel.: +33 164468269. E-mail addresses: [email protected] (S. Henrot-Versillé), [email protected] (R. Cizeron), [email protected] (F. Couchot). 1350-4495/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.infrared.2009.07.001

plane: The Low Frequency Instrument (LFI), up to 100 GHz, uses HEMT as detectors, and the High Frequency Instrument (HFI), based on bolometers, covers frequencies from 100 to 1000 GHz. This work has been motivated by specific needs of HFI bolometers ground calibration. 2.2. The Saturn setup HFI bolometers working point is around 100 mK. On the satellite, a three stages cryogenic system provides a very stable thermal environment, required by the exceptional instrument sensitivity. For ground testing and calibration at the instrument level, a specific environment has been developed at IAS Orsay, with similar requirements. This so-called ‘‘Saturn” cryostat, where HFI ground calibrations took place, allowed us also to lead the thermal characterization of the source fibers during the cooling down of the setup, for temperature ranging from 300 K to 1.7 K. Inside the cryostat, a 2 K enclosure contains the HFI Instrument and the calibration optical system detailed in [3]: an integrating sphere fed by internal sources and a path to external sources, a spherical mirror conjugating the output of the sphere and the HFI focal plane, and an instrumented support. This 2 K enclosure is surrounded by a 20 K and a 80 K shield. The fibers setup is installed on the instrumented support, on a side of a three position wheel. The light sources are facing the entrance of part of the cold optics of the instrument, as shown on Fig. 1. Two additional fibers are installed behind a dedicated small hole in the mirror facing the focal plane. They illuminate all the bolometers synchronously for time constant measurements purposes. Being further from the instrument, the mirror fibers illuminate the HFI horns about 300 times less than the wheel fibers.

160

S. Henrot-Versillé et al. / Infrared Physics & Technology 52 (2009) 159–165

Fig. 1. Schematics and picture of the fibers setup facing HFI focal plane (which cold optics is represented by the right hand side box on the drawing and which is on the left on the picture).

Fig. 2. Stacked signal of one HFI bolometer from the qualification model.

2.3. Requirements Taking into account the constraints of the previously described experimental setup, and the needs for the HFI characterizations, the requirements for developing these carbon fibers were threefold:  To get a signal ranging up to a few pW detected on the HFI bolometers for the concerned frequencies (ranging from 100 to 857 GHz) including transmission efficiency of the cold optics horns and integration over the 30% wide frequency bands.  To get a relatively small time constant (smaller than 10 ms) and a repeatable signal in a pulsed regime, allowing to stack the measurements and increase the signal over noise ratio.  To be able to switch on the fibers without introducing any parasitic signal due to EMI-EMC from the electrical pulse used to drive them. The carbon fibers did meet these requirements. Even more, on the time constant issue, the fibers were finally included in the list of sources used to characterize the time response of the instrument. Fig. 2 shows an example of one bolometer signal induced by a pulsed fiber. The quality of the fiber data revealed the presence of a second slow time constant below the percent level on some bolometers [4]. This defect has been understood and mastered thanks to an additional dedicated run during 2008 summer at the Liège CSL, for which a new device including the fiber sources has been built.

Fig. 3. Photography of one of the carbon fiber used for the Planck-HFI calibration: the diameter is 6 lm and the length  1 mm.

3. Fiber device and electronics 3.1. The fiber device One of the authors suggested to use carbon fibers, since their resistivity does not fall to zero at low temperature, which allows to easily warm them through Joule effect. Besides, the good surface over volume ratio of a fiber should allow to reach a sizeable emitting surface with a tiny amount of matter, and 1 mm long fiber should behave as good antennas for mm range or smaller wavelengths. We used on the shelf carbon fibers1 which happened to fulfill our requirements. The work presented here allowed us to quantify a posteriori the reasons of the fiber behaviour. The physical properties of materials at low temperature depend much on their fabrication procedures, so the results on heat capacity and thermal conductivity shown hereafter are specific to the fiber we used. But, they certainly give a good indication of a generic behaviour of carbon fibers. Photography on Fig. 3 shows a typical fiber. Our fibers have length L  1 mm. The diameter of a slice is 6 l. They are glued and thermalized at each end on a Kapton–Copper circuit by a drop of Ag lacquer on an extra fiber length of about :25 mm on each side. Kapton thickness is :09 mm and Copper layer thickness is :04 mm. 1 £ 7 l  12 mm fibers, labelled BESMGHT, Grade IM 500  12000, bought in 1994 by AKZO Fibers SA, F 69626 Villeurbanne Cedex, France.

161

S. Henrot-Versillé et al. / Infrared Physics & Technology 52 (2009) 159–165

drawing of this device. Seven such horns have been put on the Saturn instrumented wheel. 3.2. Dedicated electronics description The electronics used for resistance and resistance variation measurements is described in Fig. 5. The input signal V 0 is a square signal whose amplitude, frequency, and width are tunable. Ri is also finely adjustable. Offset compensations are not shown on the sketch. The output is simply given by:

! e tot ðtÞ R GV 0 ðtÞ 1 Ri

Fig. 4. The fibers inside their housing.

V 1 ðtÞ ¼

e tot ¼ Rfiber þ Rwire corresponds to the resistance of the fiber in where R series with the 15 m long Constantan wires inside the Saturn cryostat. Before any measurement, the system is balanced in order to get: e tot ðt ¼ 0Þ. Taking t ¼ 0 at the beginning of the rising pulse, Ri ’ R one gets, with this initial condition:

Rfiber

Rwire

r

V0(t)

ð1Þ

Ri − +

V 1 ðtÞ ¼ V 1 ð0Þ þ

GV 0 ðtÞ ðRfiber ðtÞ  Rfiber ð0ÞÞ Ri

ð2Þ

100 k Ω

100 k Ω

V 0 ðtÞ and V 1 ðtÞ are stored on a numerical oscilloscope. The gain factor G is measured by switching on and off the small additional resistor r whose value is known. This gives the absolute scale for Rfiber ðtÞ measurements. As shown on Fig. 6, 105 relative precision on Rfiber is reached. Sections 5.2.1 and 5.3 describe the analyses of the data taken with this electronics.

V1(t)

G

Fig. 5. Illustrating the electronic circuit used to characterize the resistance of the carbon fibers.

Fibers are bended to minimize thermal constraints on the fiber, and to prevent high linear polarisation of fiber signal. Two independent facing fiber are placed on every Kapton board. Doubling the fibers enhances system fiability. Besides, it allowed us to check the linearity of the full detector chain. The fibers are put in a small integrating cavity, at one end of a horn whose shape is conceived to focalize fiber signal at the entrance of HFI focal plane horns, a few cm in front of these source horns. The geometrical light cone half angle is ’ 24 . Fig. 4 shows a

4. The thermal approach 4.1. Equations and assumptions [5] The fiber is heated by Joule effect produced by the current generator described in former section. Heat propagates along the fiber and flows through fiber ends thanks to thermal conduction. Fiber ends are thermalized at T 0 , which we assume to be independent of fiber heating (cf. Fig. 3), and equal to the temperature measured around the source horns. For moderate fiber temperature, one can neglect the fiber cooling due the rT 4 power radiated through the Voltage(V)

Voltage(V)

0.025 0.02 0 0 -0.025 -0.02 -0.05 -0.04 -0.075 -0.06

-0.1

-0.08

-0.125

-0.1

-0.15 -0.175

-0.12 0

0.01

0.02

Time(s)

0.03

0.04

0

0.01

0.02

0.03

0.04

Time(s)

Fig. 6. Example of V 1 ðtÞ when a square signal V 0 of 0:2 V is applied: on the left with rð¼ 2:6 XÞ switched on and on the right with r off. Rfiber ðtÞð 700 XÞ is measured with a precision of the order of :05 X.

162

S. Henrot-Versillé et al. / Infrared Physics & Technology 52 (2009) 159–165

blackbody radiation. For instance, at T ¼ 200 K, a fiber with a surface of :04 mm2 and radiating as a pure blackbody would loose 4 lW by radiation, whereas the Joule power needed to heat the fiber around 200 K is in the mW range. If one assumes that T is homogeneous on all the fiber section, Tðx; tÞ, the temperature along the fiber as a function of time, is solution of the one dimensional heat equation:

@ @x



jðxÞS

 @Tðx; tÞ lðxÞI2 @Tðx; tÞ þ ¼ qC p ðxÞS @x @t S

ð3Þ

then, Eq. (3) exact solution for the mean temperature increase in steady state regime reads [5]:

2

DT mes

16 ¼ 41 

a

LI 2S

1 qffiffiffiffiffiffi tanh l0 a j



LI 2S

rffiffiffiffiffiffiffiffiffi

3

l0 a 7 5: j

ð11Þ

5. Experimental characterization Two kinds of data are analysed:

where:  x origin is set at the middle of the fiber of length L.  q is the mass per unit of volume and S the surface of a slice.  The thermal conductivity j, the heat capacity C p and the resistivity l depend on x through there temperature dependence. Rfiber is obtained through

l integration over x.

 resistance (or resistance variations) measurements at the edges of the fibers done using the dedicated electronics described in Section 3.2 for different thermalisation temperatures and for small values of the current applied on the fibers.  bolometers’ data translated in terms of incident power for a thermalisation temperature around 2 K and a wide range of current values.

4.2. Basic properties

5.1. Resistance and temperature

In the steady state regime, and in the simplest case where no parameter exhibits any temperature dependence, Eq. (3) gives:

At room temperature the measured resistivity of the fibers is 1:8  105 X m which gives for 1 mm fibers a resistance of 640 X at 300 K. This resistance does depend on the thermalisation temperature as shown on Fig. 7 for seven fibers (on the left): the dispersion of these measurements includes the spread of the resistance of the Ag-lacquer contacts and the fact that the length of the fibers lies between 0.94 and 1.06 mm. The figure on the right shows, for the same seven fibers, their resistance corrected for at their value measured at 80 K and divided by the slope dR dT T ¼ 80 K to correct for length effects. It illustrates the homogeneity of the results. The contact resistance due to the carbon–Ag lacquer interfaces is of the order of 100 X.

j

@ 2 T RI2 þ ¼0 @x2 LS

ð4Þ

leading to a parabolic temperature profile:

RI2 L2 TðxÞ ¼ T 0 þ  x2 2LSj 4

! ð5Þ

and a mean temperature increase:

DT mes ¼

1 L

Z

L=2

ðTðxÞ  T 0 Þdx ¼

L=2

RI2 L : 12jS

5.2. Thermal conductivity

ð6Þ

When one stops heating the fiber, the system relaxes to the T 0 temperature, and the transcient regime can be described by the generic Fourier expansion solution of Eq. (3):

Tðx; tÞ ¼ T 0 þ

1 X

DT k0 eðt=sk Þ cos½ð2k þ 1Þpx=L;

ð7Þ

k¼0

where the time constants

sk are given by:

2

sk ¼

qC p L ; p2 jð2k þ 1Þ2

ð8Þ

the leading term being by far the k ¼ 0 component, which corresponds to the longest time constant:

s0 ¼

qC p L2 : jp2

ð9Þ

The fiber thermal time constant is then proportional to the square of its length. Therefore, smaller time constants may be obtained with shorter fibers, at the price of a flux loss. L should stay sizeable with the longest needed wavelength, since the fiber behaves like an antenna. For small values of I; DT mes is small, and these formula apply even if j; C p and l vary with T. This approximation is used in Section 5.2.1. The fact that the resistivity is function of T can be treated analytically in the simplest case where jðTÞ is constant, assuming lðTÞ varies linearly as:

lðTÞ ¼ l0 ½1  aðT  T 0 Þ;

ð10Þ

The thermal conductivity j is measured using the two experimental setups described in Section 2. 5.2.1. With the dedicated electronics The first step of the analysis using the dedicated electronics is to estimate the resistance variation of the fibers as a function of the thermalisation temperature for very small voltages applied on their edges (typically below 100 mV), heating the fiber about 10 K above T 0 . For such a voltage V a , the amplitude V f of the signal between t ¼ 0 and its asymptotic value is measured, as well as the difference DV r of the voltage signals at the edges of the same fiber when pulsed in series with the resistance r ¼ 2:65 X and without r (cf. Fig. 6). The variation of the resistance induced by a tension V a applied on the fibers is given by:

DRa ¼

Vf r DV r

ð12Þ

The value of the mean temperature to which the fiber is heated is then deduced, inverting the polynomial function used to fit the R(T) behaviour of the fibers presented in Section 5.1. For instance, above 40 K, the resistance is proportional to the temperature (cf. Section 5.1) and one gets from (6) and (10):

j¼

V 2a La 12SDRa

ð13Þ

The results are shown on Fig. 8 for seven superimposed fibers. j is well described in this temperature range by the parametrization:

j ¼ j0 þ j2 T 2 þ j3 T 3

ð14Þ

163

S. Henrot-Versillé et al. / Infrared Physics & Technology 52 (2009) 159–165 100

1100 1050

50 1000

(R-R(80)) / (dR/dT(80))

Resistance(Ohms)

950 900 850 800

0

-50

-100

750 700

-150

650 600

0

50

100

150

200

250

300

-200

10

10

2

Temperature(K)

Temperature(K)

Fig. 7. On the left: resistance of the fibers as a function of the thermalisation temperature (a polynomial fit is superimposed). On the right: comparison of the same data when properly normalized (cf. text).

Table 1 Measured flux for each Planck-HFI frequency band for the fibers installed on the mirror and pulsed with 1 V amplitude signals.

12

10

Flux ð103 pWÞ

100 GHz

143 GHz

217 GHz

353 GHz

545 GHz

11

10

21

117

27

8

6

Measured power on 100 GHz bolometer (pW)

4

0.012 2

0.01 0

0.008 0

50

100

150

200

250

300

Temperature (K) Fig. 8. Thermal conductivity j as a function of the temperature for seven carbon fibers. The full line corresponds to a fit according to Eq. (14).

the full line curve corresponds to a fit of data with: j0 ’ 0:3 0:1 W=K m; j2 ’ ð2:5  0:8Þ104 W=K3 m, and j3 ’ ð5:  3:Þ107 W=K5 m. These results are in good agreement with other measurements on carbon fibers [6]. The small slope of jðTÞ at low temperature makes the fiber signal very stable even if T 0 is not stabilized. For instance, with T 0 around 4 K, the fiber signal varies by less than one per mil for variations of the order of 100 mK on T 0 . 5.2.2. Bolometers data In order to cross-check this result, one makes use of the Planck HFI instrument in the Saturn set up. The mirror source fibers are pulsed with square signals at approximately 1 Hz. As soon as fiber temperature is higher than 40 K, the Rayleigh– Jeans approximation of the Planck emission law is better than 10% in the whole HFI frequency range. So, we make the assumption that the emission spectrum of the fibers is proportional to the mean fiber temperature. One computes the amplitude Sbolo of the resulting synchroneous signal measured on the bolometers (translated in terms of incident power on the detectors). Sbolo is proportional to the integral of the

0.006

0.004

0.002

0.2

0.4

0.6

0.8

1

Voltage on the fiber (V) Fig. 9. Incident power measured by a 100 GHz Planck-HFI bolometer (in pW) – without corrections for optical efficiency of the cold optics of the instrument – as a function of the voltage applied at the edges of the fibers for 2 different fibers: a 1 mm length fiber (in black) and a 0.6 mm one (in red). The superimposed fitted function corresponds to the paramatrization given in Eq. (17). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

temperature increase over the length of the fiber. Table 1 gives the measured values of the flux (in pW) with the bolometers for the fibers installed on the mirror and pulsed with 1 V amplitude signals. These values have to be multiplied by a factor 350 if one wants to estimate the flux when the fiber is directly facing the entrance of the horn of the cold optics in front of the bolometers. Fig. 9 shows the power measured by one of the bolometers (at 100 GHz) for two ‘‘mirror” fibers (a small 0.6 mm fiber and a longer 1mm one) as a function of the voltage applied on the fiber.

164

S. Henrot-Versillé et al. / Infrared Physics & Technology 52 (2009) 159–165

16

1.6

14

Normalised Time Constant

1.4

Time Constant (ms)

12

10

8

6

1.2

1

0.8 4

2

0

50

100

150

200

250

300

0.6

0

50

100

Temperature(K)

150

200

250

300

Temperature(K)

Fig. 10. On the left: time constant (in ms) as a function of the thermalisation temperature. On the right: same data but normalized to the mean value obtained for each fiber with T P 250 K.

-7

-7

-7

0.8

x 10 0.8

x 10 0.8

0.6

0.6

0.6

0.4

0.4

0.4

0.2

0.2

0.2

0

100

200

0

300

-7

100

200

300

-7

0

x 10 0.8

x 10 0.8

0.6

0.6

0.6

0.4

0.4

0.4

0.2

0.2

0.2

0

100

200

0

300

x 10 0.8

100

200

300

0

250

200

100

200

300

-7

Temperature(K)

x 10

150

100

50

100

200

300

0 0

0.2

0.4

0.6

0.8

1

1.2

Voltage(V)

-7

x 10 0.8

Fig. 12. Simulation of the mean temperature of the fiber as a function of the amplitude of the applied voltage for two fibers lengths.

0.6 0.4 0.2 0

100

200

300

Fig. 11. Heat capacity C p as a function of the temperature for several carbon fibers.

These results can be understood in the light of the framework given in Section 4. In the permanent regime, using (14) to parametrize jðTÞ and neglecting RðTÞ dependence, one gets from Eq. (3):

@2 @x2

j0 T þ

j2 T 3 3

þ

j3 T 4 4

! ¼

V2 : LSR

j2 3

T 3 ðxÞ þ

j3 4

T 4 ðxÞ ¼ AðxÞ þ c

Sbolo / G0 V 2 þ

ð16Þ

is found to fit accurately both experimental (see Fig. 9) and simulated data (see Section 5.4). Here, the applied voltages are higher that the ones used in the previous section, still the model fits accurately the data and both results are in very good agreement.

where:

AðxÞ ¼

V2 j2 j3 ðL2 =4  x2 Þ; and c ¼ j0 T 0 þ T 30 þ T 40 : 2LSR 3 4

Since here T 0 2 K, it can be neglected with respect to T. Hence,

c ¼ 0.

G2 2=3 G3 1=2 V þ V 3 4

ð15Þ

Since T ¼ T 0 for x ¼ L=2, one obtains after integration:

j0 TðxÞ þ

AðxÞ simplifies to AðxÞ ¼ V 2 =ð8lÞð1  ð2x=LÞ2 Þ. Hence, the solution TðxÞ is only a function of V and the reduced variable n ¼ 2x=L. For small V values, the first term dominates the left-hand side of (16), that reduces to (5), and TðxÞ has a parabolic shape. At medium V values, TðxÞ solution on most part of the fiber implies mainly the pffiffiffiffiffiffiffiffiffiffiffiffi T 3 term, and TðxÞ has a flatter profile with a 3 1  n shape. At high V value, the negative j3 term plays a higher role and is responsible for the increase in the TðxÞ slope. The phenomenological expression:

ð17Þ

5.3. Time constants and heat capacity With the dedicated electronics of Section 3.2, and applying only small voltages to the fibers, one can extract the fiber time constant

S. Henrot-Versillé et al. / Infrared Physics & Technology 52 (2009) 159–165

165

by fitting the exponential behaviour of V 1 ðtÞ in a wide range of thermalisation temperature T 0 . The results on s are illustrated on Fig. 10: on the left hand side the data are in ms while on the right there are normalized to the mean value obtained for each fiber with T P 250 K, showing that this behaviour is coherent from one fiber to the others. In the investigated temperature range, the fiber time constants present a minimum around 100 K, and rise slowly up to 300 K. The higher time constants measured for a temperature smaller than 50 K are responsible for long decay times at the end of the pulse. During the Planck-HFI calibration [3], this drawback has been corrected for using a small permanent current on the fiber in addition to the pulse. This permanent current allows to maintain most of the fiber at temperatures where the time constant remains small, and the fiber decay time is in the same range as the rise time. C p can be deduced from the measurement of the time constant and from the j modelling through relation (9). Fig. 11 shows individual results for seven fiber heat capacity. Parametric adjustments of the standard type C p =T ¼ C p0 þ T 2 C p2 are superimposed. A good qualitative agreement is met with expected values for carbon fiber at low temperature [7].

A detailed analysis of the temperature dependence of the resistance, the heat capacity C p , and the thermal conductivity j, of theses fibers has been given, based on measurements from 1:7 K to 300 K. A self consistent modelling of these parameters was obtained, ending up with a good agreement between simulations of the 1D heat equations and data measured on the 100 mK HFI’s bolometers. This well understood picture of the fibers behaviour makes them a very useful tool for FIR instrument characterization.

5.4. Simulation

References

A simple simulation code has been developed to numerically solve Eq. (3). Using the Crank Nicholson scheme [8] and introducing the fitted dependence on the temperature of R, j and C p , the mean temperature of the fiber can be computed as a function of the input voltage: the result is shown on Fig. 12 and is in good agreement with the bolometers measurements of Fig. 9. It probes the self consistency of the modelling of the fibers.

[1] Planck. The Scientific Programme - ESA-SCI(2005)1 N. Mandolesi et al., The ESA medium size mission for measurements of CBR anisotropy, Planet. & Space Sci. 43 (10/11) (1995) 1459. [2] J.M. Lamarre et al., The Planck High Frequency Instrument a third generation CMB experiment and a full sky submillimeter survey, New Astron. Rev. 47 (11– 12) (2003) 1017–1024. [3] F. Pajot et al., HFI Calibration Plan; 30/01/2002; Edition: 3, Revision: 0, PLPHZW-100061-IAS. This report is qualified as Public. [4] F. Pajot, J.-M. Lamarre, J.-L. Puget and the calibration team, HFI Calibration and Performances Document; 14/04/08; Issue: 02, Revision: 01, CA-PH412-600824IAS. This report is qualified as Public. [5] Habilitation à diriger des recherches Henrot-Versillé S., Archeops et Planck-HFI: Etudes des systématiques pour l’analyse du fond diffus cosmologique, Université Paris Sud – Paris XI URL: http://tel.archives-ouvertes.fr/tel-00102694/en/. [6] J. Heremans et al., Thermal conductivity and Raman spectra of carbon fibers, Phys. Rev. B 32 (1985) 10. [7] C. Pradère et al., Specific-heat measurement of single metallic, carbon, and ceramic fibers at very high temperature, Rev. Sci. Instrum. 76 (2005) 064901. [8] See, e.g. W.H. Press, S.A. Teukolsky, W.T. Vetterling, B.P. Flannery, Numerical Recipes in Fortran, second ed., Cambridge University Press, Cambridge, England, 1992. vol. 7, Sec. 19.2.

6. Conclusions This article describes the extraction of the physical parameters of the carbon fibers used in the characterization of HFI. Its has been shown that, thanks to their low time constant (6 10 ms for 1 mm fibers at 2 K), they can be used in a pulsed regime without introducing any electrical parasitic signal on HFI’s bolometers, and that their emission spectrum gives a significant amount of signal in the submm domain.

Acknowledgement We wish to acknowledge J.C. Vanel and C. Rosset for their help in our first attempts to cool down the fibers, B. Maffei and R. Sudiwala for providing us with material and manpower at Cardiff, J.P. Torre for his disponibility and his setup, O. Perdereau, J. Haissinski and S. Plaszczynski for useful discussions and help, and F. Pajot, P. Lami and the Saturne Cryostat team for the Planck-HFI calibration. This work has been funded by CNES as part of LAL contribution to HFI, under contract N 737/CNES/01/8961/00.