EIGHT YEARS GEO GROUND TESTING OF THERMAL ...esmat.esa.int/Materials_News/ISME09/pdf/3-Ground/S2...The SEMIRAMIS facility main characteristics are cleanliness (very low organic residual

EIGHT YEARS GEO GROUND TESTING OF THERMAL CONTROL COATINGS

J. Marco1, S. Remaury2, C. Tonon3

1 ONERA/DESP, 2 Av. E. Belin 31055 Toulouse Cedex 4, France

E-mail: [email protected], Phone N°: (33) 562 25 27 42, Fax N°: (33) 562 25 25 69

2 CNES DCT/TV/TH, 18 Av. E. Belin, 31401 Toulouse Cedex 9, France E-mail: [email protected], Phone N°: (33) 561 27 31 33, Fax N°: (33) 561 27 34 46

3 EADS ASTRIUM, 31 Rue des Cosmonautes, 31402 Toulouse Cedex 4, France

E-mail: [email protected], Phone N°: (33) 562 19 55 03, Fax N°: (33) 562 19 55 27

Keywords: SPACE RADIATION SIMULATION – THERMAL CONTROL COATINGS – SOLAR ABSORPTIVITY DEGRADATION

ABSTRACT To achieve and optimize the thermal control during long term spacecraft life, it is necessary to forecast the degradations (thermo-optical, mechanical and electrical) of thermal control materials. The ageing of these materials is investigated on ground using UV, protons, electrons and in-situ absorptivity measurements. CNES and EADS ASTRIUM undertook to do at ONERA a long term test of eight years GEO simulation to obtain not only comparative results but also to investigate long term saturation effects. The basic feature of the space simulation is to maintain the samples under vacuum during irradiations and measurements and to apply the space degrading components in sequence. The sequence was repeated five times in order to get the degradation kinetics, to enhance the synergism and investigate the effects of each space degradation component. In paints, there is in general a continuous degradation in UV-visible spectral range while in near IR, there are alternating degradations and recoveries which depend on particles nature or time elapsed before measurements. The degradation of silicate and silicone paints is higher with electrons than with UV and protons. It is not the case of films or materials where optical properties depend mainly on extreme surface degradation. Solar absorptivity degradations results of several tested materials - OSR, white paints, films and coverglass adhesives - are presented with spectral, time variability and recovery effects which depend on irradiation nature and dose.

1. INTRODUCTION

The study presented in this paper is based on a long term test simulation and on the results analysis on several materials. In the first part of the paper, the eight years GEO test conditions are detailed covering the test procedure, the

acceleration factors limitation, the reasons of sequential simulation of space parameters and the recovery effects. In the second part, the spectral and solar degradations of several tested materials – silicate and silicone paints, films and coverglass adhesives - are presented with spectral, time variability and opposite effects which depend on space degradation components.

2. LONG TERM TEST

A long-term test (5 months in vacuum) has been performed in the SEMIRAMIS facility in order to simulate eight years on North/South faces of a GEO satellite; the simulation reproduced vacuum, UV and particles exposure.

2.1 Facilities The SEMIRAMIS facility main characteristics are cleanliness (very low organic residual partial pressures in vacuum) and reliability (vacuum keeping of samples for several months). The cleanliness is achieved by the use of stainless steel, a majority of metallic seals and the exclusive use of cryogenic pumping units which are equipped with a gate valve. The reliability is increased by the implementation of each function in a different vacuum chamber separated by its own gate valve and redundant pumping units with an associated gate valve which is automatically closed upon failure. The facility is composed of two parts: - At the lower stage, there is the irradiation chamber receiving three irradiation ports, one for protons at the normal of the samples, one for electrons and UV at 30° of incidence and one for a radiant sample temperature measurement system housed in a lateral vacuum chamber. - At the upper stage, there is a measurement and safeguard chamber in which the sample holder can be introduced by a vertical translation from the lower stage

2

and connected to an optical measurement device stored in another lateral vacuum chamber.

The proton and electron beams are supplied by 2.5 and 2.7 MeV Van de Graaff accelerators respectively. The protons are obtained from pure hydrogen plasma and separated from the other charged species by a magnetic mass analysis after acceleration. In order to irradiate the samples, the protons are swept across the sample holder surface. In the case of the electrons, the beam is diffused through a thin aluminium window. The solar UV generator is based on a short arc Xenon 6500 W source with UV spectral distribution close to sun. The source emission is filtered to reject all visible and infrared light above 400 nm to reproduce only the sun UV outer the atmosphere emission. Multiple solar constants calculated in the range 200-380 nm are applied over the sample holder surface (1 Thekaekara solar constant is 9.46 mW/cm2). The in-situ measurement system is composed of a spectrophotometer which is connected to a mobile integrating sphere in vacuum. Reflectance spectra (250-2500 nm) of about 25 samples can be obtained without breaking vacuum between irradiation phase and reflectance measurement... The SEM measurements were performed with a Hitachi S 3400N SEM. The accelerating voltage was 20 KeV, the working distance was about 10 mm and the vacuum was 10-5 Pa.

2.2 Test conditions A vacuum lower than 1x10-6 mbar was obtained after one day pumping down period and it was 7 10-8 mbar at the end of the test. The sample holder temperature was maintained at 40°C for the test duration. As thermal control of the samples depend on the thermal contact of their rear face with the sample holder and on their thermal conductivity, the irradiated surface temperature changed upon the different test phases. Surface temperature was estimated using the radiant measurement system.

2.3 UV exposure

To do a space simulation in a reasonable time, it is required to accelerate the irradiations as far as possible without modifying the degradation mechanisms which could induce wrong results. This problem is that of the reciprocity principle [1] of equivalent sun hours (ESH). When using a non filtered Xenon light source, it is recommended [2] an acceleration factor limitation to 3 in order to limit the thermal charge on the materials. With our filtered Xenon source giving only UV (Fig. 1.), we justify as requested in [3] the use of a higher acceleration factor due to the thermal charge reduction

by a factor of 10. A total amount of 8896 ESH of UV (1112 ESH per year on N/S faces) was applied at a rate of 7 ± 0.5 Solar Constants (80 mW/cm2 incident power inducing thermal charges depending on specific materials absorption) and an irradiated area uniformity better than ±10%.

200 300 400Wavelength (nm)

Ener

gy

UV Simulation

Thekaekara

Fig. 1. UV simulation

2.4 Charged particles exposure

In GEO, the dose profile inside a material due to the absorption of trapped electrons and protons decreases rapidly versus the material thickness. Fig. 2. shows the space dose profile calculation (based on trapped particles evaluated with Pole models) corresponding to 1 year in GEO inside a material of density 1.5. The dose is mainly due to trapped protons above the inflection of the curve (at about 2 microns) and it is mainly due to electrons below. The space dose profile was simulated each time in the same order with one electron energy in the bulk and with two proton energies close to the surface with the following fluences and order for one year: - First: 1x1015 electrons/cm2 of energy 400 keV at a

flux of about 5.62x1010 electrons/cm2/s (9 nA/ cm2) (penetration at about 600 μm and incident power 4 mW/cm2).

- Second: 2x1014 protons/cm2 of energy 240 keV at a flux of about 1.25x1011 protons/cm2/s (20 nA/ cm2) in order to simulate the space dose in the UV reflective (UVR) coating (penetration at about 2 μm and incident power 5 mW/cm2).

- Third: 2x1015 protons/cm2 of energy 45 keV at a flux of about 1.25x1011 protons/cm2/s (20 nA/ cm2) in order to simulate the space dose close to the surface (penetration at about 0.8μm and incident power 1 mW/cm2).

The acceleration factors of charged particles are very high compared to UV but the thermal charge induced by incident power is comparatively weak.

3

1E+51E+61E+71E+81E+9

1E+10

1E-1 1E+0 1E+1 1E+2 1E+3Thickness (micron)

DO

SE (G

y)

Dose profile 1.0 yearGEOProtons 45 keV

Protons 240 keV

Electrons 400 keV

Material density 1.5

Fig. 2. Simulated dose profiles

2.5 Test procedure In space, UV and particles are applied simultaneously. Such a synergism is impossible in laboratory because particles are applied with much higher acceleration factors for technical constraints. The synergism is compensated by the repetition of UV and particles irradiations. The simulation was performed in several steps corresponding to 0, 1, 3, 5, 6.5 and 8 years: - At the 0 year step, the samples were mounted in the

facility and the effect of the pumping down from air to vacuum was evaluated with in-situ measurements, the last measurement before irradiations being done after 4 days under vacuum.

- At the 1 year step, separated measurements were done after each successive irradiation: UV, 400 keV electrons, 240 keV protons, 45 keV protons and after only a total of 15 and 27 extra days in vacuum .

- At the 3 years step, measurements were done after UV, after completion of particles irradiations and after only 3 extra vacuum days.

- At 5 and 6.5 years step, measurements were done after UV, after completion of particles irradiations and after only 3 or 4 extra vacuum days.

- At 8 years step, measurements were done after UV, after completion of particles irradiations and after a total of 13 and 19 extra vacuum days.

- Final measurements were done after the vacuum to air operations (return at atmospheric pressure with nitrogen and exposure to air).

In order to illustrate the test procedure and the recoveries taking place in different parts of the spectra, measurements of each step are given for the white paint PSB on Fig. 4. to Fig. 9. Solar absorptivity; αS, is calculated from the spectra in the range 250-2500 nm. The estimated uncertainty is ΔαS/αS = ±0.01.

2.6 Tested materials All samples were mounted on 20 x 20 mm aluminium substrates (stuck when possible) except transparent ones which are equipped with evaporated aluminium because in-situ measurements are exclusively reflectance measurements. The samples are described in Table 1.

Temperature control was insured by the contact of the rear face of the sample with the sample holder.

Material Description Evaporated aluminium

MTO, Aluminium on glass protected SiO2 coating

OSR QIOptiq, 100 μm bonded onto bare aluminium

White paint PSB

Map, Binder: potassium silicate, Pigment: Zn2TiO4, on aluminium substrate

White paint PSBN

CNES, Binder: potassium silicate, Pigment: treated ZnO, on aluminium substrate

White paint SG121FD

MAP, Binder: polydiméthylsiloxane, Pigment: treated ZnO, on aluminium substrate.

Conductive White paint PCBE

MAP, first conductive layer: polydiméthylsiloxane binder and CuAg pigment, White surface layer: polydiméthylsiloxane binder and treated ZnO pigment, on aluminium substrate.

White paint SG122FD

MAP, Binder: polydiméthylsiloxane, Pigment: treated ZnO, on aluminium substrate.

Adhesive film P224

Permacel, Kapton 25 μm, bonded with acrylic on aluminium substrate.

Adhesive film 3M92

3M, Kapton 25 μm, bonded with silicone on aluminium substrate.

Silicone resin DC93500

Dow Corning, 300 μm silicone resin bonded between two CMO glasses, on evaporated aluminium

Silicone resin Mapsil QS1123

MAP, 300 μm silicone resin bonded between two CMO glasses, on evaporated aluminium

Silicone resin Elastosil-S690

Wacker, 20 μm silicone resin bonded between two CMO glasses, on evaporated aluminium

Table 1. Samples list

3. TEST RESULTS

3.1 OSR

The solar absorptivity variation for OSR is given in Fig. 3. The final quite low solar absorption of 0.06 is due to a low initial absorption of 0.03 due to the surface UVR coating and to a very low degradation ΔαS = 0.03 that occurs mainly during the first simulated year.

4

0

0.02

0.04

0.06

0.08

0.1

0 2 4 6 8Duration (year)

Sol

ar a

bsor

ptiv

ity

Fig. 3. Solar absorptivity of OSR.

3.2 Silicate paints During the first step, the outgassing induces minor changes. At around 1900 nm, the water absorption peak disappears (Fig. 4.).

0

0.5

1

250 750 1250 1750 2250

Wavelength (nm)

Ref

lect

ance

Air

Vacuum 4d

Fig. 4. PSB reflectance: initial air to vacuum.

During the 1 year step, UV degradation is moderated comparatively to the bulk electron irradiation that takes place in the whole spectral range. Surface protons irradiations induce additional degradations near the spectral cut off at 390 nm and degradations or recoveries in the infrared range. Recoveries take place after 15 days under vacuum and the degradation seems to be stable after 27 days (Fig. 5.).

0

0.5

1

250 750 1250 1750 2250Wavelength (nm)

Ref

lect

ance

Vacuum 4d UV 1 yearEl 1 year Prot240 1yearProt45 1year 1year 15d 1year 27d

Fig. 5. PSB reflectance: 1 year step.

During the 3 years step, UV irradiations induce recoveries in the infrared range since the degradations continue under particles over the whole spectral range (Fig. 6.).

0

0.5

1

250 750 1250 1750 2250Wavelength (nm)

Ref

lect

ance

1year 27dUV 3 yearPart. 3 year 1d3 year 3d

Fig. 6. PSB reflectance: 3 years step.

Further degradations at 6.5 and 8 years take place mainly below 1900 nm. Recoveries take place when repeating final measurements after 2 and 13 days and they are stable after 19 days (Fig. 7.).

0

0.5

1

250 750 1250 1750 2250Wavelength (nm)

Ref

lect

ance

3 year 3d 5 year 5d6.5 year 6.5 year 4d8 year 8 year 2d8 year 13d 8 year 19d

Fig. 7. PSB Reflectance : 6.5 and 8 years step.

During the final air exposure, important recoveries take place (Fig. 8.).

0

0.5

1

250 750 1250 1750 2250Wavelength (nm)

Ref

lect

ance

8 year 19dNitrogenAir air 2d

Fig. 8. PSB reflectance : final vacuum to air step.

An overview of spectral degradations during the whole test (using the last measurement of each step) shows a saturation tendency (Fig. 9.).

5

0

0.5

1

250 750 1250 1750 2250Wavelength (nm)

Ref

lect

ance

Vacuum 4d 1year 27d3 year 3d 5 year 5d6.5 year 4d 8 year 19d

Fig. 9. PSB reflectance : 1 to 8 years overview.

The effects of solar absorptivity recoveries after irradiation under vacuum due to the final vacuum to air transition appear clearly in the degradation curves (Fig. 10.). PSBN is very much less degraded than PSB due to the zinc oxide pigment. PSBN spectra are different from PSB mainly in the infrared range (Fig. 11.).

0

0.1

0.2

0.3

0.4

0.5

0.6


Sol

ar a

bsor

ptiv

ity

PSB PSBN

Fig. 10. Solar absorptivity variation of silicate paints.

0

0.5

1

250 750 1250 1750 2250Wavelength (nm)

Ref

lect

ance


Fig. 11. PSBN reflectance : 1 to 8 years overview.

In addition to those thermo optical measurements, SEM inspections were performed on paints before and after irradiation. A change on the silicate paints was observed before and after irradiation (Fig. 12) With a low enlargement cracks are observables on the paint after irradiation (those cracks are not present before irradiation). With a bigger enlargement a change of paint surface morphology is observable : there is a more accentuated relief on the irradiated area.

There was probably an erosion on the sample surface due to irradiation.

Fig. 12. SEM pictures of the PSB paint

a- before irradiation (x500) b- after irradiation (x500)

c- before irradiation (x2000) d- after irradiation (x2000)

3.3 Silicone paints

Silicone paints are less sensitive to recoveries under vacuum than silicates paints (Fig. 13.) while degradation saturation is slower. SG122FD is less degraded than SG121FD due to its initial higher value. Its reflectance spectra are less degraded in the infrared range (Fig. 14. to Fig. 16.). Final recoveries due to the vacuum to air transition are important for the two paints.

0

0.1

0.2

0.3

0.4

0.5

0.6


Sol

ar a

bsor

ptiv

ity

SG121FDPCBESG122FD

Fig. 13. Solar absorptivity variation of silicone paints.

a b

c d

6

0

0.5

1

250 750 1250 1750 2250Wavelength (nm)

Ref

lect

ance


Fig. 14. SG121FD reflectance : 1 to 8 years overview.

0

0.5

1

250 750 1250 1750 2250Wavelength (nm)

Ref

lect

ance


Fig. 15. PCBE reflectance : 1 to 8 years overview.

0

0.5

1

250 750 1250 1750 2250Wavelength (nm)

Ref

lect

ance


Fig. 16. SG122FD reflectance : 1 to 8 years overview.

3.4 Empirical models applied on paints The most common semi-empirical models used to fit solar absorptivity degradation are based on saturated exponentials:

)1( /0ταα tSS e

−−= (1) This expression is related to the solution of the differential equation describing an 1 order chemical reaction:

nAktA ][/][ −=∂∂ , n = 1 (2)

where αS is proportional to a concentration [B] of increasing absorbing species that appear through the relationship:

[B] = [A0]-[A] (3) [A] initial concentration of species to be degraded. Expression (1) cannot represent degradation mechanisms in the case of heterogeneous materials where many species can be degraded, interact within each other and involve higher reaction orders. Furthermore, it requires a linear relationship between absorption and the measured parameter in this case the solar absorptivity value. The interest of using this expression in case of paints is to describe the saturation effects with a limited number of parameters and to present an asymptotic linear limit although the multi-scattering of light due to the pigments introduces a non linear relationship. In the case of PSB and SG121FD, this empirical model is applied on the last measured points after recoveries at steps 1 and 8 years (1 year 27 days, 8 years 19 days) and on corrected points at steps 3, 5, 6.5 years. These measured points were corrected by extrapolating at 27 days recovery using a mean value of recovery kinetics found at steps 1 and 8 years. A sum of 2 saturated exponentials (τ1 = 0.5 year, τ 2= 5.68 year) gave a mean square fit of PSB points with a correlation coefficient of 99.98 % (Fig. 17.). The extrapolated αS at 15 years is 0.57 and the asymptotic value at the infinite 0.60. The degradations are compared to those obtained in previous tests. During the 7 years GEO test [4], the measurements done with less than 1 day recovery time explains the higher values. The measurements of the 3 years GEO test [5] show equivalent degradation and recovery levels.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7


Sol

ar a

bsor

ptiv

ity

Meas & correctedModel (2 sat. Exp.)3 year GEO 20037 year GEO, 4th Sym 88

Fig. 17. PSB empirical model.

A sum of 2 saturated exponentials of nearly equal time constant was found in the case of SG121FD with a correlation coefficient of 99.63 % (Fig. 18.). It is equivalent to a single saturated exponential with τ = 4.5 years, (αS 15 years = 0.57, αS infinite = 0.59). The measurements of the 3 years GEO test show equivalent degradation and recovery levels.

7

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7


Sola

r abs

orpt

ivity

Meas & correctedModel (1 sat. Exp)3 year GEO 2003

Fig. 18. SG121FD empirical model.

3.5 Adhesive films

The adhesive films were stuck on the aluminium substrate for a good thermal conductivity, the warmest surface temperature was 46°C under UV. The reflectance spectra are very similar mainly determined by the Kapton film (Fig. 20. and Fig. 21.). The solar absorptivity of Kapton with silicone adhesive is more degraded than the one of Kapton with acrylic one (Fig. 19.). It could be due to the additional light reflected by the aluminium substrate and absorbed by the Kapton due to the lower absorption of the silicone adhesive. Mechanical tests are scheduled to evaluate the degradation effect on the adhesive films bonding on aluminium. Previous tests (with shorter simulation duration) showed a higher degradation for the acrylic adhesive than for the silicone one.

0.3

0.4

0.5

0.6

0.7


Sola

r abs

orpt

ivity

Adhesive film P224

Adhesive film 3M92

Fig. 19. Solar absorptivity variation of adhesive films.

0

0.5

1

250 750 1250 1750 2250Wavelength (nm)

Ref

lect

ance


Fig. 20. Adhesive P224 film reflectance : 1 to 8 years

overview.

0

0.5

1

250 750 1250 1750 2250Wavelength (nm)

Ref

lect

ance


Fig. 21. Adhesive 3M92 film reflectance : 1 to 8 years

overview.

3.6 Coverglass adhesives

The samples were composed of three silicone adhesives applied between two CMO 100 μm glasses. For CMO, the wavelength cut-off is 330 nm [6]. The DC 93500 and Mapsil QS1123 layers were 300 μm thick and Elastosil S690 layer was 20 μm thick. Each assembly was put onto a reflective evaporated aluminium during the test to enable the in-situ reflectance measurements. Due to bad thermal conduction with the sample holder (from the use of the rear evaporated aluminium), the irradiated surface temperature was about 70°C (for DC93500 and Mapsil QS1123) and 80°C (for Elastosil S690) under UV irradiation and about 35°C in other test phases instead of 40°C. Additional transmission measurements were performed before and after the tests. As shown in dose profile curves (Fig. 2.), proton irradiations are blocked into the first CMO glass whereas UV and electron go through the adhesive. First CMO glass degradation is probably negligible under UV and particles irradiations due to the applied fluences [6]. Without thickness correction, reflectance degradation is equivalent for DC93500 and Mapsil QS1123 (Fig. 22. and Fig. 23.) and a little higher for Elastosil S690 (Fig. 24.).

8

0

0.5

1

250 750 1250 1750 2250Wavelength (nm)

Ref

lect

ance


Fig. 22. DC93500 reflectance : 1 to 8 years overview.

The most absorbing peak at 2295 nm is representative of silicones (pointed by an arrow, Fig. 22. and 23.).

0

0.5

1

250 750 1250 1750 2250Wavelength (nm)

Ref

lect

ance


Fig. 23. Mapsil QS1123 reflectance : 1 to 8 years

overview.

The magnitude of this peak is equivalent for DC93500 and Mapsil QS1123 whereas it is weak for Elastosil S690.

0

0.5

1

250 750 1250 1750 2250Wavelength (nm)

Ref

lect

ance


Fig. 24. Elastosil S690 reflectance : 1 to 8 years

overview.

The ex-situ transmission measurements of DC93500 and Mapsil QS1123 were corrected at each wavelength by applying a proportional change of the corresponding absorption coefficient in order to obtain a spectrum for a 20 μm thickness instead of 300 μm (Fig. 25.). The peak at 2295 nm is consequently of equal amplitude for the three silicones and the degradation of Elastosil S690 seems more pronounced when applying this correction.

0

20

40

60

80

100

250 750 1250 1750 2250Wavelength (nm)

Tran

smis

sion DC93500 Before

QS1123 BeforeS690 BeforeDC93500 8 year GEOQS1123 8 year GEOS690 8 year GEO

Fig. 25. Comparison of silicones transmission

degradation.

4. CONCLUSIONS

Different materials like OSR, silicate and silicone paints, adhesive films and coverglass adhesives were tested during a five months laboratory space simulation representing 8 years in GEO. It shed light on degradations instability and on final air exposure recoveries that are major simulation constraints. The strong recoveries of the final vacuum to air transition which affects the solar absorptivity shows the necessity of performing measurements under vacuum or to get a precise knowledge on recovery mechanisms in order to take into account these effects if ex-situ measurements are done. The degradation of white paints (particularly silicates) is unstable after each irradiation step under vacuum. It shows that it is necessary to make several measurements to estimate the recovery kinetics and to obtain stable degradation values. The acceleration factors can also influence the recovery effects but it is not mentioned in this paper. Empirical models were applied to paints in order to describe the solar absorptivity degradations versus time and try to find saturation values. A maximal value of 0.6 was calculated. The optical degradation of adhesive films is higher for silicone adhesive than for acrylic. The degradation of CMO glasses stuck with DC93500, Mapsil QS1123 and Elastosil S690 were compared in severe conditions (due to the high transmission in the UV of CMO). The three adhesives present a low degradation compared to the environment and the impact on the silicon solar cell performance is limited for all of them.

REFERENCES

[1] Laszlo T S 1966, J. Environ. Sci., p. 13, Dec. 1966.

[2] ASTM E512, Standard Practice for combined Simulated Space Environment Testing of Thermal

9

Control Materials with Electromagnetic and Particulate Radiation, American Society for Testing and Materials, Philadelphia, USA.

[3] ECSS-Q-ST-70-06C, 31 July 2008, Space product assurance: Particle and UV radiation testing for space materials

[4] Marco. J., Paillous A., Levadou F., Combined Radiation Effects on Optical Reflectance of Thermal Control Coatings, 4th European Symposium on Spacecraft Materials in Space Environment, Toulouse,

France 1988.

[5] Marco J., Remaury S., Evaluation of Thermal Control Coatings Degradation in Simulated Geo-Space Environment, High Performance Polymers, vol. 16 177-196, 2004.

[6] Russell J., Jones J., Radiation testing of coverglasses, www.qioptiqspace.com/Data/Documents/Radiation_Testing_of_Coverglasses.pdf