Characterization of UV Irradiated Space Application Polymers by Spectroscopic Ellipsometry.
Orbital spacecraft such as the International Space Station (ISS) and Hubble Space Telescope (HST) are subject to the harsh Low Earth Orbit (LEO) environment. Neutral species of atomic oxygen, UV radiation, thermal cycling, and micrometeroid impacts all contribute to degradation of space-application materials [1-3]. Space system designs demand consideration of the LEO environment to ensure proper function, reliability, and lifetime.
Polymers are susceptible to altered chemistry from UV radiation. The absorption of UV radiation can promote breakage of important functional groups and structural bonds, such as C = C and C = O (2). It is important to understand the damage mechanisms to predict their long-term durability. When an organic molecule absorbs UV radiation it will be excited into higher energy states, and possibly dissociate bonds. Dissociated radical species can then participate in additional reactions.
The effects of UV radiation on three different polymers were studied. The polymers included: DuPont Kapton(R) H polyimide, polyarylene ether benzixnidazole (PAEBI) [4,5], and DuPont Teflon(R) fluorinated ethylene propylene (FEP) . Kapton(R) is used on spacecraft for structural support, aluminized Teflon(R) FEP is used on the Hubble telescope for thermal control, and PAEBI is an oxide-forming polymer with similar physical and optical properties as Kapton(R). The chemical structures of these polymers are given in Fig. 1.
In this study, spectroscopic ellipsometry (SE) is used to monitor changes in polymers induced by UV radiation. SE measures changes in polarization of light after reflection from a sample. This change in polarization is expressed in terms of two parameters, [psi] and [delta], defined by:
Tan ([psi]) [e.sup.i[delta]] = [r.sub.p]/[r.sub.s] (1)
where [r.sub.p,s] are the complex Fresnel reflection coefficients for light parallel and perpendicular to the plane of incidence, respectively [6-8]. The sample structure is modeled to determine the "theoretical" results expected. Unknown properties, such as film thickness or optical properties can be determined by "fitting" the model generated data to experimental data through a non-linear regression. The quality of the "fit" is interpreted by a weighted mean squared error (MSE) . It is also important to consider correlation between "fit" parameters. If parameters become strongly correlated, a unique solution can not be determined. Therefore, a "best fit" is one that achieves the lowest MSE with low parameter correlation.
Bulk optical constants of each polymer (each approximately 127 [micro]m thick) were obtained with a variable angle spectroscopic ellipsometer (VASE) over a spectral range from 190 to 1700 nm. The polymers were modeled as optically isotropic, however they are potentially anisotropic. To avoid complex anisotropic modeling, care was taken to orient each sample in the same direction for all measurements. Also, roughening the backside of the films with sandpaper eliminated backside reflections. By only collecting light reflected from the sample surface the effects of anisotropy were insignificant.
Each polymer was irradiated by UV light from a 150 watt Xe lamp in normal room ambient. In situ SE monitored the sample during irradiation, as seen m Fig. 2. Kapton(R) H and PAEBI are known to absorb water in air over time. Since the experiment was performed in air over a long enough period of time to absorb water the samples were exposed to room ambient for a long period of time to ensure saturation. In this way the samples would not be absorbing water during the experiment and changing properties.
The lamp power supply current and voltage were fixed at 8.9 amps and 17 volts, respectively, which coincide with settings during calibration. The calibrated spectral output of the lamp, measured by Opto-Cal. Inc., is given in Fig. 3. The polymers studied were significantly absorbing only in the UV region of the spectrum ([greater than] 3.1 eV) thus any modification of the polymers were assumed due to photons at energies greater than 3.1 eV. Photon energies required to dissociate many molecular bonds found in space application polymers are given by Dever (2).
The UV light was transmitted through a fiber optic bundle onto the polymers, with 25 mm from the end of the bundle to the sample. The in situ ellipsometer acquired data at eighty-eight wavelengths simultaneously, spanning a spectral range from 280 to 760 nm. After long-time UV irradiance, the optical data saturated (changed little with further irradiance) and ex situ VASE measurements were again taken over a spectral range from 190 to 1700 nm.
III. OPTICAL MODELING OF ELLIPSOMETRIC DATA
Optical constants of bulk and heavily UV affected layers were obtained from analysis of ex situ VASE data. Both sets of optical constants (bulk and heavily modified samples) were parameterized using three Gaussian oscillators. Each Gaussian oscillator is described in the data analysis software by an amplitude, a center energy, and a broadening term, plus an additive constant, [[epsilon].sub.[infinity]], to account for effects from outside the spectral range .
Bulk polymer optical constants (those determined before irradiation) remained fixed as "substrates" in irradiated polymer optical models. The UV radiation damage was initially modeled as a single homogeneous layer on top of the unaffected polymer substrate. This single-layer model did not provide a satisfactory "fit". Therefore, "graded layer" models that vary the optical constants throughout the damage layer were used. In this approach, the layer is subdivided into N sublayers (N = 15) of equal thickness. The effective dielectric function of each sublayer is determined from a linearly varying volume fraction of two constituent materials, in this case undamaged (bulk) and "heavily UV exposed" materials. The effective dielectric function, [epsilon], at a given depth from the surface, d, is given as:
[epsilon] = [f.sub.a](d)[[epsilon].sub.a] + (1 - [f.sub.a] = (d)) [[epsilon].sub.b] (2)
where [[epsilon].sub.a,b] are the dielectric functions of the two constituent materials and [f.sub.a] is the relative volume fraction of material a. The effective dielectric function, [epsilon], is related to the refractive index by:
[epsilon] = [[epsilon].sub.1] + i[[epsilon].sub.2] = [n.sup.2] - [k.sup.2] + i2nk (3)
where [[epsilon].sub.1] [[epsilon].sub.2] and n, k are the real and imaginary parts of the dielectric function and refractive index, respectively.
Linearly graded and exponentially graded layers were investigated. Both graded layer models use Eq 2 to obtain the dielectric function. For a layer thickness, t, the normalized depth from the surface is given as:
[delta] = d/t (4)
The volume fraction for linear grading is directly related to the depth:
[f.sub.a]([delta]) = [delta] (5)
Whereas, the volume fraction for an exponentially graded layer is given as:
[f.sub.a]([delta]) = [A.sub.s][e.sup.-B[delta]] (6)
where [A.sub.s] is the percent conversion at the surface and B is a constant. After long exposures, [A.sub.s] will equal 100.
Model simplicity, MSE, fit parameter 90% confidence limits, parameter correlation, and physical realism of all three models were considered in determining the best optical model to use. As seen in Table 1, the MSE was greatly reduced by using a graded layer over a single layer model in all cases. The MSE was only slightly reduced from a linearly graded to an exponentially graded model. This small reduction in MSE alone, probably was not enough to warrant use of the more complex exponential model. However, by looking at the 90% confidence limits, as well as at how physical the results were, we concluded that the exponential grading was the best model for this analysis. Based on the above discussion, results are now presented in detail using the exponential model.
The VASE data taken before irradiation were parameterized, as discussed above, using three Gaussian oscillators. To obtain the optical constants at the top surface after heavy UV irradiation, the exponentially graded layer model was used, in which the damaged material was also modeled with three Gaussian oscillators. The Gaussian parameter values (amplitude, center energy, broadening, and [[epsilon].sub.[infinity]]) for bulk polymer optical constants were used as starting values in the regression for the top layer parameters. Only the amplitudes of the first two longer wavelength oscillators, [[epsilon].sub.[infinity]], and the exponential B parameter were allowed to vary in the regression fit to the VASE data. Thus, center energy and broadening values for all three oscillators, and the shortest wavelength oscillator amplitude value were fixed at the bulk polymer values. This resulted in good fits to the data and minimal correlation.
For in situ SE data analysis, the exponential grading parameters ([A.sub.s] and B) were fit to the spectroscopic data taken at a series of exposure times. The optical constants for the bulk substrate and "heavily UV exposed" materials were obtained from ex situ VASE data, as discussed above, and were not allowed to vary.
IV. RESULTS AND DISCUSSION
As seen from Fig. 4, the index of refraction and extinction coefficient decreased when PAEBI was irradiated. Data were fit using three oscillators centered at 190, 280, and 330 nm. The 190 nm Gaussian oscillator amplitude of the modified layer was fixed at the bulk PAEBI value because there was no significant change with irradiance. Regression analysis results showed significant decreases in oscillator amplitudes at 280 and 330 nm. Furthermore, the regression fit was significantly improved by varying the [[epsilon].sub.[infinity]] model parameter. Varying this parameter causes a broad band shift of the index up or down. A comparison of bulk and UV affected PAEBI refractive indices at 633 nm, oscillator amplitudes, and [[epsilon].sub.[infinity]] are given in Table 2.
Data analysis suggests a broad band decrease in the index ([[epsilon].sub.[infinity]] decreased). This is likely due to the material becoming less dense and/or the polymer heating up and water evaporated out leaving small pockets of void. Photons of high enough energy to dissociate molecular bonds within the PAEBI molecule (some dissociation energies given in reference 2) were striking the PAEBI, thus altering the chemistry and changing the materials density. In addition, the PAEBI is known to absorb water and high intensity light incidence on the sample is likely to have some heating effect. However, the contributing factor of each to the change in the ellipsometric data was not determined. Separating the effect of each of these factors is worthwhile for further experiments.
Amplitude decreases in UV absorption bands (seen in Fig. 4) are likely due to higher energy radiation dissociating molecular bonds. Molecular unit structures absorb at selective wavelengths, so small molecular structural changes can bring about perceptible changes in the UV absorption spectra. The most important electronic energy changes occurring when polymers absorb UV radiation include 1) transitions between bonding orbitals and antibonding orbitals or 2) promotion of non-bonding electrons (unshared electrons) into antibonding sigma or antibonding pi orbitals (11). Low photon fluxes below 300 nm suggests most effects are probably due to excitation of pi-electrons.
Analysis of in situ ellipsometry data provided refractive index depth profiles for increasing UV exposure. The index depth profiles (determined from regression fits to Eq 6) for various exposure times at a wavelength of 500 nm are given in Fig. 5. The index exhibits a steady decrease until reaching saturation, with the saturation index at the surface being 1.71 compared to the bulk index of 1.815 at 500 nm wavelength. The damage depth, at saturation, is approximately 250 nm from the surface.
As seen in Fig. 6, a plot of the "[A.sub.s]" fit parameter from Eq 6 (% of surface converted to heavily UV affected optical constants) shows that UV radiation effects reach a near steady-state condition at approximately 375 minutes of exposure. This is a small fraction of the exposure plastics receive in long-duration space flights. However, the photon penetration depth is shallow at high absorption wavelengths, which for most plastics is in the UV region. Thus, UV radiation will dissociate all allowed bonds within that penetration depth and then no further damage is expected. Figure 6 shows that UV damage progresses towards the expected saturation.
B. Kapton(R) H Polyimide
Kapton(R) was irradiated for approximately 33 hours, with in situ ellipsometry data acquired during the first 762 minutes only. Figure 7 compares Kapton(R) and UV affected Kapton(R) optical constants. Similar to the PAEBI, UV radiation appeared to cause decreases in the index of refraction over all the measured spectrum ([[epsilon].sub.[infinity]] decreased), as well as the amplitudes of the UV absorption bands (Gaussian oscillators) at 290 and 220 nm. A comparison of the measured values is given in Table 2.
The index of refraction depth profile at a wavelength of 500 nm, seen in Fig. 8, shows little effect until approximately 100 minutes of irradiation (there was no measurable change before that). At 100 minutes of irradiation, the index only decreased by approximately 0.008. At saturation, the index at the surface decreased by approximately 0.066, and there were measurable effects nearly 500 nm into the film.
Kapton(R) took longer to reach saturation than PAEBI. The times required were 1970 minutes and 375 minutes, respectively. A possible reason is that PAEBI has an absorption band in the region corresponding to the largest UV photon flux in the Xenon lamp spectrum (300 to 400 am). Figure 9 is a plot of the extinction coefficient and penetration depths in the UV spectral region of both Kapton(R) and PAEBI for comparison. From this Figure it is seen that UV photons will penetrate deeper into the Kapton(R), and thus dissociation of more molecular bonds is required to reach saturation. This is also why the damaged layer is nearly twice as deep in the Kapton(R) as in the PAEBI. Figure 10 shows that the Kapton(R) damage layer requires nearly 2000 minutes (5.25 times longer than PAEBI) to begin to saturate.
C. Teflon(R) FEP
There was little or no change in the UV irradiated Teflon(R) experimental data after 300 minutes of exposure. The Teflon(R) appears to be less susceptible to UV damage. However, actual Teflon samples used as a thermal control blanket on Hubble telescope have shown considerable damage [12,13]. The Hubble Teflon(R) thermal blanket is exposed to visible and ultraviolet photon radiation as well as soft X-rays, electrons, protons, and thermal cycling. From the present experiments and analysis one may conclude that photons energies less than 6.2 eV are not the main degradation factor of the Hubble Teflon(R). It is likely due to one or a more of the other sources.
By combining transmission intensity and ellipsometric data, very small absorption at wavelengths between 190 and 300 nm were measured (k [approximately equals] 1 X [10.sup.-4]). Thus, very little energy is being absorbed at wavelength greater than 190 nm and only after very long exposure times would one expect to see any change in the Teflon(R). Other constituents present in Low Earth Orbit likely will cause degradation much quicker. A recommended future experiment would be to expose FEP Teflon(R) to the other energized particles present in LEO and try to separate out the effects of each.
Ellipsometric analysis of irradiated space-application polymers helped produce better understanding of degradation mechanisms and their time dependencies involved. Ellipsometric data taken on samples after UV exposure were fit to a layer model that graded the index of refraction exponentially. From this analysis the degree of degradation of these polymers due to photons greater than 200 nm wavelength was quantified. Ellipsometric analysis before and after exposure to UV radiation provided optical constants for the bulk and UV treated polymers over a spectral range of 190 to 1700 nm. Dynamic ellipsometric measurements, taken during exposure, provided information about the degree, depth and time dependence of degradation.
UV radiation appears to cause decreases in the index of refraction and extinction coefficient in both PAEBI and Kapton(R). The damage appears to reach saturation within approximately 370 and 1970 minutes for PAEBI and Kapton(R), respectively. These exposure times are much lower than for long duration space flights. However, once the materials reach saturation, further significant damage due to the UV radiation is not expected. The changes observed from the UV radiation (photon energies [less than] 6.2 eV) are not likely significant enough to threaten the lifetime of a spacecraft.
After approximately 300 minutes of radiation, Teflon(R) showed no measurable change. By combining ellipsometric and transmission data, very small k values were measured. Absorption in Teflon(R) is very low for non-vacuum UV thus very little energy is absorbed to dissociate molecular bonds. However, at long enough exposure times, some changes may occur. These possible changes are also not likely to endanger the life-time of a spacecraft since it would take very long exposure times for changes to take place. The extensive cracking and embrittling observed in Hubble Telescope Teflon(R) samples are likely due to other the species present in Low Earth Orbit (higher energy photons, soft X-rays, electrons, protons, etc.). One or more of these constituents appear to cause degradation much quicker than the photon energies used in the present work.
Research funded by NASA Lewis contract #NAG3-2086 and NASA EPSCOR contract #NCC5-169. Triton Systems, Inc., provided samples of polyarylene ether benzimidazole. DuPont Corp. provided samples of Kapton(R) polyimide and Teflon(R) FEP.
(1.) "U.S. Standard Atmosphere, 1976," NASA TM-X-74335 (1976).
(2.) J. A. Dever, "Low Earth Orbital Atomic Oxygen and Ultraviolet Radiation Effects on Polymers," NASA Tech. Memo. 103711 (1991).
(3.) B. A. Banks, K. de Groh, S. K. Rutledge, and F. J. DiFfllippo, "Prediction of In-space Durability of Protected Polymers Based on Ground Laboratory Thermal Energy Atomic Oxygen," NASA Tech. Memo. 107209 (1996).
(4.) A. Shepp, R Haghighat, J. Lennhoff, P. Schuler, T. L. St. Clair, J. Connell, J. Vaughn, and J. Zwiener, "TOR an COR AO-VUV Resistant Polymers for Space, ICPMSE-3 Protection of Materials and Structures from the Low Earth Orbit (LEO) Space Environment, Proceedings of the Third International Conference, Toronto. Canada, April 25-26, 1996, J. Klesman, editor.
(5.) A. Shepp, R. Haghighat, J. Lennhoff, P. Schuler, T. L. St. Clair, J. Connell, J. Vaughn, and J. Zwiener, "TOR and COR AO-VUV Resistant Polymers for Space," Proceedings of the 1997 Space Contamination and Coatings Workshop, Annapolis, Maryland, July 9-10, 1997, NASA conference publication 3361, P. Chen, editor.
(6.) R. M. A. Azzam and N. M. Bashara, Ellipsometry and Polarized Light, North Holland Publishing, New York (1977).
(7.) J. A. Woollam, B. Johs, W. A. McGaban, P. G. Snyder, J. Hale, and H. W. Yao, Mat Res. Soc. Symp. Proc., 324, 15 (1994).
(8.) J. A. Woollam and P. G. Snyder, in Encyclopedia of Material Characterization, p. 401, Butterworth-Heinemann, Publishers, Boston (1992).
(9.) J. Jellison, Proceedings of the First international Conference on Spectroscopic Ellipsometry, Thin Solid Films, 234, 416 (1993).
(10.) F. Wooten, Optical Properties of Solids, Academic Press, New York (1972).
(11.) S. F. Dyke, A. J. Floyd, M. Sainbury, and R. S. Theobald, Organic Spectroscopy, p. 13. Longman Group Limited, London (1971).
(12.) T. M. Zuby, K. K. de Groh, and D. C. Smith, "Degradation of FEP Thermal Control Materials Returned from the Hubble Space Telescope," NASA Tech. Memo. 104627 (1995).
(13.) K. K. de Groh and D. C. Smith, "investigation of Teflon FEP Embrittlement on Spacecraft in Low Earth Orbit," Proc. 7th Int. Symp. on Mat, in Space Env., p. 255, Toulouse, France (1997).
Three Optical Models are Compared to Determine the Best One to Use to Describe UV Damage. The Resulting Fit Parameter Values, MSE, 90% Confidence Limit, and Parameter Correlation are Included in the Table and Compared. PAEBI Fit Parameters Time (min) Thickness (nm) 90% Confidence Single Layer 10.7 1.6 0.6 50.4 17.4 1.2 100.3 26.1 1.3 200.2 33.7 1.3 300.2 38.3 1.3 Thickness (nm) Linear Grading 10.7 43.7 9.6 50.4 238.0 6.7 100.3 244.9 5.2 200.2 297.1 5.2 300.2 310.1 7.1 B Parm. Exponential Grading 10.7 fix at 8, 0.000 no sensitivity 50.4 9.2 1.1 100.3 7.0 0.6 200.2 5.8 0.4 300.2 7.3 0.5 Kapton(R) Thickness (nm) Single Layer 100.7 11.0 1.0 301.2 28.3 0.9 531.7 37.2 0.9 731.5 41.023 0.9 Thickness (nm) Linear Grading 100.7 no sensitivity 301.2 76.8 8.7 531.7 105.1 4.7 731.5 104 3.5 B Parm. Exponential Grading 100.7 12.4 1.5 301.2 46.4 1.4 531.7 71.6 1.8 731.5 82.9 1.6 PAEBI MSE Correlation Single Layer 9.5 none 28.3 none 33.7 none 37.8 none 38.8 none Surface Node % 90% Confidence Linear Grading 11.3 3.1 6.4 none 39.6 1.0 8.4 none 59.9 1.1 9.3 none 75.8 1.2 11.1 none 86.8 1.9 17 none [A.sub.s] Parm. (%) Exponential Grading 4.9 0.9 7.3 none 41.1 1.0 6.7 none 61.6 0.7 5.4 none 79.0 0.6 5.3 none 92.2 1.1 8.4 none Kapton(R) Single Layer 2.45 none 4.7 none 7.3 none 8.26 none Surface Node % Linear Grading no sensitivity yes 28.5 3.9 2.62 a little 60.3 2.2 4.9 very little 68.8 1.9 4.9 very little [A.sub.s] Parm. (%) Exponential Grading 10.1 6.1 1.9 yes 7.5 1.8 2.44 very little 10.2 1.2 4.3 very little 10.8 0.9 4.2 very little Comparison of Bulk and UV Affected Polymer Gaussian Oscillator Amplitudes (Osc. Amp.) at Their Corresponding Center Energies (C.E.) Show Decreases in Some UV Absorption Bands. The Irradiation Caused Decreases in the Index of Refraction at 633 nm for Both PAEBI and Kapton(R). index [epsilon] Osc. C. E. 1 Osc. [lambda] = 633 nm infinity Amp. 1 (eV) Amp. 2 Bulk PAEBI 1.765 1.68 0.500 3.628 0.2328 UV Affected PAEBI 1.680 1.63 0.1666 3.628 0.1412 Bulk Kapton(R) 1.790 1.69 0.2275 4.325 0.8070 UV Affected Kapton(R) 1.726 1.64 0.0597 4.325 0.7431 C. E. 2 Osc. C. E. 3 (eV) Amp 3 (eV) Bulk PAEBI 4.782 0.5429 5.984 UV Affected PAEBI 4.782 0.5429 5.984 Bulk Kapton(R) 5.587 2.287 6.316 UV Affected Kapton(R) 5.587 2.287 6.316
|Printer friendly Cite/link Email Feedback|
|Author:||BUNGAY, COREY L.; TIWALD, THOMAS E.; DEVRIES, MICHAEL J.; DWORAK, BRAD J.; WOOLLAM, JOHN A.|
|Publication:||Polymer Engineering and Science|
|Article Type:||Brief Article|
|Date:||Feb 1, 2000|
|Previous Article:||Kinetic Parameters Estimation for Unsaturated Polyester-Styrene Systems.|
|Next Article:||Effects of Gauge Length and Strain Rate on Fracture Toughness of Polyethylene Terephthalate Glycol (PETG) Film Using The Essential Work of Fracture...|