Spectroscopic Investigations on Thin Adhesive Layers in Multi-Material Laminates Yuliya Voronko,a,b Boril S. Chernev,c Gabriele C. Edera,* a b c
OFI Austrian Research Institute for Chemistry and Technology, 2700 Vienna, Austria Institute of Materials Science and Technology, Vienna University of Technology, 1040 Vienna, Austria Institute for Electron Microscopy and Nanoanalysis, Graz University of Technology, Steyrergasse 17, A-8010 Graz, Austria
Three different spectroscopic approaches, Raman linescans, Raman imaging, and attenuated total reflection Fourier transform infrared spectroscopy (ATR FT-IR) imaging were evaluated for the visualization of the thin adhesive layers (3–6 lm) present in polymeric photovoltaic backsheets. The cross-sections of the multilayer laminates in the original, weathered, and artificially aged samples were investigated spectroscopically in order to describe the impact of the environmental factors on the evenness and thickness of the adhesive layers. All three methods were found to be suitable tools to detect and visualize these thin layers within the original and aged polymeric laminates. However, as the adhesive layer is not very uniform in thickness and partly disintegrates upon weathering and/or artificial aging, Raman linescans yield only qualitative information and do not allow for an estimation of the layer thickness. Upon increasing the measuring area by moving from one-dimensional linescans to two-dimensional Raman images, a much better result could be achieved. Even though a longer measuring time has to be taken into account, the information on the uniformity and evenness of the adhesive layer obtainable using the imaging technique is much more comprehensive. Although Raman spectroscopy is known to have the superior lateral resolution as compared with ATR FT-IR spectroscopy, the adhesive layers of the samples used within this study (layer thickness 3–6 lm) could also be detected and visualized by applying the ATR FT-IR spectroscopic imaging method. However, the analysis of the images was quite a demanding task, as the thickness of the adhesive layer was in the region of the resolution limit of this method. The information obtained for the impact of artificial aging and weathering on the adhesive layer obtained using Raman imaging and ATR FT-IR imaging was in good accordance. Index Headings: Vibrational spectroscopy; Raman microscopy; Raman imaging; Infrared spectroscopy; ATR FT-IR imaging; Photovoltaic; Backsheet; Adhesive.
INTRODUCTION Thin layers of functional materials are widely present in material composites as thin barrier layers in packaging films, adhesive layers in multi-material laminates, or corrosive protection for metallic layers, to name a few. As the functionality of the whole multi-material composite is often critically dependent on the reliable and efficient performance of these thin layers, the knowledge of their structure, chemical identity, thickness, or compactness is crucial in product development, quality control, failure analysis, and lifetime assessment. Received 16 September 2013; accepted 6 December 2013. * Author to whom correspondence should be sent. E-mail: gabriele. eder@ofi.at. DOI: 10.1366/13-07291
584
Volume 68, Number 5, 2014
The work described is focused on the characterization of thin adhesive layers in multilayer laminates that are commercially used as photovoltaic (PV) backsheets.1 The role of the backsheet within the multi-material composite of a PV module is to provide protection for the photovoltaic active layer against environmental, mechanical, and chemical influences and to ensure electrical insulation.2 The backsheets are usually comprised of a core layer (e.g., polyester (PET) or polyamide), which provides the mechanical stability and electrical isolation. This core is laminated on both sides with a thin adhesive film to highly protective fluoropolymer outer layers.3,4 The adhesiveness of the layers within the backsheet5,6 is of high importance for the functionality of the backsheet over a demanded lifetime of up to 25 years. Thus, it is of great interest to investigate the changes in the adhesive layer induced by the impact of environmental factors such as radiation, temperature, humidity, and corrosive gases on the material and ply adhesion within the multilayer laminate.4,5,6 Among the analytical tools used to visualize and characterize these thin layers within the polymeric laminates in the original and aged state, spectroscopic imaging techniques such as attenuated total reflection Fourier transformed infrared (ATR FT-IR) imaging and Raman imaging seem to be the most promising methods.7 These techniques combine the spatial information (microscopy) with the chemical information (vibrational spectroscopy) and, thus, might yield a complete description on the variances in the compactness, thickness, and chemical structure of the adhesive upon aging. We will evaluate whether a correlation of the results derived from molecular spectroscopy to macroscopic properties (e.g., ply adhesion and/or mechanical characteristics or the elongation at break) of the backsheet exists.
EXPERIMENTAL Material. According to the datasheets, the two types of backsheets chosen for this study are comprised of a PET core layer (which provides the mechanical stability and electrical isolation of the multilayer material) laminated with an adhesive to weathering resistant outer layers of a fluoropolymer: (i) polyvinyl fluoride (PVF, Tedlar) = sample 1, and (ii) polyvinylidene fluoride (PVDF) = sample 2. As an example, the light microscopic picture of the cross-section of sample 1 is given in Fig. 1. The composition of the laminates, as derived from light microscopy, can be described as a symmetrical structure with a core layer from approximately 250 lm
0003-7028/14/6805-0584/0 Q 2014 Society for Applied Spectroscopy
APPLIED SPECTROSCOPY
FIG. 1. Microscopic picture of the cross-section of backsheet 1.
(sample 2) to approximately 260 lm (sample 1) laminated with the outer layers. These in turn show an approximate thickness of 30 and 40 lm for samples 1 and 2, respectively. The adhesive layer is difficult to resolve optically. For the microscopic and spectroscopic investigations, polished cross-sections of the samples, embedded in an epoxide resin, were prepared. Environmental Simulation/Artificial Aging. The backsheets were characterized in the original state and after treatment in various weathering and artificial aging experiments where temperature, humidity, and corrosive gas (ammonia, NH3) were applied as isolated or combined influencing factors. For the ultraviolet (UV)-weathering experiments, the samples were exposed to UV-A irradiation according to the standard EN-ISO 4892-3 (Plastics—Methods of Exposure to Laboratory Light Sources—Part 3: Fluorescent UV Lamps).8 The irradiance of the UV fluorescence lamp was 0.72 W/m2 at 340 nm. The Xe weathering (simulated solar light spectrum) of the samples was performed corresponding to standard EN-ISO 4892-2/ method A (Plastics—Methods of Exposure to Laboratory Light Sources—Part 2: Xenon-Arc Lamps).8 The irradiance was 60 6 2 W/m2 (300–400 nm). For the artificial aging experiments, the samples were stored (i) at elevated temperature, and (ii) at increased humidity for 2000 h. The two influencing factors were then combined in a so-called ‘‘damp heat test’’ according to standard ISO 61 215 (Crystalline Silicon Terrestrial Photovoltaic (PV) Modules—Design Qualification and Type Approval; 10.13).9 Furthermore, the effect of corrosive gases was evaluated by storing the backsheets under dry or humid ammonia at 80 8C. The conditions used in the various weathering and aging experiments as well as the abbreviations used for the samples throughout the paper are summarized in Table I. Confocal Raman Microscopy. Raman linescans and Raman images of the cross-sections of the original and aged backsheets were recorded using a confocal Raman system Horiba Jobin Yvon LabRam 800 HR with visible laser excitation (473 and 514 nm) and motorized XYZ stage. The detection system used was a charge coupled
device camera (1 inch chip with 1024 3 256 pixel) with Peltier cooling. The spatial resolution obtainable in xand y-direction with such a setup is below 1 lm.10–12 For the imaging experiments described in this paper, an Olympus MPlan N objective with magnification of 1003 and a numerical aperture = 0.9 was used, the pinhole was closed to 400 lm, the spectral slit set to 100 lm, the integration time per pixel was 100 ms. The spectra were recorded in the spectral region 100–4000 cm1 with approximate spectral resolution of 5 cm1 (300 lines/mm grating). For the linescans, the spectra were recorded in the spectral range 100–3200 cm1 with approximate spectral resolution of 2 cm1 (600 lines/mm grating). The pinhole was closed to 300 lm, the spectral slit set to 100 lm, the integration time per pixel was 2 s. Averaged spectra from the various layers of the backsheet crosssections could be extracted from the linescans and images. In order to determine the thickness of the adhesive layer within the backsheet laminate, first linescans with a length between 15 and 25 lm and an increment of 1 lm were recorded. The sampling line was chosen in a way that the adhesive layer was placed in the middle of the line. To get additional information on the uniformity and compactness of the adhesive layers, Raman images of the laminated regions were measured. The image size was 30 lm 3 50 lm, the step size for the x and y increment was 1 lm. The linescans and images were evaluated in two ways: (i) either by integrating the area of a characteristic band for each component or (ii) by manually extracting the full spectrum for each component from the image and modeling the image with this spectral dataset, using a classical least squares (CLS) fitting. The resulting CLS model reveals the component distributions in the Raman images. The thickness of the adhesive layer was then determined by fitting the intensity profile of the adhesive band/spectrum using a Gaussian curve and determining the full width at half maximum (FWHM). For the Raman images, the average thickness value of at least five measuring points was calculated. Infrared Microscopy. Attenuated total reflection FT-IR images of the cross-sections of the backsheets were performed using a Perkin Elmer Spotlight 400 equipped with an ATR FT-IR imaging device and liquid nitrogen cooled mercury–cadmium–telluride array detector (linear array detector with 16 individual, gold-wired infrared detector elements). The ATR FT-IR imaging system is comprised of a germanium (Ge) crystal with a flattened tip (d 600 lm). For the measurements presented in this paper an image size of 500 lm 3 100 lm was chosen, which allowed for an image over the whole profile of the backsheet. According to the specifications of the manufacturer, the spatial resolution achievable with this equipment is below 5 lm (3.1 lm at 1700 cm1).13,14 The experimentally determined spatial resolution with this type of system (as measured by van Dalen et al.), using polymer multilayers for wavenumbers ranging from 1000 to 1730 cm1, was about 4 lm.15 The spatial resolution in that paper15 was calculated as the FWHM of the fitted first derivative of the absorbance profile. However, we applied a more stringent approach, as described by
APPLIED SPECTROSCOPY
585
TABLE I. Weathering and aging conditions.
Sample
Treatment
Temperature (8C)
Relative humidity (%)
Irradiation or NH3 concentration
Time (h)
S S_DH S_D S_H S_UVa S_Xea S_A S_DA
Original Damp heat Damp Heat UV weathering Xenon weathering Ammonia storage dry Ammonia storage wet
— 85 — 85 — — 80 80
— 85 85 — — — — 95
— — — — 340 nm 300–2500 nm 0.5% NH3 in N2 0.5% NH3 in N2
0 1000, 2000 1000, 2000 1000, 2000 1000 1000 1000 1000
a
For the spectroscopic imaging measurements the irradiated part of the backsheet was chosen.
Kazarian and Chan16,17 and Sommer18 where the distance between points corresponding to 5% and 95% of the maximum absorbance profile was taken as the actual spatial resolution. Best values for the actual spatial resolution of micro-ATR FT-IR imaging systems with Ge crystals could be as high as 4 lm16,17 and 8 lm18 at 1700 cm1. With this test we obtained with the equipment described above a practical resolution of about 9 lm at 1510 cm1, which is in good agreement with the results reported for the same system by Patterson and Havrilla,19 who found a minimum spatial resolution at the center of the hemisphere of the ATR crystal to be 10 lm. Averaged spectra of the different layers of the crosssection were extracted from the images. For the evaluation, the ATR FT-IR images were on the one hand treated with principle component analyses (PCA) resulting in a false color representation, where each spectroscopically distinguishable component is displayed in a different color. On the other hand, the images were analyzed using a spectra comparison correlation, where each spectrum of the image is set in comparison to a given reference spectrum, in our case, the adhesive spectrum in the original state. However, as the adhesive spectrum could not be extracted from the images in its pure form due to its limited thickness, the contributions of the neighboring layers had to be spectrally subtracted. The averaged thickness of the adhesive layer was determined from the grayscale representation of the images (after spectral comparison correlation) with the DigitalMicrographTM (Gatan, Inc.) software package. A mean value for the thickness was determined over the whole height of the image (100 lm).
in the profile of the samples. As an example, Figs. 2a and 2b show the profiles obtained for the PET, the adhesive, and the PVF layer obtained for sample S1_DH1000 by applying two different types of evaluation. First, one characteristic band in the spectra of each of the three components/layers was chosen, and the integrated area of these bands was plotted as a function of the scanned distance (Fig. 2a): the aromatic vibration at 1610 cm1 for the PET layer, the CH-stretching vibration at 2925 cm1 for the adhesive, and the TiO2 mode at 610 cm1 for the PVF layer filled with titania (TiO2) (see rectangles in Fig. 2c). However, as in the spectral region of the aliphatic CH vibrations (2800–3000 cm1), all three components showed absorptions, and no clear separation of the layers could be obtained (Fig. 2a). Thus, another type of evaluation, making use of a model based on the extraction of full spectra for the three independent compounds, was applied. The profiles calculated on the basis of these extracted spectra for
RESULTS AND DISCUSSION The thin adhesive layers interconnecting the PET core layer with the weather-resistant fluoropolymer outer layers of the PV backsheet were investigated with different spectroscopic techniques. After preparing cross-sections of the samples, Raman linescans, Raman images, and ATR FT-IR images of the profile of the laminated layers were recorded. The changes in the chemical structure, thickness, and compactness of the adhesive layer induced by the impact of environmental factors such as radiation, temperature, humidity, and corrosive gases were then assessed using various evaluation processes. Confocal Raman Microscopy. Raman Linescans. Raman linescans were recorded over the adhesive layer
586
Volume 68, Number 5, 2014
FIG. 2. Linescan profiles obtained (a) upon integration of one characteristic band (1610 cm1/PET layer, 2925 cm1/adhesive, and 610 cm1 for the PVF layer filled with titania) and (b) by applying a model based on extracted spectra; (c) Raman spectra of the individual layers, S1_DH1000h.
FIG. 4. Linescan profiles obtained by applying a model based on extracted spectra of samples (a) and (c) S1_UV and (b) S1_Xe. FIG. 3. (a) Linescan profiles calculated using a four-component model and (b) extracted Raman spectra; S2.
the PET, adhesive, and PVF/titania layer are given in Fig. 2b, the extracted spectra in Fig. 2c. With S2, however, we had to face a more complicated situation, as it became obvious from the first results that the PVDF outer film of this backsheet consists of three individual layers: PVDF/PVDF filled with titania/PVDF. Thus, four independent components (PVDF, PVDF plus titania, adhesive, and PET) had to be taken into account for the evaluation. As the CH-stretching bands used as characteristic bands for the adhesive are also the main bands in the pure PVDF layer and, additionally, the two other components show absorptions in this spectral region (see Fig. 3b), a separation of the individual layers just by choosing specific bands was not possible. Consequently, all evaluations were performed using the model calculating the linescan profiles from extracted spectra. These profiles obtained from the modeled linescan as well as the extracted spectra for S2 are given in Figs. 3a and 3b, respectively. By comparing the linescan profiles obtained for the adhesive layers after the various weathering and accelerated aging procedures and determining the thickness of the layer at half height of maximum, clear changes in shape and width of the adhesive bands could be observed upon aging (three typical examples are shown in the Fig. 4). Furthermore, the adhesive layer was in some cases partially split in two or sometimes more maxima upon accelerated aging. The partial overlap of the profiles obtained for the PET (green line in Fig. 4) and the adhesive (red line in Fig. 4) layers suggests that in the artificially weathered samples, S1, a partial transport of the adhesive into the PET layer took place. In these cases, it is very difficult to estimate the adhesive thickness correctly. Furthermore it became obvious that the thickness of the adhesive layer was not uniform over the sample length but rather heterogeneous with deviations of up to
625%. By measuring, e.g., ten linescans of the sample S_A on different sample positions, we obtained values for the thickness of the adhesive layer of 4.8 6 1.2 lm. As this deviation is high in comparison with the effects we expect to see upon weathering and accelerated aging, we decided to record images to get a better overview of the inhomogeneity of the adhesive and to be able to derive more reliable results on the adhesive thickness. Raman images would allow for the at-once observation of a larger sampling area and, thus, permit the direct inspection of the thickness deviations of the adhesive layers. Raman Images. Raman images with dimensions of 30 lm 3 50 lm were recorded in the laminated areas of the backsheets and evaluated in a first step in a comparable manner to the Raman linescans by choosing one characteristic band from the spectrum of each component/layer: the aromatic vibration at 1610 cm1 for the PET, the CH-stretching vibration at 2925 cm1 for the adhesive, and the band at 610 cm1 for the PVF layer filled with titania (TiO2) of sample 1. In a second step, the evaluation was made by means of a model using extracted full spectra for each component. The comparison of the images obtained with the two evaluation approaches is shown exemplarily for samples S1 and S1_DH1000 in Fig. 5. It was observed that the adhesive within the S1 samples stayed intact as a uniform layer within the selected sample area of about 30 lm, and the averaged thickness is only slightly decreased upon prolonged damp heat storage (4.5 lm ! 4.3 lm for 1000 h ! 3.8 lm for 2000 h; model evaluation). The Raman images of the samples S2, S2_A, and S2_DA obtained after evaluation with the model (extracted Raman spectra) are given in Fig. 6. Upon treatment with dry ammonia (S2_A) the adhesive layer was found to stay intact, while the storage of the sample in humid ammonia atmosphere (S2_DA) led to a strong decline in the layer thickness, from 4.3 lm to 2.6 lm, as well as a partial overlap with the adjacent PVDF material.
APPLIED SPECTROSCOPY
587
FIG. 5. Comparison of the Raman images obtained (a) and (c) using evaluation of characteristic bands and (b) and (d) by applying a model based on extracted spectra (a) and (b) for S1, and (c) and (d) S1_DH1000.
The values derived for the averaged thicknesses of the adhesive layers in the original and aged backsheets S1 with the two evaluation approaches are summarized in Fig. 7. It can be shown that for this backsheet (uniform
FIG. 6. Raman images of samples (a) S2, (b) S2_A, and (c) S2_DA evaluated via the model.
588
Volume 68, Number 5, 2014
outer layer, three independent components) both evaluation methods yield satisfactory results. Especially with the aged samples, the images obtained upon model evaluation gave more homogeneous layers. This in turn allows for the more reliable determination of the averaged layer thicknesses. Weathering with UV-light (S1_UV) and storage in a humid ammonia atmosphere (S1_DA) seem to have the strongest impact on the adhesive layer. The results obtained for backsheet S2 after evaluation with the model based on extracted spectra are presented in Fig. 8. With these samples (with four components) the evaluation through band integration was not possible (see above). While some treatments, such as xenon weathering (S2_Xe) or storage in ammonia containing
FIG. 7. Results of the thickness of the adhesive layer of backsheet S1 in the original and aged state as derived from the Raman images (blue = evaluating using characteristic bands, red = model with extracted spectra).
FIG. 8. Results of the thickness of the adhesive layer of backsheet 2 in the original and aged state as derived from the Raman images.
atmosphere (S2_A), seem to have no effect on the thickness of the adhesive layer, UV exposure (S2_UV), storage under damp heat (S2_DH), or humid ammonia atmosphere (S2_DA) cause a reduction of the adhesive layer thickness by approximately one-third. The Raman spectra of the aged adhesive layer show significant fluorescence, especially with the samples after damp heat treatment. Therefore and in order to obtain more detailed information on the chemical changes occurring within these layers upon aging, we decided to additionally record ATR FT-IR images of these samples. Infrared Microscopy. Attenuated Total Reflection Fourier Transform Infrared Imaging. In addition to the Raman images, ATR FT-IR images of the cross-sections of the backsheets were collected. The sample area was chosen in a way that the whole profile of the sample could be visualized within one image (500 lm 3 100 lm).
FIG. 9. ATR FT-IR image (a) after PCA and (b)–(d) single contributions of three components (red = PET layer; blue = PVF layer; green = adhesive).
FIG. 10. ATR FT-IR images of samples S1 after performing spectra comparison correlation.
A typical example (S1) for the obtained images after PCA as well as the images of the principal components contributions is given in Fig. 9. Although the thin adhesive layer (thickness 5 lm) could be detected in the original sample, the adhesive could not be visualized satisfactorily after PCA or band integration after some accelerated aging tests. Thus, for the evaluation of the thickness and compactness of the adhesive layers after weathering and/or artificial aging, a different kind of data treatment had to be chosen: we evaluated the ATR FT-IR images via a compare correlation with a reference adhesive spectrum. However, as the thickness of the adhesive layers has comparable dimensions with the diffraction limited spatial resolution of ATR FT-IR microscopes13–19 (see above), the isolated spectrum of the adhesive cannot be directly extracted from the ATR FT-IR images of the cross-sections of these backsheet samples. The spectrum of the unaltered adhesive was, thus, extracted from the ATR FT-IR image after PCA analysis of the sample in the original state (S1 or S2), and the contributions of the neighboring layers (PET and PVF or PVDF, respectively) were spectrally subtracted. The spectrum of the pure adhesive in the original state then was taken as reference spectrum for the spectra comparison correlation analysis of the ATR FT-IR images of the various aged samples. This evaluation procedure was performed for all samples; the resulting images are
APPLIED SPECTROSCOPY
589
FIG. 11. Extracted and corrected spectra of the adhesive for sample S1 after different aging conditions.
presented in part in Fig. 10. Treatment with heat (S1_H), humidity (S1_D), ammonia gas storage (S1_A), and Xenon exposure (S1_Xe) had no strong effect on the appearance of the adhesive layer in the image, while storage under damp heat conditions (S1_DH2000), humid ammonia atmosphere (S1_DA), or UV exposure (S1_UV) caused a drastic change in the degree of conformance of the adhesive with the reference adhesive spectrum (Fig. 10). Thus, the ATR FT-IR spectra of the region of the adhesive between the PET and fluoropolymer layers were extracted from the ATR FT-IR images of the aged samples, and the contributions of the adjacent layers were subtracted. The resulting spectra of the aged/ weathered adhesive were then compared with the reference spectrum (for sample S1, see Fig. 11) in order to get information on the aging-induced changes in the chemical structure of the adhesive. It becomes visible that upon aging spectral changes have occurred in the spectra of the adhesive (especially in the spectral region between 1150 cm1 and 1750 cm1) upon storage under combined damp/heat and damp/ammonia conditions as well as after UV irradiation (Fig. 11). From the FT-IR spectra we suggest that the original adhesive has to contain an ester–urethane compound (e.g., at 1727, 1707, 1530, and 1226 cm1, Fig. 11) but also shows additional absorptions typical for epoxide-based adhesives (1610, 1510, and 830 cm1). Therefore we consider that the adhesive is based on a polyester–urethane polymer modified with an epoxide (on the basis of bisphenol A). While the IR spectrum of the adhesive stays nearly unchanged for the samples S1_A, S1_D, and S1_H, few changes were observed in the ester bands (1727 and 1226 cm1) for samples S1_DH1000 and S1_Xe. Strong degradation of the carbonyl group at 1727 cm1 is observed upon humid ammonia storage (S1_DA), prolonged damp heat treatment (S1_DH2000), and UV weathering (S1_UV). The urethane absorption at 1707 cm1 stays clearly visible after all artificial aging procedures, but decreases in intensity upon UV weath-
590
Volume 68, Number 5, 2014
FIG. 12. ATR FT-IR images of samples S2 after performing spectra compare correlation.
ering. The bands at (1610) 1510, 1464 1240, 1180 and 1118 cm1, to which we attribute an epoxide modification (particularly bisphenol A), dominate the degraded adhesive spectra. For sample S2, the resulting ATR FT-IR images of the adhesive layer (after performing spectral compare correlation) are shown in Fig. 12. The application of individual environmental factors such as heat (S2_H), humidity (S2_D), or corrosive gas (S2_A) had no strong effect on the appearance of the adhesive layer. For the weathered samples S2_Xe and S2_UV and the artificially aged samples S2_DH and S2_DA, however, a change in the evenness and thickness of the adhesive was observed, partially strongly different for the two adhesive layers within the backsheet. Especially with the samples S2_UV and S2_DA, a partial disintegration of the adhesive layer was observed with strongly varying layer thicknesses. The average thickness of the adhesive layers in the ATR FT-IR images, obtained after comparison correlation with the original reference spectrum, was evaluated using an image analysis procedure. The mean values of the layer thickness (averaged over the image height to equal 100 lm) obtained from the grayscale images of the various samples, are presented in Figs. 13 (S1) and 14 (S2). With this type of evaluation, where the spectral conformity with the original adhesive is analyzed, a spreading of the adhesive is observed for part of the
FIG. 13. Adhesive thickness estimated from the ATR FT-IR images for samples S1.
FIG. 14. Adhesive thickness estimated from the ATR FT-IR images for samples S2.
samples, e.g., S_DH, S_UV, and S_DA. These findings together with the results obtained for the chemical changes in the adhesive structure upon aging suggest that the unchanged polyester–urethane basis of the adhesive is partially transported into the neighboring layers while the epoxide component is enriched in the interlayer region. It seems that especially this epoxide layer is detected in the Raman images as the individual adhesive layer. In addition, especially with backsheet S2, the two adhesive layers showed nonuniform aging effects within the sample, making the results strongly dependent on the measuring field chosen for the imaging measurement. The smaller the chosen measurement field is, the stronger this effect gets: Raman linescan . Raman image . ATR FT-IR image.
technique is much higher. For the analysis of the Raman linescans and images, two different approaches were evaluated. In principle both methods, analysis of the Raman spectra using band integration of individual bands and using a model making use of extracted spectra, yield suitable Raman images. For the linescans, however, the analysis using the model was clearly favorable. Especially in the case when each of the independent components of the sample does not show a characteristic absorption band usable for the evaluation using band integration, the analysis using a model based on extracted spectra instead of individual bands proved to be the more constructive and reliable approach. Although Raman spectroscopy is known to have the superior lateral resolution as compared with ATR FT-IR spectroscopy,10–25 the adhesive layers of the samples used within this study (layer thickness 3–6 lm) could also be detected and visualized by applying the ATR FTIR imaging method. However, the analysis of the images was quite a demanding task, as the thickness of the adhesive layer was in the magnitude of the resolution limit of this method, and the FT-IR spectra of the pure adhesives could not be directly extracted from the images. In addition, to make things worse, the spectrum of the adhesive showed strong similarities to the spectrum of the next neighboring layer, PET. By applying alternative methods for the data evaluation, such as principal component analysis, spectral subtraction of the disturbing contributions of the neighboring layers, and using spectra comparison correlation with reference spectra, we also obtained reliable images for the adhesive layers within the various samples with this method. The information obtained for the impact of artificial aging and weathering on the adhesive layer obtained using Raman imaging and ATR FT-IR imaging was, however, complementary.20,21,26,27 As two completely different approaches were used for the evaluation of the images, namely, (i) analysis with a model using extracted spectra from the image for the Raman images and (ii) spectra comparison correlation with a reference spectrum of the adhesive in the original state for the ATR FT-IR images, different information for the aging-induced
CONCLUSION Three different spectroscopic approaches, Raman linescans, Raman imaging, and ATR FT-IR imaging, were evaluated for the visualization of the thin adhesive layers (3–6 lm) present in polymeric PV backsheets. The cross-sections of the multilayer samples in the original, weathered, und artificially aged samples were investigated spectroscopically in order to describe the impact of the environmental factors on the evenness and thickness of the adhesive layers. All three methods were found to be suitable tools to identify and detect the thin adhesive layer within the original and aged polymeric laminates. However, as the adhesive layer is not very uniform in thickness and partly disintegrates upon weathering and/or artificial aging, Raman linescans yield only qualitative information and do not allow for an estimation of the layer thickness, as deviations of up to 6 25% in the adhesive thickness were determined when measuring ten independent positions within one sample. Upon increasing the measuring area by moving from one-dimensional linescans to two-dimensional Raman images, a much better result could be achieved. Although a higher measuring time has to be taken into account, the information on the uniformity and evenness of the adhesive layer obtainable using the imaging
APPLIED SPECTROSCOPY
591
changes could be derived from the spectroscopic images. A broadening/spreading of the adhesive (polyester–urethane compound) becomes evident upon evaluation of the ATR FT-IR images after spectral comparison correlation, especially with those samples (DH and UV) where the adhesive is severely degraded (hydrolyzed) during aging. The epoxide part of the adhesive, however, seems to be enriched in the interlayer region. For the evaluation of the Raman images, the signal of this epoxide fraction seems to be the main spectral feature. Thus, a decrease in the adhesive layer thickness upon aging is detected with this method. In the case of chemical alteration, the Raman spectra of the adhesive only show increasing fluorescence upon accelerated aging (especially after damp heat treatment). The IR spectra thereof, however, were a valuable source for information regarding the aging induced chemical changes (e.g., degradation of the ester group) occurring within these layers. The fact that the samples showing strong changes in the thickness and chemistry of the adhesive layer (after damp heat test, ammonia/water storage, and after UV irradiation), as derived from molecular spectroscopy, are the same samples showing strong changes in macroscopic properties such as ply adhesion and/or mechanical characteristics, such as, e.g., the elongation at break, seems to be quite promising. The sample showing the strongest delamination behavior within the backsheet is S2_DH2000, also S2_DA shows a light tendency to delamination. ACKNOWLEDGMENT The present work was funded by the project ‘‘Analysis of PV Ageing’’ of the Austrian Klima- und Energiefonds, ‘‘Neue Energien 2020’’. The Raman microscope was financially supported by the European Regional Development Fund (EFRE), the Graz University of Technology, and the Government of Styria. 1. S. Kurtz, J. Wohlgemuth, T. Sample, M. Yamamichi, J. Amano, P. Hacke, M. Kempe, M. Kondo, T. Doi, K. Otani. ‘‘Ensuring Quality of PV Modules’’. Paper presented at: 37th IEEE Photovoltaic Specialists Conference (PVSC). Seattle, WA; June 19-24, 2011. Print ISBN 978-1-4244-9966-3: 842-847. 2. W. Gambogi, S. Kurian, B. Hamzavytehrany, J. Trout, O. Fu, Y. Chao. ‘‘The Role of Backsheet in Photovoltaic Module Performance and Durability’’. Paper presented at: 26th EU PVSEC. Hamburg, Germany; September 5-9, 2011. 3. C.E. Packard, J.H. Wohlgemuth, S.R. Kurtz, U. Jahn, K.A. Berger, T. Friesen, M. Koentges. ‘‘Fielded PV Module Condition’’. Workshop presented at: 27th EU PVSEC Parallel Event Workshop from IEA PVPS Task 13: ‘‘Characterising and Classifying Failures of PV Modules’’. Frankfurt, Germany; September 24-28, 2012. 4. A.J. Steiner, W. Krumlacher, H. Muckenhuber, M. Kraxner. ‘‘PV Backsheet Materials under Accelerated Aging Conditions: A Performance Study’’. Paper presented at: 26th EU PVSEC. Frankfurt, Germany; September 5-9, 2011. 5. G. Oreski, G.M. Wallner. ‘‘Delamination Behavior of Multi-Layer Films for PV Encapsulation’’. Sol. Energy Mater. Sol. Cells. 2005. 89: 139-151. 6. E.J. Lightfoot, M. Debergalis, W.J.M. Gambogi, B. Fu, A.Z. Bradley, T.J. Trout. ‘‘Adhesion in Photovoltaic Modules’’. Paper presented at: 34th Annual Meeting of the Adhesion Society USA. Savannah, GA; February 13-16, 2011. 7. K.J. Geretschla¨ger, G.M. Wallner, J. Fischer. ‘‘Raman and Infrared Microscopical Analysis of Multilayer Backsheets’’. Paper presented at: 27th EU PVSEC. Frankfurt, Germany; September 24-28, 2012.
592
Volume 68, Number 5, 2014
8. International Organization for Standardization. ‘‘Plastics: Methods of Exposure to Laboratory Light Sources. Part 2: ‘‘Xenon-Arc Lamps’’ O¨NORM EN ISO 4892-2; Part 3: ‘‘Fluorescent UV Lamps’’. O¨NORM EN ISO 4892-3. 2006. 9. International Organization for Standardization. ‘‘Crystalline Silicon Terrestrial Photovoltaic (PV) Modules–Design Qualification and Type Approval: 10 Test Procedures, 10.13 Damp-Heat Test’’. ISO 61215. 2005. 10. N.J. Everall. ‘‘Confocal Raman Microscopy: Performance, Pitfalls, and Best Practice’’. Appl. Spectrosc. 2009. 63(9): 245A-262A. 11. N.J. Everall. ‘‘Confocal Raman Microscopy: Why the Depth Resolution and Spatial Accuracy Can Be Much Worse Than You Think’’. Appl. Spectrosc. 2000. 54(10): 1515-1520. 12. K.R. Allakhverdiev, D. Lovera, V. Altstadt, P. Schreier, L. Kador. ‘‘Confocal Raman Microscopy: Non-Destructive Materials Analysis with Micrometer Resolution’’. Rev. Adv. Mater. Sci. 2009. 20: 77-84. 13. Perkin Elmer. Application note: ‘‘Large Area ATR-FTIR Imaging Using the Spotlight FT-IR Imaging System’’. 2006. http://www. perkinelmer.com/ CMSResources/Images/44- 74854TCH_ LargeareaATRFT-IRImaging.pdf [accessed Sept 16 2013]. 14. Perkin Elmer. Technical note: ‘‘Spatial Resolution in ATR-FTIR Imaging: Measurement and Interpretation’’. 2006. http://www. perkinelmer.ca/CMSResources/Images/44-74872TCH_ SpatialResolutioninATRFT-IRImaging.pdf [accessed Sept 16 2013]. 15. G. van Dalen, P.C.M. Heussen, R. den Adel, R.B.J. Hoeve. ‘‘Attenuated Total Internal Reflection Infrared Microscopy of Multilayer Plastic Packaging Foils’’. Appl. Spectrosc. 2007. 61(6): 593-602. 16. K.L.A. Chan, S.G. Kazarian. ‘‘New Opportunities in Micro- and Macro-Attenuated Total Reflection Infrared Spectroscopic Imaging: Spatial Resolution and Sampling Versatility’’. Appl. Spectrosc. 2003. 57(4): 381-389. 17. S.G. Kazarian, K.L.A. Chan. ‘‘Micro- and Macro-Attenuated Total Reflection Fourier Transform Infrared Spectroscopic Imaging’’. Appl. Spectrosc. 2010. 64(5): 135A-152A. 18. A.J. Sommer, L.G. Tisinger, C. Marcott, G.M. Story. ‘‘Attenuated Total Internal Reflection Infrared Mapping Microspectroscopy Using an Imaging Microscope’’. Appl. Spectrosc. 2001. 55(3): 252256. 19. B.M. Patterson, G.J. Havrilla. ‘‘Attenuated Total Internal Reflection Infrared Microspectroscopic Imaging Using a Large-Radius Germanium Internal Reflection Element and a Linear Array Detector’’. Appl. Spectrosc. 2006. 60(11): 1256-1266. 20. B. Chernev, P. Wilhelm. ‘‘Vibrational Spectroscopy and Spectral Imaging Techniques—Special Applications in the Polymer Science’’. Monatsh. Chem. 2006. 137(7): 963-967. 21. P. Wilhelm, B.S. Chernev. ‘‘Microscale Chemical Imaging Using Vibrational Spectroscopy Methods’’. In: A. Mendez-Vilas, J. Diaz, editors. Microscopy: Science, Technology, Applications and Education. Badajoz, Spain: Formatex Research Center, 2010. Vol. 3. Pp. 2062-2071. 22. P. Wilhelm, B. Chernev. ‘‘Spatial Imaging/Heterogeneity’’. In: J.M. Chalmers, R.J. Meier, editors. Compr. Anal. Chem. 2008. 53: 527560. 23. A. Gupper, P. Wilhelm, M. Schmied, S.G. Kazarian, K.L.A. Chan, J. Reußner. ‘‘Combined Application of Imaging Methods for the Characterization of a Polymer Blend’’. Appl. Spectrosc. 2002. 56(12): 1515-1523. 24. K.L.A. Chan, S.G. Kazarian. ‘‘Detection of Trace Materials with Fourier Transform Infrared Spectroscopy Using a Multi-Channel Detector’’. Analyst. 2006. 131(1): 126-131. 25. S.G. Kazarian, K.L.A. Chan. ‘‘ATR-FTIR Spectroscopic Imaging: Recent Advances and Applications to Biological Systems’’. Analyst. 2013. 138(7): 1940-1951. 26. G.C. Eder, L.S. Lukacic, B.S. Chernev. ‘‘Visualisation and Characterisation of Ageing Induced Changes of Polymeric Surfaces by Spectroscopic Imaging Methods’’. Anal. Bioanal. Chem. 2012. 403(3): 683-695. 27. B.S. Chernev, G.C. Eder. ‘‘Spectroscopic Characterisation of the Oligomeric Surface Structures on Polyamide Materials Formed During Accelerated Ageing’’. Appl. Spectrosc. 2011. 65(10): 11331144.
OFI Austrian Research Institute for Chemistry and Technology, 2700 Vienna, Austria Institute of Materials Science and Technology, Vienna University of Technology, 1040 Vienna, Austria Institute for Electron Microscopy and Nanoanalysis, Graz University of Technology, Steyrergasse 17, A-8010 Graz, Austria
Three different spectroscopic approaches, Raman linescans, Raman imaging, and attenuated total reflection Fourier transform infrared spectroscopy (ATR FT-IR) imaging were evaluated for the visualization of the thin adhesive layers (3–6 lm) present in polymeric photovoltaic backsheets. The cross-sections of the multilayer laminates in the original, weathered, and artificially aged samples were investigated spectroscopically in order to describe the impact of the environmental factors on the evenness and thickness of the adhesive layers. All three methods were found to be suitable tools to detect and visualize these thin layers within the original and aged polymeric laminates. However, as the adhesive layer is not very uniform in thickness and partly disintegrates upon weathering and/or artificial aging, Raman linescans yield only qualitative information and do not allow for an estimation of the layer thickness. Upon increasing the measuring area by moving from one-dimensional linescans to two-dimensional Raman images, a much better result could be achieved. Even though a longer measuring time has to be taken into account, the information on the uniformity and evenness of the adhesive layer obtainable using the imaging technique is much more comprehensive. Although Raman spectroscopy is known to have the superior lateral resolution as compared with ATR FT-IR spectroscopy, the adhesive layers of the samples used within this study (layer thickness 3–6 lm) could also be detected and visualized by applying the ATR FT-IR spectroscopic imaging method. However, the analysis of the images was quite a demanding task, as the thickness of the adhesive layer was in the region of the resolution limit of this method. The information obtained for the impact of artificial aging and weathering on the adhesive layer obtained using Raman imaging and ATR FT-IR imaging was in good accordance. Index Headings: Vibrational spectroscopy; Raman microscopy; Raman imaging; Infrared spectroscopy; ATR FT-IR imaging; Photovoltaic; Backsheet; Adhesive.
INTRODUCTION Thin layers of functional materials are widely present in material composites as thin barrier layers in packaging films, adhesive layers in multi-material laminates, or corrosive protection for metallic layers, to name a few. As the functionality of the whole multi-material composite is often critically dependent on the reliable and efficient performance of these thin layers, the knowledge of their structure, chemical identity, thickness, or compactness is crucial in product development, quality control, failure analysis, and lifetime assessment. Received 16 September 2013; accepted 6 December 2013. * Author to whom correspondence should be sent. E-mail: gabriele. eder@ofi.at. DOI: 10.1366/13-07291
584
Volume 68, Number 5, 2014
The work described is focused on the characterization of thin adhesive layers in multilayer laminates that are commercially used as photovoltaic (PV) backsheets.1 The role of the backsheet within the multi-material composite of a PV module is to provide protection for the photovoltaic active layer against environmental, mechanical, and chemical influences and to ensure electrical insulation.2 The backsheets are usually comprised of a core layer (e.g., polyester (PET) or polyamide), which provides the mechanical stability and electrical isolation. This core is laminated on both sides with a thin adhesive film to highly protective fluoropolymer outer layers.3,4 The adhesiveness of the layers within the backsheet5,6 is of high importance for the functionality of the backsheet over a demanded lifetime of up to 25 years. Thus, it is of great interest to investigate the changes in the adhesive layer induced by the impact of environmental factors such as radiation, temperature, humidity, and corrosive gases on the material and ply adhesion within the multilayer laminate.4,5,6 Among the analytical tools used to visualize and characterize these thin layers within the polymeric laminates in the original and aged state, spectroscopic imaging techniques such as attenuated total reflection Fourier transformed infrared (ATR FT-IR) imaging and Raman imaging seem to be the most promising methods.7 These techniques combine the spatial information (microscopy) with the chemical information (vibrational spectroscopy) and, thus, might yield a complete description on the variances in the compactness, thickness, and chemical structure of the adhesive upon aging. We will evaluate whether a correlation of the results derived from molecular spectroscopy to macroscopic properties (e.g., ply adhesion and/or mechanical characteristics or the elongation at break) of the backsheet exists.
EXPERIMENTAL Material. According to the datasheets, the two types of backsheets chosen for this study are comprised of a PET core layer (which provides the mechanical stability and electrical isolation of the multilayer material) laminated with an adhesive to weathering resistant outer layers of a fluoropolymer: (i) polyvinyl fluoride (PVF, Tedlar) = sample 1, and (ii) polyvinylidene fluoride (PVDF) = sample 2. As an example, the light microscopic picture of the cross-section of sample 1 is given in Fig. 1. The composition of the laminates, as derived from light microscopy, can be described as a symmetrical structure with a core layer from approximately 250 lm
0003-7028/14/6805-0584/0 Q 2014 Society for Applied Spectroscopy
APPLIED SPECTROSCOPY
FIG. 1. Microscopic picture of the cross-section of backsheet 1.
(sample 2) to approximately 260 lm (sample 1) laminated with the outer layers. These in turn show an approximate thickness of 30 and 40 lm for samples 1 and 2, respectively. The adhesive layer is difficult to resolve optically. For the microscopic and spectroscopic investigations, polished cross-sections of the samples, embedded in an epoxide resin, were prepared. Environmental Simulation/Artificial Aging. The backsheets were characterized in the original state and after treatment in various weathering and artificial aging experiments where temperature, humidity, and corrosive gas (ammonia, NH3) were applied as isolated or combined influencing factors. For the ultraviolet (UV)-weathering experiments, the samples were exposed to UV-A irradiation according to the standard EN-ISO 4892-3 (Plastics—Methods of Exposure to Laboratory Light Sources—Part 3: Fluorescent UV Lamps).8 The irradiance of the UV fluorescence lamp was 0.72 W/m2 at 340 nm. The Xe weathering (simulated solar light spectrum) of the samples was performed corresponding to standard EN-ISO 4892-2/ method A (Plastics—Methods of Exposure to Laboratory Light Sources—Part 2: Xenon-Arc Lamps).8 The irradiance was 60 6 2 W/m2 (300–400 nm). For the artificial aging experiments, the samples were stored (i) at elevated temperature, and (ii) at increased humidity for 2000 h. The two influencing factors were then combined in a so-called ‘‘damp heat test’’ according to standard ISO 61 215 (Crystalline Silicon Terrestrial Photovoltaic (PV) Modules—Design Qualification and Type Approval; 10.13).9 Furthermore, the effect of corrosive gases was evaluated by storing the backsheets under dry or humid ammonia at 80 8C. The conditions used in the various weathering and aging experiments as well as the abbreviations used for the samples throughout the paper are summarized in Table I. Confocal Raman Microscopy. Raman linescans and Raman images of the cross-sections of the original and aged backsheets were recorded using a confocal Raman system Horiba Jobin Yvon LabRam 800 HR with visible laser excitation (473 and 514 nm) and motorized XYZ stage. The detection system used was a charge coupled
device camera (1 inch chip with 1024 3 256 pixel) with Peltier cooling. The spatial resolution obtainable in xand y-direction with such a setup is below 1 lm.10–12 For the imaging experiments described in this paper, an Olympus MPlan N objective with magnification of 1003 and a numerical aperture = 0.9 was used, the pinhole was closed to 400 lm, the spectral slit set to 100 lm, the integration time per pixel was 100 ms. The spectra were recorded in the spectral region 100–4000 cm1 with approximate spectral resolution of 5 cm1 (300 lines/mm grating). For the linescans, the spectra were recorded in the spectral range 100–3200 cm1 with approximate spectral resolution of 2 cm1 (600 lines/mm grating). The pinhole was closed to 300 lm, the spectral slit set to 100 lm, the integration time per pixel was 2 s. Averaged spectra from the various layers of the backsheet crosssections could be extracted from the linescans and images. In order to determine the thickness of the adhesive layer within the backsheet laminate, first linescans with a length between 15 and 25 lm and an increment of 1 lm were recorded. The sampling line was chosen in a way that the adhesive layer was placed in the middle of the line. To get additional information on the uniformity and compactness of the adhesive layers, Raman images of the laminated regions were measured. The image size was 30 lm 3 50 lm, the step size for the x and y increment was 1 lm. The linescans and images were evaluated in two ways: (i) either by integrating the area of a characteristic band for each component or (ii) by manually extracting the full spectrum for each component from the image and modeling the image with this spectral dataset, using a classical least squares (CLS) fitting. The resulting CLS model reveals the component distributions in the Raman images. The thickness of the adhesive layer was then determined by fitting the intensity profile of the adhesive band/spectrum using a Gaussian curve and determining the full width at half maximum (FWHM). For the Raman images, the average thickness value of at least five measuring points was calculated. Infrared Microscopy. Attenuated total reflection FT-IR images of the cross-sections of the backsheets were performed using a Perkin Elmer Spotlight 400 equipped with an ATR FT-IR imaging device and liquid nitrogen cooled mercury–cadmium–telluride array detector (linear array detector with 16 individual, gold-wired infrared detector elements). The ATR FT-IR imaging system is comprised of a germanium (Ge) crystal with a flattened tip (d 600 lm). For the measurements presented in this paper an image size of 500 lm 3 100 lm was chosen, which allowed for an image over the whole profile of the backsheet. According to the specifications of the manufacturer, the spatial resolution achievable with this equipment is below 5 lm (3.1 lm at 1700 cm1).13,14 The experimentally determined spatial resolution with this type of system (as measured by van Dalen et al.), using polymer multilayers for wavenumbers ranging from 1000 to 1730 cm1, was about 4 lm.15 The spatial resolution in that paper15 was calculated as the FWHM of the fitted first derivative of the absorbance profile. However, we applied a more stringent approach, as described by
APPLIED SPECTROSCOPY
585
TABLE I. Weathering and aging conditions.
Sample
Treatment
Temperature (8C)
Relative humidity (%)
Irradiation or NH3 concentration
Time (h)
S S_DH S_D S_H S_UVa S_Xea S_A S_DA
Original Damp heat Damp Heat UV weathering Xenon weathering Ammonia storage dry Ammonia storage wet
— 85 — 85 — — 80 80
— 85 85 — — — — 95
— — — — 340 nm 300–2500 nm 0.5% NH3 in N2 0.5% NH3 in N2
0 1000, 2000 1000, 2000 1000, 2000 1000 1000 1000 1000
a
For the spectroscopic imaging measurements the irradiated part of the backsheet was chosen.
Kazarian and Chan16,17 and Sommer18 where the distance between points corresponding to 5% and 95% of the maximum absorbance profile was taken as the actual spatial resolution. Best values for the actual spatial resolution of micro-ATR FT-IR imaging systems with Ge crystals could be as high as 4 lm16,17 and 8 lm18 at 1700 cm1. With this test we obtained with the equipment described above a practical resolution of about 9 lm at 1510 cm1, which is in good agreement with the results reported for the same system by Patterson and Havrilla,19 who found a minimum spatial resolution at the center of the hemisphere of the ATR crystal to be 10 lm. Averaged spectra of the different layers of the crosssection were extracted from the images. For the evaluation, the ATR FT-IR images were on the one hand treated with principle component analyses (PCA) resulting in a false color representation, where each spectroscopically distinguishable component is displayed in a different color. On the other hand, the images were analyzed using a spectra comparison correlation, where each spectrum of the image is set in comparison to a given reference spectrum, in our case, the adhesive spectrum in the original state. However, as the adhesive spectrum could not be extracted from the images in its pure form due to its limited thickness, the contributions of the neighboring layers had to be spectrally subtracted. The averaged thickness of the adhesive layer was determined from the grayscale representation of the images (after spectral comparison correlation) with the DigitalMicrographTM (Gatan, Inc.) software package. A mean value for the thickness was determined over the whole height of the image (100 lm).
in the profile of the samples. As an example, Figs. 2a and 2b show the profiles obtained for the PET, the adhesive, and the PVF layer obtained for sample S1_DH1000 by applying two different types of evaluation. First, one characteristic band in the spectra of each of the three components/layers was chosen, and the integrated area of these bands was plotted as a function of the scanned distance (Fig. 2a): the aromatic vibration at 1610 cm1 for the PET layer, the CH-stretching vibration at 2925 cm1 for the adhesive, and the TiO2 mode at 610 cm1 for the PVF layer filled with titania (TiO2) (see rectangles in Fig. 2c). However, as in the spectral region of the aliphatic CH vibrations (2800–3000 cm1), all three components showed absorptions, and no clear separation of the layers could be obtained (Fig. 2a). Thus, another type of evaluation, making use of a model based on the extraction of full spectra for the three independent compounds, was applied. The profiles calculated on the basis of these extracted spectra for
RESULTS AND DISCUSSION The thin adhesive layers interconnecting the PET core layer with the weather-resistant fluoropolymer outer layers of the PV backsheet were investigated with different spectroscopic techniques. After preparing cross-sections of the samples, Raman linescans, Raman images, and ATR FT-IR images of the profile of the laminated layers were recorded. The changes in the chemical structure, thickness, and compactness of the adhesive layer induced by the impact of environmental factors such as radiation, temperature, humidity, and corrosive gases were then assessed using various evaluation processes. Confocal Raman Microscopy. Raman Linescans. Raman linescans were recorded over the adhesive layer
586
Volume 68, Number 5, 2014
FIG. 2. Linescan profiles obtained (a) upon integration of one characteristic band (1610 cm1/PET layer, 2925 cm1/adhesive, and 610 cm1 for the PVF layer filled with titania) and (b) by applying a model based on extracted spectra; (c) Raman spectra of the individual layers, S1_DH1000h.
FIG. 4. Linescan profiles obtained by applying a model based on extracted spectra of samples (a) and (c) S1_UV and (b) S1_Xe. FIG. 3. (a) Linescan profiles calculated using a four-component model and (b) extracted Raman spectra; S2.
the PET, adhesive, and PVF/titania layer are given in Fig. 2b, the extracted spectra in Fig. 2c. With S2, however, we had to face a more complicated situation, as it became obvious from the first results that the PVDF outer film of this backsheet consists of three individual layers: PVDF/PVDF filled with titania/PVDF. Thus, four independent components (PVDF, PVDF plus titania, adhesive, and PET) had to be taken into account for the evaluation. As the CH-stretching bands used as characteristic bands for the adhesive are also the main bands in the pure PVDF layer and, additionally, the two other components show absorptions in this spectral region (see Fig. 3b), a separation of the individual layers just by choosing specific bands was not possible. Consequently, all evaluations were performed using the model calculating the linescan profiles from extracted spectra. These profiles obtained from the modeled linescan as well as the extracted spectra for S2 are given in Figs. 3a and 3b, respectively. By comparing the linescan profiles obtained for the adhesive layers after the various weathering and accelerated aging procedures and determining the thickness of the layer at half height of maximum, clear changes in shape and width of the adhesive bands could be observed upon aging (three typical examples are shown in the Fig. 4). Furthermore, the adhesive layer was in some cases partially split in two or sometimes more maxima upon accelerated aging. The partial overlap of the profiles obtained for the PET (green line in Fig. 4) and the adhesive (red line in Fig. 4) layers suggests that in the artificially weathered samples, S1, a partial transport of the adhesive into the PET layer took place. In these cases, it is very difficult to estimate the adhesive thickness correctly. Furthermore it became obvious that the thickness of the adhesive layer was not uniform over the sample length but rather heterogeneous with deviations of up to
625%. By measuring, e.g., ten linescans of the sample S_A on different sample positions, we obtained values for the thickness of the adhesive layer of 4.8 6 1.2 lm. As this deviation is high in comparison with the effects we expect to see upon weathering and accelerated aging, we decided to record images to get a better overview of the inhomogeneity of the adhesive and to be able to derive more reliable results on the adhesive thickness. Raman images would allow for the at-once observation of a larger sampling area and, thus, permit the direct inspection of the thickness deviations of the adhesive layers. Raman Images. Raman images with dimensions of 30 lm 3 50 lm were recorded in the laminated areas of the backsheets and evaluated in a first step in a comparable manner to the Raman linescans by choosing one characteristic band from the spectrum of each component/layer: the aromatic vibration at 1610 cm1 for the PET, the CH-stretching vibration at 2925 cm1 for the adhesive, and the band at 610 cm1 for the PVF layer filled with titania (TiO2) of sample 1. In a second step, the evaluation was made by means of a model using extracted full spectra for each component. The comparison of the images obtained with the two evaluation approaches is shown exemplarily for samples S1 and S1_DH1000 in Fig. 5. It was observed that the adhesive within the S1 samples stayed intact as a uniform layer within the selected sample area of about 30 lm, and the averaged thickness is only slightly decreased upon prolonged damp heat storage (4.5 lm ! 4.3 lm for 1000 h ! 3.8 lm for 2000 h; model evaluation). The Raman images of the samples S2, S2_A, and S2_DA obtained after evaluation with the model (extracted Raman spectra) are given in Fig. 6. Upon treatment with dry ammonia (S2_A) the adhesive layer was found to stay intact, while the storage of the sample in humid ammonia atmosphere (S2_DA) led to a strong decline in the layer thickness, from 4.3 lm to 2.6 lm, as well as a partial overlap with the adjacent PVDF material.
APPLIED SPECTROSCOPY
587
FIG. 5. Comparison of the Raman images obtained (a) and (c) using evaluation of characteristic bands and (b) and (d) by applying a model based on extracted spectra (a) and (b) for S1, and (c) and (d) S1_DH1000.
The values derived for the averaged thicknesses of the adhesive layers in the original and aged backsheets S1 with the two evaluation approaches are summarized in Fig. 7. It can be shown that for this backsheet (uniform
FIG. 6. Raman images of samples (a) S2, (b) S2_A, and (c) S2_DA evaluated via the model.
588
Volume 68, Number 5, 2014
outer layer, three independent components) both evaluation methods yield satisfactory results. Especially with the aged samples, the images obtained upon model evaluation gave more homogeneous layers. This in turn allows for the more reliable determination of the averaged layer thicknesses. Weathering with UV-light (S1_UV) and storage in a humid ammonia atmosphere (S1_DA) seem to have the strongest impact on the adhesive layer. The results obtained for backsheet S2 after evaluation with the model based on extracted spectra are presented in Fig. 8. With these samples (with four components) the evaluation through band integration was not possible (see above). While some treatments, such as xenon weathering (S2_Xe) or storage in ammonia containing
FIG. 7. Results of the thickness of the adhesive layer of backsheet S1 in the original and aged state as derived from the Raman images (blue = evaluating using characteristic bands, red = model with extracted spectra).
FIG. 8. Results of the thickness of the adhesive layer of backsheet 2 in the original and aged state as derived from the Raman images.
atmosphere (S2_A), seem to have no effect on the thickness of the adhesive layer, UV exposure (S2_UV), storage under damp heat (S2_DH), or humid ammonia atmosphere (S2_DA) cause a reduction of the adhesive layer thickness by approximately one-third. The Raman spectra of the aged adhesive layer show significant fluorescence, especially with the samples after damp heat treatment. Therefore and in order to obtain more detailed information on the chemical changes occurring within these layers upon aging, we decided to additionally record ATR FT-IR images of these samples. Infrared Microscopy. Attenuated Total Reflection Fourier Transform Infrared Imaging. In addition to the Raman images, ATR FT-IR images of the cross-sections of the backsheets were collected. The sample area was chosen in a way that the whole profile of the sample could be visualized within one image (500 lm 3 100 lm).
FIG. 9. ATR FT-IR image (a) after PCA and (b)–(d) single contributions of three components (red = PET layer; blue = PVF layer; green = adhesive).
FIG. 10. ATR FT-IR images of samples S1 after performing spectra comparison correlation.
A typical example (S1) for the obtained images after PCA as well as the images of the principal components contributions is given in Fig. 9. Although the thin adhesive layer (thickness 5 lm) could be detected in the original sample, the adhesive could not be visualized satisfactorily after PCA or band integration after some accelerated aging tests. Thus, for the evaluation of the thickness and compactness of the adhesive layers after weathering and/or artificial aging, a different kind of data treatment had to be chosen: we evaluated the ATR FT-IR images via a compare correlation with a reference adhesive spectrum. However, as the thickness of the adhesive layers has comparable dimensions with the diffraction limited spatial resolution of ATR FT-IR microscopes13–19 (see above), the isolated spectrum of the adhesive cannot be directly extracted from the ATR FT-IR images of the cross-sections of these backsheet samples. The spectrum of the unaltered adhesive was, thus, extracted from the ATR FT-IR image after PCA analysis of the sample in the original state (S1 or S2), and the contributions of the neighboring layers (PET and PVF or PVDF, respectively) were spectrally subtracted. The spectrum of the pure adhesive in the original state then was taken as reference spectrum for the spectra comparison correlation analysis of the ATR FT-IR images of the various aged samples. This evaluation procedure was performed for all samples; the resulting images are
APPLIED SPECTROSCOPY
589
FIG. 11. Extracted and corrected spectra of the adhesive for sample S1 after different aging conditions.
presented in part in Fig. 10. Treatment with heat (S1_H), humidity (S1_D), ammonia gas storage (S1_A), and Xenon exposure (S1_Xe) had no strong effect on the appearance of the adhesive layer in the image, while storage under damp heat conditions (S1_DH2000), humid ammonia atmosphere (S1_DA), or UV exposure (S1_UV) caused a drastic change in the degree of conformance of the adhesive with the reference adhesive spectrum (Fig. 10). Thus, the ATR FT-IR spectra of the region of the adhesive between the PET and fluoropolymer layers were extracted from the ATR FT-IR images of the aged samples, and the contributions of the adjacent layers were subtracted. The resulting spectra of the aged/ weathered adhesive were then compared with the reference spectrum (for sample S1, see Fig. 11) in order to get information on the aging-induced changes in the chemical structure of the adhesive. It becomes visible that upon aging spectral changes have occurred in the spectra of the adhesive (especially in the spectral region between 1150 cm1 and 1750 cm1) upon storage under combined damp/heat and damp/ammonia conditions as well as after UV irradiation (Fig. 11). From the FT-IR spectra we suggest that the original adhesive has to contain an ester–urethane compound (e.g., at 1727, 1707, 1530, and 1226 cm1, Fig. 11) but also shows additional absorptions typical for epoxide-based adhesives (1610, 1510, and 830 cm1). Therefore we consider that the adhesive is based on a polyester–urethane polymer modified with an epoxide (on the basis of bisphenol A). While the IR spectrum of the adhesive stays nearly unchanged for the samples S1_A, S1_D, and S1_H, few changes were observed in the ester bands (1727 and 1226 cm1) for samples S1_DH1000 and S1_Xe. Strong degradation of the carbonyl group at 1727 cm1 is observed upon humid ammonia storage (S1_DA), prolonged damp heat treatment (S1_DH2000), and UV weathering (S1_UV). The urethane absorption at 1707 cm1 stays clearly visible after all artificial aging procedures, but decreases in intensity upon UV weath-
590
Volume 68, Number 5, 2014
FIG. 12. ATR FT-IR images of samples S2 after performing spectra compare correlation.
ering. The bands at (1610) 1510, 1464 1240, 1180 and 1118 cm1, to which we attribute an epoxide modification (particularly bisphenol A), dominate the degraded adhesive spectra. For sample S2, the resulting ATR FT-IR images of the adhesive layer (after performing spectral compare correlation) are shown in Fig. 12. The application of individual environmental factors such as heat (S2_H), humidity (S2_D), or corrosive gas (S2_A) had no strong effect on the appearance of the adhesive layer. For the weathered samples S2_Xe and S2_UV and the artificially aged samples S2_DH and S2_DA, however, a change in the evenness and thickness of the adhesive was observed, partially strongly different for the two adhesive layers within the backsheet. Especially with the samples S2_UV and S2_DA, a partial disintegration of the adhesive layer was observed with strongly varying layer thicknesses. The average thickness of the adhesive layers in the ATR FT-IR images, obtained after comparison correlation with the original reference spectrum, was evaluated using an image analysis procedure. The mean values of the layer thickness (averaged over the image height to equal 100 lm) obtained from the grayscale images of the various samples, are presented in Figs. 13 (S1) and 14 (S2). With this type of evaluation, where the spectral conformity with the original adhesive is analyzed, a spreading of the adhesive is observed for part of the
FIG. 13. Adhesive thickness estimated from the ATR FT-IR images for samples S1.
FIG. 14. Adhesive thickness estimated from the ATR FT-IR images for samples S2.
samples, e.g., S_DH, S_UV, and S_DA. These findings together with the results obtained for the chemical changes in the adhesive structure upon aging suggest that the unchanged polyester–urethane basis of the adhesive is partially transported into the neighboring layers while the epoxide component is enriched in the interlayer region. It seems that especially this epoxide layer is detected in the Raman images as the individual adhesive layer. In addition, especially with backsheet S2, the two adhesive layers showed nonuniform aging effects within the sample, making the results strongly dependent on the measuring field chosen for the imaging measurement. The smaller the chosen measurement field is, the stronger this effect gets: Raman linescan . Raman image . ATR FT-IR image.
technique is much higher. For the analysis of the Raman linescans and images, two different approaches were evaluated. In principle both methods, analysis of the Raman spectra using band integration of individual bands and using a model making use of extracted spectra, yield suitable Raman images. For the linescans, however, the analysis using the model was clearly favorable. Especially in the case when each of the independent components of the sample does not show a characteristic absorption band usable for the evaluation using band integration, the analysis using a model based on extracted spectra instead of individual bands proved to be the more constructive and reliable approach. Although Raman spectroscopy is known to have the superior lateral resolution as compared with ATR FT-IR spectroscopy,10–25 the adhesive layers of the samples used within this study (layer thickness 3–6 lm) could also be detected and visualized by applying the ATR FTIR imaging method. However, the analysis of the images was quite a demanding task, as the thickness of the adhesive layer was in the magnitude of the resolution limit of this method, and the FT-IR spectra of the pure adhesives could not be directly extracted from the images. In addition, to make things worse, the spectrum of the adhesive showed strong similarities to the spectrum of the next neighboring layer, PET. By applying alternative methods for the data evaluation, such as principal component analysis, spectral subtraction of the disturbing contributions of the neighboring layers, and using spectra comparison correlation with reference spectra, we also obtained reliable images for the adhesive layers within the various samples with this method. The information obtained for the impact of artificial aging and weathering on the adhesive layer obtained using Raman imaging and ATR FT-IR imaging was, however, complementary.20,21,26,27 As two completely different approaches were used for the evaluation of the images, namely, (i) analysis with a model using extracted spectra from the image for the Raman images and (ii) spectra comparison correlation with a reference spectrum of the adhesive in the original state for the ATR FT-IR images, different information for the aging-induced
CONCLUSION Three different spectroscopic approaches, Raman linescans, Raman imaging, and ATR FT-IR imaging, were evaluated for the visualization of the thin adhesive layers (3–6 lm) present in polymeric PV backsheets. The cross-sections of the multilayer samples in the original, weathered, und artificially aged samples were investigated spectroscopically in order to describe the impact of the environmental factors on the evenness and thickness of the adhesive layers. All three methods were found to be suitable tools to identify and detect the thin adhesive layer within the original and aged polymeric laminates. However, as the adhesive layer is not very uniform in thickness and partly disintegrates upon weathering and/or artificial aging, Raman linescans yield only qualitative information and do not allow for an estimation of the layer thickness, as deviations of up to 6 25% in the adhesive thickness were determined when measuring ten independent positions within one sample. Upon increasing the measuring area by moving from one-dimensional linescans to two-dimensional Raman images, a much better result could be achieved. Although a higher measuring time has to be taken into account, the information on the uniformity and evenness of the adhesive layer obtainable using the imaging
APPLIED SPECTROSCOPY
591
changes could be derived from the spectroscopic images. A broadening/spreading of the adhesive (polyester–urethane compound) becomes evident upon evaluation of the ATR FT-IR images after spectral comparison correlation, especially with those samples (DH and UV) where the adhesive is severely degraded (hydrolyzed) during aging. The epoxide part of the adhesive, however, seems to be enriched in the interlayer region. For the evaluation of the Raman images, the signal of this epoxide fraction seems to be the main spectral feature. Thus, a decrease in the adhesive layer thickness upon aging is detected with this method. In the case of chemical alteration, the Raman spectra of the adhesive only show increasing fluorescence upon accelerated aging (especially after damp heat treatment). The IR spectra thereof, however, were a valuable source for information regarding the aging induced chemical changes (e.g., degradation of the ester group) occurring within these layers. The fact that the samples showing strong changes in the thickness and chemistry of the adhesive layer (after damp heat test, ammonia/water storage, and after UV irradiation), as derived from molecular spectroscopy, are the same samples showing strong changes in macroscopic properties such as ply adhesion and/or mechanical characteristics, such as, e.g., the elongation at break, seems to be quite promising. The sample showing the strongest delamination behavior within the backsheet is S2_DH2000, also S2_DA shows a light tendency to delamination. ACKNOWLEDGMENT The present work was funded by the project ‘‘Analysis of PV Ageing’’ of the Austrian Klima- und Energiefonds, ‘‘Neue Energien 2020’’. The Raman microscope was financially supported by the European Regional Development Fund (EFRE), the Graz University of Technology, and the Government of Styria. 1. S. Kurtz, J. Wohlgemuth, T. Sample, M. Yamamichi, J. Amano, P. Hacke, M. Kempe, M. Kondo, T. Doi, K. Otani. ‘‘Ensuring Quality of PV Modules’’. Paper presented at: 37th IEEE Photovoltaic Specialists Conference (PVSC). Seattle, WA; June 19-24, 2011. Print ISBN 978-1-4244-9966-3: 842-847. 2. W. Gambogi, S. Kurian, B. Hamzavytehrany, J. Trout, O. Fu, Y. Chao. ‘‘The Role of Backsheet in Photovoltaic Module Performance and Durability’’. Paper presented at: 26th EU PVSEC. Hamburg, Germany; September 5-9, 2011. 3. C.E. Packard, J.H. Wohlgemuth, S.R. Kurtz, U. Jahn, K.A. Berger, T. Friesen, M. Koentges. ‘‘Fielded PV Module Condition’’. Workshop presented at: 27th EU PVSEC Parallel Event Workshop from IEA PVPS Task 13: ‘‘Characterising and Classifying Failures of PV Modules’’. Frankfurt, Germany; September 24-28, 2012. 4. A.J. Steiner, W. Krumlacher, H. Muckenhuber, M. Kraxner. ‘‘PV Backsheet Materials under Accelerated Aging Conditions: A Performance Study’’. Paper presented at: 26th EU PVSEC. Frankfurt, Germany; September 5-9, 2011. 5. G. Oreski, G.M. Wallner. ‘‘Delamination Behavior of Multi-Layer Films for PV Encapsulation’’. Sol. Energy Mater. Sol. Cells. 2005. 89: 139-151. 6. E.J. Lightfoot, M. Debergalis, W.J.M. Gambogi, B. Fu, A.Z. Bradley, T.J. Trout. ‘‘Adhesion in Photovoltaic Modules’’. Paper presented at: 34th Annual Meeting of the Adhesion Society USA. Savannah, GA; February 13-16, 2011. 7. K.J. Geretschla¨ger, G.M. Wallner, J. Fischer. ‘‘Raman and Infrared Microscopical Analysis of Multilayer Backsheets’’. Paper presented at: 27th EU PVSEC. Frankfurt, Germany; September 24-28, 2012.
592
Volume 68, Number 5, 2014
8. International Organization for Standardization. ‘‘Plastics: Methods of Exposure to Laboratory Light Sources. Part 2: ‘‘Xenon-Arc Lamps’’ O¨NORM EN ISO 4892-2; Part 3: ‘‘Fluorescent UV Lamps’’. O¨NORM EN ISO 4892-3. 2006. 9. International Organization for Standardization. ‘‘Crystalline Silicon Terrestrial Photovoltaic (PV) Modules–Design Qualification and Type Approval: 10 Test Procedures, 10.13 Damp-Heat Test’’. ISO 61215. 2005. 10. N.J. Everall. ‘‘Confocal Raman Microscopy: Performance, Pitfalls, and Best Practice’’. Appl. Spectrosc. 2009. 63(9): 245A-262A. 11. N.J. Everall. ‘‘Confocal Raman Microscopy: Why the Depth Resolution and Spatial Accuracy Can Be Much Worse Than You Think’’. Appl. Spectrosc. 2000. 54(10): 1515-1520. 12. K.R. Allakhverdiev, D. Lovera, V. Altstadt, P. Schreier, L. Kador. ‘‘Confocal Raman Microscopy: Non-Destructive Materials Analysis with Micrometer Resolution’’. Rev. Adv. Mater. Sci. 2009. 20: 77-84. 13. Perkin Elmer. Application note: ‘‘Large Area ATR-FTIR Imaging Using the Spotlight FT-IR Imaging System’’. 2006. http://www. perkinelmer.com/ CMSResources/Images/44- 74854TCH_ LargeareaATRFT-IRImaging.pdf [accessed Sept 16 2013]. 14. Perkin Elmer. Technical note: ‘‘Spatial Resolution in ATR-FTIR Imaging: Measurement and Interpretation’’. 2006. http://www. perkinelmer.ca/CMSResources/Images/44-74872TCH_ SpatialResolutioninATRFT-IRImaging.pdf [accessed Sept 16 2013]. 15. G. van Dalen, P.C.M. Heussen, R. den Adel, R.B.J. Hoeve. ‘‘Attenuated Total Internal Reflection Infrared Microscopy of Multilayer Plastic Packaging Foils’’. Appl. Spectrosc. 2007. 61(6): 593-602. 16. K.L.A. Chan, S.G. Kazarian. ‘‘New Opportunities in Micro- and Macro-Attenuated Total Reflection Infrared Spectroscopic Imaging: Spatial Resolution and Sampling Versatility’’. Appl. Spectrosc. 2003. 57(4): 381-389. 17. S.G. Kazarian, K.L.A. Chan. ‘‘Micro- and Macro-Attenuated Total Reflection Fourier Transform Infrared Spectroscopic Imaging’’. Appl. Spectrosc. 2010. 64(5): 135A-152A. 18. A.J. Sommer, L.G. Tisinger, C. Marcott, G.M. Story. ‘‘Attenuated Total Internal Reflection Infrared Mapping Microspectroscopy Using an Imaging Microscope’’. Appl. Spectrosc. 2001. 55(3): 252256. 19. B.M. Patterson, G.J. Havrilla. ‘‘Attenuated Total Internal Reflection Infrared Microspectroscopic Imaging Using a Large-Radius Germanium Internal Reflection Element and a Linear Array Detector’’. Appl. Spectrosc. 2006. 60(11): 1256-1266. 20. B. Chernev, P. Wilhelm. ‘‘Vibrational Spectroscopy and Spectral Imaging Techniques—Special Applications in the Polymer Science’’. Monatsh. Chem. 2006. 137(7): 963-967. 21. P. Wilhelm, B.S. Chernev. ‘‘Microscale Chemical Imaging Using Vibrational Spectroscopy Methods’’. In: A. Mendez-Vilas, J. Diaz, editors. Microscopy: Science, Technology, Applications and Education. Badajoz, Spain: Formatex Research Center, 2010. Vol. 3. Pp. 2062-2071. 22. P. Wilhelm, B. Chernev. ‘‘Spatial Imaging/Heterogeneity’’. In: J.M. Chalmers, R.J. Meier, editors. Compr. Anal. Chem. 2008. 53: 527560. 23. A. Gupper, P. Wilhelm, M. Schmied, S.G. Kazarian, K.L.A. Chan, J. Reußner. ‘‘Combined Application of Imaging Methods for the Characterization of a Polymer Blend’’. Appl. Spectrosc. 2002. 56(12): 1515-1523. 24. K.L.A. Chan, S.G. Kazarian. ‘‘Detection of Trace Materials with Fourier Transform Infrared Spectroscopy Using a Multi-Channel Detector’’. Analyst. 2006. 131(1): 126-131. 25. S.G. Kazarian, K.L.A. Chan. ‘‘ATR-FTIR Spectroscopic Imaging: Recent Advances and Applications to Biological Systems’’. Analyst. 2013. 138(7): 1940-1951. 26. G.C. Eder, L.S. Lukacic, B.S. Chernev. ‘‘Visualisation and Characterisation of Ageing Induced Changes of Polymeric Surfaces by Spectroscopic Imaging Methods’’. Anal. Bioanal. Chem. 2012. 403(3): 683-695. 27. B.S. Chernev, G.C. Eder. ‘‘Spectroscopic Characterisation of the Oligomeric Surface Structures on Polyamide Materials Formed During Accelerated Ageing’’. Appl. Spectrosc. 2011. 65(10): 11331144.
