Round‐robin weathering test of various polymeric back‐sheets for PV modules with different ultraviolet irradiation and sample temperatures

Durability assessment of materials exposed to natural weathering with expected service lifetimes of more than 20 years requires accelerated testing of the UV stability, especially for polymeric materials. Fraunhofer ISE organized an interlaboratory comparison for testing different back‐sheets in various test laboratories using different UV sources. The five participating test laboratories used three different types of UV sources. The interaction of the UV radiation with the polymers used in PV modules is main subject of this round‐robin. Laminates were produced by using solar glass and a suitable EVA encapsulant combined with 10 different back‐sheets. The simultaneous usage of different optical filters allowed to investigate roughly the spectral sensitivity and the intensity impact.


| INTRODUCTION
Photovoltaic modules are used for the conversion of solar energy to electricity. Durability testing of materials exposed to natural weathering requires testing of the UV stability, especially for polymeric materials. Numerous papers dealing with the two most prominent polymeric components, namely, the encapsulant of the cells and the weather-protecting back-sheet, can be found in the literature. The type approval testing of PV modules according to the standard IEC 61215 ed. 3 and IEC 61730 includes a so-called UV-preconditioning test with a total UV dose of 60 kWh/m 2 . About 3%-10% should be in the UV-B wavelength range (280-320 nm). However, the spectral distribution of the radiation source is very important to the degradation because the samples show unique spectral sensitivities to the radiation applied. A filtered xenon light is the ideal artificial light source because of the sun-like spectrum. But, in addition to the high costs and relatively low efficiency, the use of a xenon light bears another disadvantage, namely, the fact that the samples are strongly heated up by the broadband irradiation from the UV to the infrared. Less than 6% of the intensity of solar radiation is in the UV range. In case of increasing the intensity of the light source for accelerating the UV test, the overheating of the samples would have to be prevented more strictly, and the temperature of the samples have to be measured in order to avoid misinterpretation of the test results.
Less temperature impact can be expected from UV-A fluorescent tubes, because they have an intensity maximum around 340 nm and decreasing with the longer wavelengths. However, they are limited in intensity and not an ideal solar simulator. Therefore, there remains the question whether the longer wavelength radiation (VIS-NIR) might cause degradation phenomena or contribute to degradation that could not be revealed by those UV tests.  The emission spectra allow the integration of the UV range (280-400 nm) and the photodegrading short wavelength range below 340 nm (see Figure 2) that might be more accurate than integral UV measurements. 4 One big problem is the application of integral UV sensors for controlling the UV dose because they are mostly calibrated with other types of light sources like sun or tungsten-halogen lamps and have different spectral sensitivities at the long wavelength edge (400 nm) as was found in a previous comparison of various UV sources for material testing. 4

| Sample design
The focus of the round-robin testing was put on polymeric backsheets of photovoltaic modules. They are hit by direct UV radiation from the normally light-facing glass side, which is filtered by the frontglazing and the polymeric encapsulant as in the gap between the solar cells (in case of modules with silicon solar cells) or unfiltered by diffuse UV radiation from the back side (see Figure 3).
Measurements of the spectral UV radiation on Canary Island resulted in a ratio of 5% to 10% of radiation on tilted or vertical backside compared with the front in plane of array (tilt angle of 22.5 to the south corresponding to the latitude and roughly 1 m above the F I G U R E 1 Spectral irradiation of a selection of different UV light sources in the participating labs and natural UV irradiation for AM1.5 F I G U R E 2 Integrated spectral irradiation of a selection of different UV light sources in the participating labs and natural UV irradiation for AM1.5 F I G U R E 3 Sketch of the design of a photovoltaic module with a polymer back-sheet natural reddish ground). The yearly loads of UV radiation at the Canary Island test site are described elsewhwere. 5 Sample laminates were produced with the same solar glass from Interfloat GmbH and the same encapsulant (Bridgestone S18) but without solar cells and 10 different back-sheet materials. Some of the back-sheets had a weak UV stability in order to facilitate degradation investigation by the accelerated life tests. Therefore, the industry partners did not want to disclose the composition of the test samples. They were exposed with the front side and with the back side towards the UV source, separately. The size of the samples (200 mm × 130 mm) was selected in order to allow the partial cover with spectral long-pass edge filters and neutral density filters for simultaneous variation of the UV irradiation spectrally and with intensity (see Figure 4).
The spectral transmittance of the edge filters ( Figure 5) is to be multiplied with the transmittance of the glass cover and the encapsulant allowing the calculation of the irradiation on the normally light-facing (front) side of the back-sheet in the laminate. Gratings that were used are a less expensive alternative for neutraldensity filters. The effect of a single grating and two stacked gratings can be seen in Figure 6. The spectral transmittance is sufficiently wavelength independent for the wavelength range of interest. The integral UV transmittance is 56% for the single grating and 32% for crossed gratings.
The effect of the filters could be seen visually as different yellowing or even blistering (see Figure 7). Therefore, the yellowness index (YI) seems to be an appropriate degradation indicator. It has been shown that YI can be correlated with the mechanical properties (strain to break) of polymeric foils. 6

| Degradation indicator
Most of the back-sheet samples were filled with diffusely reflecting inorganic compounds (usually TiOx). The spectral hemispherical reflectance was measured with an integrated sphere from the irradiated side (front or back). Extremely stable samples show hardly any changes of the spectral reflectance of sample spots behind the applied filters after 120 kW/m 2 UV irradiation (see Figure 8), proving that the used EVA and the glass were not degraded by the UV exposure.
Another example (shown in Figures 7 and 9) shows the yellowing of a sample depending on the applied filters. Color coordinates X, Y, and Z were calculated from the spectra and used for the computation of YI with the following formula according to ASTM E313 7 : with C x = 1,2871 and C z = 1,0781.

| RESULTS
The difference of the YI of the samples after 120 kWh/m 2 UV irradiation (corresponding to about 2 years outdoor exposure 3 ) is shown in Figure 11 for the EVA (front) side of the laminates and in Figure 12  are shown in Figures 12 and 14. Note the factor of 10 in the scaling of the y-axis taking into account the one magnitude less degradation, when filtering out the short wavelength range of the irradiation. This effect can be seen in all labs, whereas the cutoff at 320 nm has different effects in the different labs (Figures 15 and 16). The two samples 4 and 10 behave differently because of their different spectral sensitivity. Figure 17 shows the degradation as function of the dose behind the edge filters. Similar curves can be found for the degradation of the samples behind the grating filters ( Figure 18). The constant degradation ratio proves a linear relation between YI and the UV dose.
The temperature dependence can be seen clearly in Figure 19.  F I G U R E 2 1 Sample temperature of three differently colored back-sheets at 60 C cabinet temperature after 60-min irradiation with an HQI-UV source F I G U R E 2 3 Sample temperature of three differently colored back-sheets at 60 C cabinet temperature after 60-min irradiation with a fluorescent tube UV source F I G U R E 2 2 Sample temperature of three differently colored back-sheets at 60 C cabinet temperature after 60-min irradiation with an HMI solar simulator