field technique for photovoltaic module defect detection was ... appear or have been reached during module operation. ... The EVA of the second manufacturer.
Ultraviolet fluorescence of ethylene-vinyl acetate in photovoltaic modules as estimation tool for yellowing and power loss Arnaud Morlier, Michael Siebert, Iris Kunze, Susanne Blankemeyer and Marc Köntges Institute for Solar Energy Research Hamelin (ISFH), 31860 Emmerthal, Lower Saxony, Germany Abstract — The potential of ultraviolet (UV) fluorescence as a field technique for photovoltaic module defect detection was demonstrated recently. Here we study the formation rate of the fluorophores in module encapsulating material under UV illumination. We observe a correlation between the decrease in visible light transmission of the encapsulant and UV fluorescence intensity. This correlation allows estimating the yellowing-induced power loss via fluorescence measurements. Index Terms — Fluorescence, nondestructive testing, solarmodule reliability.
colored products formed in the material during ageing are also suspected to be responsible for fluorescence emission. In this paper, we study the formation of fluorophores under various conditions in three EVA materials. We investigate the kinetics of fluorophore formation over time and demonstrate the correlation between the fluorescence emission intensity and the yellowing of embedding material in the PV modules. To conclude, we simulate in the extended paper the impact of the observed yellowing on the efficiency of a PV module and show a forecast of this degradation as a function of the UV dose.
I. INTRODUCTION UV fluorescence inspection has revealed as a promising alternative for the rapid detection of photovoltaic (PV) module failures in the field recently. This method has proven useful for the detection of cell cracks or abnormal cell operating temperatures. It is applicable during daytime without requiring any electrification of PV modules [1,2] and can be an alternative or a complement to the usual electroluminescence [3] or thermography imaging [4] methods. The UV fluorescence method relies on the excitation of fluorophores present in a PV module encapsulating material with UV light, and the observation of the fluorescence emission pattern in the visible spectral range. The emission can be detected with the human eye or a camera equipped with filters eliminating the reflected excitation UV light [5]. These fluorophores are formed in the material over time when the module is exposed to sunlight. The formation rate of the fluorophores is increased by higher temperatures, resulting in more intensively fluorescing spots where higher temperatures appear or have been reached during module operation. The fluorophores are extinguished in the presence of oxygen. Therefore, oxygen diffusing through backsheet in the module will extinct locally the fluorescence of the material between the cell edges. The encapsulating material laminated between the cells and the front glass panel is typically not reached by the oxygen diffusing through the backsheet excepted in case the cell present some cracks through which gases can diffuse. Thus the cracks are revealed indirectly by fluorescence extinction on the fluorescence emission pattern [6]. Cell damages are not the only cause of power loss in photovoltaic systems. The weathering of polymeric encapsulant materials such as ethylene-vinyl acetate (EVA) may lead to a discoloration and a decrease of its optical transmission. The
II. METHOD We capture fluorescence images in the laboratory with a device consisting of a digital camera and two UV light emitting diodes arrays as described in [2]. Capturing with the camera integrates the whole fluorescence signal between 450 nm and 750 nm. The camera and the light sources are mounted on a structure ensuring the reproducibility of the measurements by maintaining the same relative position of the samples to the camera and the LEDs from measurement to measurement. Fluorescence measurements are carried on at room temperature. The three EVAs investigated in this paper are from two different manufacturers. From one manufacturer we employ one material with a UV cut-off wavelength at 377 nm (EVA A) and one EVA with a UV cut-off at 322 nm (EVA B). This latter EVA is designed for having a higher transmission at short wavelengths, allowing the most recent PERC cells to collect more photons, thus increasing the potential efficiency of the cells in a module [7]. The EVA of the second manufacturer (EVA C) has a cut-off wavelength of 377 nm. EVA samples are prepared by laminating two polymer layers between two 120 mm x 120 mm glass windows. We choose this sample configuration to reproduce the conditions of the oxygen free EVA area over the middle of a solar cell between cell and glass front cover. This zone corresponds to the glass/glass sample middle. Moreover, we simulate the EVA at the cell edge area in the module where oxygen diffuse through the backsheet and the EVA along the cell edge. This zone corresponds to the edge of the glass/glass sample. Furthermore, laminating without backsheet allows for avoiding taking the backsheet yellowing into account by studying polymer yellowing. We irradiate the samples in a UV chamber equipped with low pressure Hg UV lamps delivering an average power of 290 W/m² in the range 300 nm to 400 nm at the sample surface. The
samples are held during the irradiation at temperatures of 27°C (EVA A) or 57°C (EVA A, B and C). To be able to discriminate between thermal effects and UV aging we hold samples of EVA A and C in the dark at a temperature of 57°C, which is approximately the normal operating cell temperature (NOCT) in a module. The UV-visible transmission spectra are collected in the range 280 nm to 700 nm with a Perkin-Elmer Cary 5000 spectrometer.
Fig.2 shows the fluorescence intensity of different EVA samples as function of the UV dose and EVA samples maintained in the dark at 57°C for the same duration as the UV treated samples are exposed.
III. RESULTS Fig.1 shows the fluorescence intensity IF as function of the position across the EVA C sample after various UV irradiation doses. On top of the graph is a grayscale image of the sample cross-section from which the fluorescence intensity is extracted. The sample edges are at the x=0 mm and x=120 mm marks. They appear brighter as they reflect the fluorescence of the center of the laminate. The sample shows an initial fluorescence effect after lamination. This initial fluorescence has vanished after an UV dose exposure of 6 kWh/m². With further UV exposure the fluorescence increases in the middle region of the sample. In the following we measure the fluorescence of the samples in the central 4 cm x 4 cm square.
Fig.2. Fluorescence intensity of the central area as a function of the dose for EVA A, B and C under various conditions.
Samples maintained at 57°C without UV exposure show no increased fluorescence over time. Samples of EVA A irradiated over the same duration with the same dose at temperatures of respectively 27 °C and 57°C show different fluorescence intensities, the development of fluorophores being faster at 57°C. The samples of EVA B do not develop fluorophores under any of the studied conditions. The increase of fluorescence intensity in EVA A and C as a function of the dose does not increase linearly with increasing dose. Fig.3 shows the derivative of the fluorescence intensity dIF/dD over the dose as a function of the dose D for the EVA A and C samples. The derivative of the fluorescence intensity dIF/dD over the dose as a function of the dose D for EVA A is increasing linearly with the dose D.
Fig.1. Fluorescence intensity across a glass/glass EVA A sample at various UV exposure doses. On top of the graph a middle section fluorescence image of sample EVA A after a UV dose of 265 kWh/m² is shown.
700
𝛥𝜏 =
∫380 (𝜏(𝜆)−𝜏0 (𝜆))d𝜆 700
∫380 d𝜆
(1)
Fig.5 shows the integrated loss of transmittance Δτ in the wavelength range 380 nm to 700 nm as a function of the fluorescence intensity for EVA A and C exposed to UV.
Fig.3. Fluorescence intensity derivative dIF/dD over the dose as a function of the dose D for the EVA A and C samples. The dashed lines are linear fits.
Transmission spectra in the UV and visible wavelength range after different UV exposures are shown in Fig.4. The transmission of EVA A and C decreases with increasing UV dose in the range 380 nm to 550 nm. This decrease of transmittance in the violet domain leads to the yellowing of the EVA. Furthermore, the transmission in the range 350 nm to 370 nm increases. For EVA B, the transmission value decreases slightly in the range 290 nm to 370 nm. This negligible variation is also not accompanied by any visible yellowing.
Fig.5. Integrated transmittance decrease of EVA A and C under UV exposure at 27°C and 57°C as a function of fluorescence intensity. The dashed line is a guide to the eye.
There is a strong correlation between the yellowing of the EVA and the fluorescence intensity. We observe this correlation for samples of two different materials and for samples irradiated at different temperatures, which suggests that this correlation is neither depending on the ageing conditions nor specific to a material. IV. DISCUSSION
Fig.4. Transmission spectra of EVA A, B and C after irradiation with different UV doses at 57°C. Note that the wavelength scale (abscissa) for EVA B is different than for EVA A and C.
We integrate the transmission loss for λ>380 nm over the spectra of EVA A and C according to (1).
The initial decrease of fluorescence intensity observed in Fig.1 is due to the depletion of the fluorophores formed uniformly during the lamination process [1,2,8]. The decrease of the fluorescence at the edges of the sample is mainly due to the oxygen diffusion and interaction with the fluorophores in presence of UV light [6]. As our aim is to investigate the material comprised in the oxygen free zone comprised between the cells and the front glass of a PV module we base our further investigations on this central domain. Fig. 2 shows that a higher temperature is a favoring condition but not a sufficient condition for the formation of fluorophores, as samples heated in the dark are not forming any fluorophores. Furthermore the EVA without UV absorber studied her is not developing any fluorophores under UV irradiation. This suggests that UV absorbers are potential precursors of the fluorophores.
As we have shown that the fluorescence intensity is correlated to the yellowing and the power loss, it is of interest to propose a model describing the yellowing of the encapsulation material over time. We observe a non linear increase of the fluorescence intensity with increasing dose over the first 800 kWh/m² of UV irradiation, suggesting an acceleration with increasing dose of the fluorophore formation and yellowing process. The formation of fluorophores is due to the degradation of chemical species present in the material. As these species are depleted, the probability for a photon to reach a fluorophore precursor and initiate a degradation reaction decreases with increasing fluorescence intensity, leading to a decrease of dIF/dD. Furthermore, after depletion of these precursors, the fluorescence would reach a saturation plateau. The total UV irradiation dose received by the samples equals 26 years of outdoor exposure in northern Germany [9]. As the saturation regime is not reached with this dose, we choose not to consider the saturation regime. Fig.3 shows that the derivative of the fluorescence intensity increases linearly with increasing dose. Therefore, we can write (2). 𝑑𝐼𝐹 𝑑𝐷
= 𝑘(𝑇)𝐷
𝐼𝐹 = ∫ 𝑘(𝑇) 𝐷𝑑𝐷 = ∫ 𝑘(𝑇(𝑡))𝑃(𝑡)𝑑𝑡
(2) (3)
where k(T(t)) is a kinetic constant depending on the temperature at the time t and P(t) the incident UV irradiation power at time t. The observed fluorescence intensity is also depending on the sensitivity of the camera used for the fluorescence image capture. As a simplification in our case we do not introduce this parameter. The calculated kinetics constants in the following are specific for our camera setup. Under isothermal conditions and constant irradiation power P, the equation (3) can be written as in (4). 1
𝐼𝐹 = 𝐴 + 𝑘(𝑇, 𝑃)𝐷² 2
(4)
where A is a constant and k(T,P) the apparent kinetic constant at temperature T and irradiation P. Fig.6 shows the fluorescence intensity IF of EVA A and C samples as a function of the dose D. The data is fitted with the equation (4) for each sample.
Fig.6. Fluorescence intensity IF of EVA A and C samples as a function of the dose D. The dashed curves are fits according to (5).
As observed in Fig.4, the yellowing leads to an increased photon absorption in the encapsulation material. This increased absorption in visible domain should have a consequence on the current delivered by a cell encapsulated in the material. We calculate the current density of a hypothetical cell encapsulated in the studied material after exposure to a given UV-dose with the following equation (5) 𝐽𝑆𝐶 = 𝑞 ∫ 𝜏(𝜆)𝐸𝑝 (𝜆)𝑄𝑒𝑥𝑡 (𝜆)𝑑𝜆 (5) where τ(λ) the transmittance of the polymer sample, Ep(λ) the energy of the solar spectrum AM1.5G at the wavelength λ and Qext(λ) the quantum efficiency of the solar cell at the wavelength λ. Fig.7 shows the variation of JSC due to the UV-induced yellowing compared to the initial value of Jsc as a function of the fluorescence intensity at different irradiation doses. All material samples show the same linear correlation between the current density loss and the fluorescence intensity. The EVA B is not showing a significant yellowing and is not forming fluorophores. This supports the interdependence of these two properties observed for the other EVAs.
Fig.7. Calculated current density loss due to yellowing in an encapsulated solar cell as a function of the fluorescence intensity of the material. The dashed line is a linear regression of the data of all samples with a fluorescence intensity IF such as 2000
