By Yun Luo, Laure-Emmanuelle Perret-Aebi, Heng-Yu Li
Photovoltaic (PV) modules produced today have a warrantied lifetime of 20-25 years1). One of the most common failures of PV modules is delamination2), which is caused by poor control over the various interfaces, mainly the encapsulant/glass and the encapsulant/back sheet interfaces. To ensure good interfacial adhesion, Ethylene-co-vinyl Cetate (EVA) encapsulants are typically formulated with a silane-based adhesion promoter3), which generally contains an alkoxy-silane and a (meth)acrylic functional group. The alkoxy-silane functionality ensures bonding to the surfaces of the glass and the back sheet, whereas the (meth)acrylic functionality is covalently attached to the EVA via a radical reaction during the curing process. Both interfacial adhesion and EVA crosslinking density have crucial influence on the PV module reliability, especially lifetime. Historically, EVA gel content has been measured by a Soxhlet extraction method using organic solvents such as toluene or xylene. The gel content is defined as the weight percentage of the non-soluble part of the EVA and should typically exceed 80%4). From a polymer point of view, the non-soluble part of the EVA polymer has no correlation with its cross-linking density, which reflects the exact curing degree. The latter can be determined by the G’ and G’’ ratio obtained from Dynamical Mechanical Analysis (DMA) measurements for example. However, in the PV field, the Soxhlet method is established, and one has to adapt to this term while discussing the curing level of EVA in PV modules.
Next to the fact that the Soxhlet extraction is time consuming (usually requiring about 24 h), a piece of EVA has to be taken from the PV module, making it a destructive method. In today’s high-throughput PV module production lines, a much faster quality control method is demanded to ensure a high PV module production yield. To achieve a fast quality control method, Differential Scanning Calorimetry (DSC) and DMA were introduced to reduce the EVA gel content determination time down to 30-60 min5,6). The DSC method measures the amount of curing agent remaining in the EVA after encapsulation, and a calibration curve is used to provide the relation between the DSC and Soxhlet methods. Even though DSC and DMA analyzes allow significantly faster quality control compared with the Soxhlet method, they are also destructive, preventing the quality control of every module produced. Here, we describe a fast and non-destructive optical method to determine the EVA cross-linking density, which allows for in-line analysis of every PV module immediately after the lamination step and could significantly contribute to the quality control that is needed in future PV module production lines. At this stage, the technique presented applies to glass/glass PV modules.
Various commercially available EVA grades were tested for the lamination, and optical studies7). Because of similar results, we decided to proceed with one representative EVA grade for the studies presented here (unless otherwise mentioned). The lamination and thus curing of EVA were performed in a controlled manner in a flat-bed laminator (3S Swiss Solar Systems laminator S1815, 3S Modultec, Lyss, Switzerland). The temperature and pressure profiles of a typical lamination process are shown in Figure 1. After a preheating step on pins for approximately 300 s (plate temperature at 140°C), the pins were set down and a 1-bar pressure was immediately applied to the laminates. The pressure was applied starting at the beginning of the curing time, which occurred under vacuum at about 1 mbar. After a set curing time between 0 and 1300 s, the pressure was removed and the cooling process started8).
The sample architecture used for the optical study is illustrated in Figure 2. For simulation of a glass/glass PV module configuration, a layer of EVA was sandwiched between two identical 4-mm-thick glass plates acting as front and back substrates. The thickness of the EVA layer after the lamination was controlled by placing 0.3-mm-thick stainless steel spacers between the glass plates. The curing times varied from 500 to 1300 s. All samples were cooled naturally in ambient air after the lamination step.
The optical characterization was carried out with an ultraviolet/visible/near-infrared spectrometer (Lambda 900, Perkin Elmer, Waltham, MA, USA). The measured wavelength range was set to 320-2000 nm. Total transmission and total reflection were obtained using an integrating sphere.
Results and Discussions
Fast and Non-Destructive Optical Method to Determine EVA Curing State
From our previous DSC study, the average crystallite size in the cured EVA after cooling shrinks with the EVA curing time9). This relationship is inspiring. The refractive index of the EVA crystallites differs from that of the amorphous phase. This suggests that the crystallites will scatter light differently at defined wavelengths. Therefore, the size and concentration of the crystallites in the EVA encapsulant will strongly affect the optical diffuse transmission through it. If optical transmission is able to detect changes in EVA crystallinity as a function of its gel content, a novel, fast, and non-destructive method for measuring EVA gel content can be developed. For examination of this method, a systematic optical transmission study was performed on glass/EVA/glass laminates with curing times of 500, 600, 700, 900, 1100, 1200, and 1300 s.
As shown in Figure 3, several values are obtained from each laminate in the optical studies, namely Diffuse Transmission (DT), Specular Transmission (ST), Diffuse Reflection (DR), Specular Reflection (SR), Total Transmission (TT), and Total Reflection (TR). The haze factor is defined as the ratio between DT and TT and is commonly used to describe the extent to which light is diffused after travelling through a medium.
The absorption spectra of the samples are shown in Figure 4. The curves overlap in nearly all measured wavelength ranges for all the samples. According to the Beer-Lambert law, the absorption (or ransmittance) is a function of the absorption coefficient and the length of the light path. Assuming a similar concentration of absorbing species in the samples, the overlapped curves reflect satisfactory control over the thickness of the EVA in all samples. A typical absorption spectrum of the glass/EVA/glass laminate possesses several features. The sharp cutoff at around 380 nm is caused by the UV absorber added to the EVA formulation. The other peaks can be assigned by comparison to characteristic absorption spectra reported in the literature10,11). The peaks at roughly 1200 and 1400 nm and the doublet around 1750 nm are all vibrational modes of the C-H group in the EVA. The peaks in the range of 1850-2200 nm are assigned to the vibrational modes of the acetate group of VA segments. Please note that the noisy region at about 870 nm is caused by a detector change in the spectrometer.
Figure 5-a) shows DT curves for 350-600 nm. The colors of the curves correspond to samples with different curing times. For all curves, DT is largest at about 400 nm and monotonically decreases with wavelength. For wavelengths below 400 nm, DT is sharply reduced to nearly zero because of the cutoff of the UV absorber in the EVA. It is clear that DT increases at all wavelengths with increasing curing time. This is further illustrated in Figure 5-b), which shows the haze factor at 400 nm. A gradual decrease of the haze factor is observed for the longer-cured samples (about 2.5% for the sample cured for 500 s and about 0.5% for the sample cured for longer than 1100 s). It is important to point out that the glass used in all laminates is transparent and has no influence on DT of light from 400 to 600 nm, as shown in the Optiwhite reference curve in Figure 5-a). Hence, the observed phenomenon reflects changes of the EVA layers in the laminates. Finally the correlation between EVA gel content and the haze factor is plotted in Figure 5-c). One can see that the EVA gel content is higher for samples showing lower haze factors. The samples for both the gel content determination and the optical study were prepared in parallel following the same lamination process, thus allowing us to compare them directly.
The observed correlation in Figure 5 is explained as follows. The DSC study showed the semi-crystalline nature of the EVA. The crystalline portion of a polymer exhibits a different refractive index from the amorphous portion. The refractive index of EVA is lower for a higher content of VA. For EVA containing 33% VA, the refractive index is about 1.48212). Because the crystallites in EVA mainly consist of polyethylene, the crystallites possess a higher refractive index than the surrounding amorphous phase. The fluctuation in refractive index of EVA containing crystallites results in scattering of transmitted light at those crystallites. The DSC study showed that the average size of the crystallites in cured EVA gradually decreases with increasing curing time. It is also reported in the literature that the crystallinity of EVA drops when reaching higher gel contents13). Therefore, the dependence of DT on the curing time can be attributed to the differing concentration and size of the crystallites in the EVA after the lamination process. As the crystallite sizes are probably much smaller than the wavelength range studied, the Rayleigh scattering mechanism can be used to understand this effect. In principle the DT curve can be fit with the Rayleigh scattering equation after taking into account the wavelength-dependent optical absorption and multiple reflections. This topic will be treated in a future work.
In summary, DT and haze factor of light from 350 to 600 nm are found to depend on the curing time and thus the gel content of the EVA. This dependency can be explained by the different levels of light scattering related to the different crystallinity of the EVA. The discovery of this correlation allows the development of an optical transmission tool for a fast and non-destructive determination of EVA gel content in PV modules, especially in the case of glass/glass modules.
A non-destructive optical method for determining the EVA gel content in a PV module is proposed on the basis of measuring the diffuse transmission (or the haze factor) through the PV module. This fast and non-destructive optical approach holds strong potential for an in-line monitoring of the EVA gel content of every PV module produced, which will significantly contribute to the process quality control that is needed in future high-throughput production lines of PV modules.
Dr. Yun Luo studied physics in Ecole Polytechnique in France and obtained her Ph.D. in material science under the supervision of Prof. Ulrich Gosele from Max-Planck Institute of Microstructure Physics in Halle, Germany in 2005. From August 2008, she joined 3S Swiss Solar Systems in Switzerland and has been holding various research positions. Currently she works as senior research project manager in 3S Modultec (www.3-s.ch) of Meyer Burger Technology Ltd.
Dr. Laure-Emmanuelle Perret-Aebi studied Chemistry at the University of Neuchatel in Switzerland and obtained in 2004 her Ph.D. in Chemistry under the supervision of Prof. Alexander von Zelewsky at the University of Fribourg in Switzerland. From 2004 to 2006 she did a postdoc research at the University of Edinburgh in Scotland and from 2006 to 2008 a postdoc research at the Institute of Physics at the University of Neuchatel. She joined the PVLAB headed by Prof. Christophe Ballif at IMT EPFL in Switzerland in 2009 as a group leader of the Module Design group, in charge of the packaging, reliability and BIPV activities.
MSc. Ir. Heng-Yu Li obtained his bachelor degree in Polymer materials from Beijing University of Chemical Technology. He also held the master degree in Nanoscience from the Technical University of Delft in the Netherlands. Since 2009 till now, he has been a Ph.D. candidate in material science under the supervision of Prof. Christophe Ballif in the Ecole Polytechnique Federale De Lausanne, Switzerland. His research topic is on the understanding of the encapsulation process of Photovoltaic modules with the aim of improving their reliability.
1) Dunlop ED, Halton D. The performance of crystalline silicon photovoltaic solar modules after 22 years of continuous outdoor exposure . Progress in Photovoltaics: Research and Applications 2006; 14: 53-64, DOI: 10.1002/pip.627.
2) Vazquez M, Rey-Stolle I. Photovoltaic module reliability model based on field degradation studies. Progress in Photovoltaics: Research and Applications 2008; 16:419-433, DOI: 10.1002/pip.825
3) Jorgensen GJ, Terwilliger KM, et al. Moisture transport, adhesion, and corrosion protection of PV module packaging materials. Solar Energy Materials and Solar Cells 2006; 90: 2739-2775, doi:10.1016/j.solmat.2006.04.003.
4) ASTM D2765 - 01 Standard Test Methods for Determination of Gel Content and Swell Ratio of Crosslinked Ethylene Plastics 2006
5) Xia Z, Cunningham DW, Wohlgemuth JH. A new method for measuring cross-link density in ethylene vinyl acetate-based encapsulant. Photovoltaics International lite 2009; 5: 16-18;
6) Schubnell M. Investigation of the curing reaction of EVA by DSC and DMA. Photovoltaics International 2010; 7: 131-137.
7) 3S Swiss Solar Systems. Internal technical note 2009.
8) Lange RFM, Luo Y, Polo R, Zahnd J The lamination of (multi)crystalline and thin film based photovoltaic modules. Progress in Photovoltaics: Research and Applications 2010; n/a. DOI: 10.1002/pip.993.
9) Li HY, Perret-Aebi LE, et al. Towards In-line Determination of EVA Gel Content during PV Modules Lamination Processes. Proceedings of 25th EU PVSEC 2010: 4044- 4046.
10) Masahiro M, Yukihiro O. Calibration Models for the Vinyl Acetate Concentration in Ethylene-Vinyl Acetate Copolymers and its On-Line Monitoring by Near-Infrared Spectroscopy and Chemometrics: Use of Band Shifts Associated with Variations in the Vinyl Acetate Concentration to Improve the Models. Applied spectroscopy 2005; 59: 912-919.
11) Shimoyama M, Hayano S, et al. Discrimination of ethylene/vinyl acetate copolymers with different composition and prediction of the content of vinyl acetate in the copolymers and their melting points by near-infrared spectroscopy and chemometrics. Journal of Polymer Science part B: polymer physics 1998; 36: 1529-1537, DOI: 10.1002/(SICI)1099-0488(19980715)36:9<1529::AID-POLB10>3.0.CO;2-7.
12) Sung YT, Kum CK, et al. Effects of crystallinity and crosslinking on the thermal and rheological properties of ethylene vinyl acetate copolymer. Polymer 2005; 46: 11844-11848.
13) TexLoc Refractive Index of Polymers. http://www.texloc.com/closet/cl_refractiveindex.html
For more information, please send your e-mails to firstname.lastname@example.org.
ⓒ2011 www.interpv.net All rights reserved.