Analysis of polyamide and fluoropolymer backsheets: Degradation and insulation failure in field‐aged photovoltaic modules

Durability of photovoltaic (PV) modules is of great concern not only from the point of view of cost‐effectiveness but also from the point of view of safety and sustainability. The backsheet of a PV module is one of the most critical parts of the PV module from the point of view of protection and also one of the most important sources of PV modules' failure; hence, it is of great importance to understand its different forms of failure. In this paper we analyze the case of an 8‐MW PV plant, which had suffered a rapid degradation of their PV modules' backsheets. The case is especially relevant as all the PV modules are from the same model and manufacturer but with different backsheet materials (polyamide and fluoropolymer) and different times of exposure: on one hand, all PV modules originally installed in the plant (i.e., 6 years under operation when tested), and also, extra modules that had been stored indoors for replacement and had been mounted in the plant for less than 1 year when tested, serving as reference modules. In this paper we present the signs of degradation of these PV modules after different times of exposure under real operation using different on‐field and laboratory tests. We propose different techniques for rapid diagnosis of backsheet degradation so that the problem can be detected at a very early stage, before it results in major energy losses or in safety issues.

module, making manufacturers to keep looking for the most economical solutions.
The efforts for cost reduction are reaching every part of a PV plant and PV module, hence, also affecting the protective backsheet of PV modules. Backsheets have two main goals: electrical insulation and protection against outdoor conditions. For this purpose, for years, there has been a preference for a three-layer type of backsheet consisting on a PET core (which provides electrical insulation but has bad weatherability) between two protective fluoropolymer layers. 7,8 This combination has proved to be very reliable in contrast with nonfluoropolymer backsheets, mainly based on polyamides. 9 However, laminated fluoropolymer-based backsheets typically based on PVF or PVDF pose two main disadvantages against polyamides. First, polyamide-based backsheets are cheaper than fluoropolymer backsheets due to the material itself 9,10 but also due to the manufacturing process when coextruded. 11 Moreover, trim and scrap materials from the polyamide backsheet manufacturing can be collected and re-processed, which opens the door to future PV modules' recycling, although separation of this type of backsheet from the module is still a challenge to be solved. 12 For these reasons there has been a growing interest in avoiding fluoropolymers and using other polymers such as polyamides for the manufacturing of backsheets.
Polyamides, which comprise a big family of polymers, including nylons, are widely used in the industry (textile or automotive for instance). Being a thermoplastic, it is easy to use for manufacturing fibers or solid parts but has good chemical and mechanical properties when solid. However, its use in the PV industry has proved that polyamide-based backsheets have poor weatherability, mainly due to UV degradation. For this reason, some innovations are already appearing such as polyamide-ionomer alloys 12 or even new approaches as the use of polyolefins like polypropylene. 13 However, while these products are being introduced in the market, polyamides have already been used massively in previous years and are present in PV plants that should last for around 25 years. Thus, the understanding of the degradation modes of polyamides is still of high interest in order to have an early identification of this problem.
Polyamide degradation in PV modules' backsheets is a known issue and has been reported in several papers. [14][15][16] In these works, polyamide in PV modules' backsheet has been seen to be degraded by UV, thermal stress, or dampness; and, overall, by a combination of these and other mechanisms.
One factor that may affect the degradation of polyamide backsheets, as explained in previous works, 17,18 is the use of titanium dioxide, TiO 2 , which is mixed in the polyamide giving to the backsheet its characteristic white color, used to reflect the light that otherwise could harm the backsheet and would also heat up the module. Unfortunately, TiO 2 also acts as a photocatalyst, absorbing some of the UV light and giving of charge carriers, which facilitates reduction and oxidation reactions with the surrounding organic compounds. This wellknown process is in fact used for the degradation of organic pollutants from wastewater [18][19][20] or, in the textile industry, where TiO 2 is used to deluster precisely nylon, a type of polyamide. 21 This same gloss loss is in fact observed in degraded polyamide backsheets. Moreover, this reaction is also a common problem with some TiO 2 -based pigments, which tend to decompose in fine detached particles, which have the appearance of chalk, hence known as chalking. [22][23][24] Again, this chalking process is also observed in some of the polyamide backsheets.
This material loss, in turn, may expose new material to UV light, which will follow the same process, further weakening the backsheet.
Nevertheless, the exact mechanism relies on different aspects like the concentration and structure of TiO 2 (rutile and anatase, the most common polymorphs of TiO 2 , are known to behave differently in this reaction, 16,18 ) or the exact composition of the polyamide and other organic compounds that may be present. 17 Moreover, as previously stated, backsheets may be degraded without any radiation but under other stressing factors, for example, under the combination of thermal stress and dampness, revealing that different aging mechanisms coexist in the field. 14,15,25 In fact, these mechanisms change between different places; hence, the degradation mode may also depend on climate; thus, the observed degradation modes of polyamide backsheets are different as seen in several studies. The complexity of all these factors is further studied in Lyu et al., 26 which comprises a detailed degradation analysis of polyamide backsheets of field-aged PV modules from different locations around the world and accelerated tests, which led, in turn, to the design of more reliable accelerated tests.
In other different studies, 27,28 an analysis is conducted in order to visually assess the degradation of PV modules following NREL's Visual Inspection Checklist. 29 Different PV modules are gathered for the study from different climates and that have been installed in different years. Although the material of the backsheet is not specified, it is interesting that chalking appeared in practically all climates (only cold climate has no modules with chalking but there are only three samples in this category). It is also interesting to note that in this study chalking only appeared in modules aged 11-20 years old. Most of the PV modules in this age range had also encapsulant discoloration. Despite this apparent correlation 30 shows how all PV module in a 22-year-old PV system showed chalking, but only 10% suffered encapsulant discoloration. More importantly, despite the presence of chalking, insulation resistance for most of the modules was above the threshold level recommended by the norm IEC 61215 (40 MΩ m 2 measured between the frame and the connectors, which are short circuited for the test), showing that chalking may appear before insulation resistance fails.
In Gebhardt 17 chalking and its relationship with cracks in the backsheet and insulation failure is studied. As it is explained, chalking may appear in many kinds of polymers including fluoropolymers, however, only polyamide backsheets have shown chalking as the root of further problems as micro-cracks (superficial cracks not visible to the naked eye). However, the outcome highly depends on several factors such as the structure of TiO 2 , its location in the backsheet (outer or inner layer), presence of other additives, and so on. In any case, chalking appears as an early indicator (but not the necessarily the cause) of an accelerated degradation process that may end up with visible cracks (macro-cracks) and lower insulation resistance values, which may fall below values that result in complete failure. To sum up, this paper shows how the relation between chalking, cracks and low insulation resistance is not a simple relationship but rather depends on many physical factors of the backsheet.
More specifically, previous works 11,16 suggest that although micro-cracks can be related to chalking, macro-cracks are not and are rather caused by thermo-mechanical stress. Moreover, previous works 31,32 show that macro-cracks in polyamide backsheets appear due to a combination of thermo-mechanical stress in presence of acetic acid from EVA encapsulant degradation, thus resulting in through cracks formation from the inner layers outwards.
Finally, Omazic et al. 33 show that climate also plays an important role in the degradation of PV modules. Based on the understanding of the reactions underlying the chalking process, warmer climates are thought to accelerate the chalking process as well as delamination.
However, all types of degradation modes are encountered in all climates and one of the main conclusions is that more studies with field-aged PV modules are needed in order to find stronger conclusions.
In this paper, we provide an in-depth study of backsheet failure in were some modules that had been stored indoors in case they were needed as replacement, which were finally installed in 2015 as substitutes of some broken modules; that is, they had been installed in the PV plant for less than 1 year when they were tested. These modules showed almost no sign of degradation and served as reference modules.
As expected, different materials and different times of exposure lead to different evolution of the backsheet ranging from no degradation at all, chalking, cracking and total insulation failure. Backsheets have been thoroughly tested in order to gain insight in their degradation process. Tests are divided in two main categories: (1) on-field tests, that is, insulation resistance measurement and visual inspection for the identification of cracks and chalking; and (2) laboratory tests, that is, hydrophobicity, surface roughness, Fourier transform infrared spectroscopy (FT-IR) for finer material identification, scanning electron microscopy (SEM) for the visualization of chalking particles and microcracks, and surface light reflection via UV-Vis-NIR spectroscopy. Therefore, this paper offers an in-depth analysis of a set of backsheet samples that will help gain understanding in their degradation process by studying different materials and time exposures in otherwise similar conditions. The paper is structured as follows: Section 2 gives an overview of the PV plant and PV modules analyzed as well as the faults encountered after some years of operation. Section 3 shows the results of the on-field tests while Section 3 gives an insight of the properties of the PV modules by analyzing samples of their backsheets at the laboratory. Finally, conclusions are discussed in Section 5.

| Performance issues
Once installed in 2011, the PV plant operated normally for 5 years.
During these first years, some ground fault alarms were reported by the inverters but not too many and, more importantly, very rarely should be said that minor ground faults should always be regarded as a prompt sign of some health problem in the PV plant, and it is important to assess the plant before the number of alarms grow, that is, before more energy production is lost. In this case, as seen in Figure 2, it would have been better to thoroughly check the PV plant around December of 2014. However, note that ground faults may not only be caused by the PV modules, as it was the case of this PV plant, but also could arise from wiring or other parts of the PV plant. As such, it is very important to have the tools and techniques to rapidly diagnose each part of the plant, as some of the techniques we show in this paper regarding PV modules. In the case of this PV plant however, it was not until the beginning of 2017 when PV modules where tested, and it was found that many of the PV panels' backsheets already showed substantial chalking and some visible cracks. Subsequently, a thorough analysis of the PV modules, both on-field and in the laboratory, was carried out as shown in the following sections.

| Overview
Following the insulation failure events, three on-field tests were carried out in order to quantify the problem, understand its source, and try to predict its evolution.
First, as there was an evident problem of visible cracks on the backsheets, which can be related to low insulation resistance, all PV modules were checked for visible cracks and located in the PV plant, in order to identify whether the problem was located in a particular area or was rather widespread. Second, in order to assess the insulation failure events, the insulation resistance of every PV module was measured. Finally, assuming that chalking may be the primary stage of the degradation problem, backsheets were checked for chalking, sampling one out of every seven PV modules in the plant. All three tests are described individually in the following subsections and summarized in Section 3.5.

| Visible cracks on backsheet
First, all PV modules were inspected for visible cracks (or macrocracks) like those shown in Figure 3. This inspection was made not only to quantify the extent of the problem, but also to localize the failing modules in order to realize whether the problem was occurring in a particular part of the PV plant or, on the contrary, whether the problem was widespread.
As a result, this inspection revealed that 4932 PV modules, that is, 14.1% of the PV plant, were affected by visible cracks. Moreover, as shown in Figure 4, even though there were some areas more affected than others, the problem was seen to be spread out throughout all the PV plant.

| Insulation resistance test
Second, in order to understand were all the insulation failure alarms came from, all PV modules were tested for low insulation resistance.
The measurement was carried out following the manufacturer's guide, using the "Benning PV 1" tester. The results showed that 1257 modules (3.7% of the PV plant) had an insulation resistance value below the accepted value in the norm IEC 61215 (24.9 MΩ for these modules' area), which justifies the insulation alarms. More importantly, PV modules with a fluoropolymer backsheet, which showed no cases of visible cracks, also presented no cases of low insulation resistance.
Note that right after this measuring campaign, all 1257 defective modules found were replaced by extra PV modules that had been provided at the time of the installation of the PV plant and that have been kept indoors and, hence, have not suffered the outdoors conditions that lead to degradation of the backsheet. This is of great importance as these modules make good reference PV modules as they are from the same manufacturer's batch. Moreover, both types of backsheet (polyamide and fluoropolymer) were present in these newly installed PV modules.

| Chalking test
Finally, speculating that chalking may be an early sign of the degradation process that leads to low insulation resistance values, one out of seven PV modules were checked for chalking following the visual inspection procedure defined by NREL and also adopted by NCPRE. 28,29 In this case it was revealed that most PV modules' backsheets (93.6%) suffered from chalking, as the one shown in Figure 5.
Moreover, there were no cases of chalking among the fluoropolymer backsheet modules.

| Summary of the on-field tests
Analyzing the on-field results, it was readily clear that fluoropolymer backsheets had no apparent signs of degradation. On the other hand, polyamide backsheets could be classified in two groups according to their insulation resistance and appearance of visible cracks. In order to make this classification, PV modules were firstly grouped by their serial number, assuming that manufacturing processes may vary in different batches. In particular, the PV modules were classified by their eight first numbers in their serial number, resulting in 77 families with an average of 455 PV modules per family. Then, a family was classed as defective (named PA1) if the number of PV modules with cracks or low insulation resistance was above 3%. Otherwise, the subfamily was classed as healthy (PA2). Note that most PV modules in both groups PA1 and PA2 suffered from chalking, showing that all polyamide PV modules were undergoing a degradation process, only that those in group PA1 seem to be at a later stage than those in group PA2.
According to this classification, as shown in Table 1   • 1 PV modules from group PA1 with substantial chalking, visible cracks and low insulation resistance (test module name: MPA1).
• 2 PV modules from group PA2. Both with substantial chalking but no cracks. One below and another above the minimum insulation resistance stated by the norm (named MPA2_1 and MPA2_2).   The identification names of the tested PV modules along with their main characteristics are summarized in Table 2.

| Chalking, and hydrophobicity tests
The first tests carried out in the laboratory were a chalking test, by means of passing the hand over the backsheet as indicated in Packard et al. 29 and a hydrophobicity test, by spraying water and observing the reaction, either soaking or repelling.
The chalking test showed that the only backsheets not presenting chalking are the fluoropolymer ones (both old and new, i.e., MFP_1, MFP_2, MFPref_1, and MFPref_2) and the newly installed polyamide one (MPAref). All the rest, that is, polyamide backsheets of PV modules originally installed in the PV plant (MPA1, MPA2_1, and MPA2_2) did present chalking. As an example, Figure 6 shows the chalking results for modules MPA2_1 and MPAref, both polyamides from the original batches, the first one exposed for 6 years and the other one recently installed.
Regarding hydrophobicity, every module's backsheet was sprayed with water showing that those presenting chalking soaked the water while those not presenting chalking, including the newly installed T A B L E 1 Summary of on-field tests  The most significant result from this test is not only that older polyamide backsheets tend to soak the water but, most importantly, that those recently installed (from the same batch) are hydrophobic, proving that, with the pass of time under normal operating conditions, they lose this property, hence, revealing a degradation process.

| Infrared spectroscopy test
In order to assess the degree of degradation of the material, an infrared spectroscopy test was carried out to the selected PV modules using a FT-IR Perkin-Elmer Frontier. Samples were cleaned with ethanol and sliced with a microtome in order to obtain a sample with a thickness lower than 10 μm that can be analyzed by transmission. Although not shown in the figures, this test was carried out also against laboratory samples of different polymers in order to confirm their nature, that is, whether they were polyamides or fluoropolymers.
As it can be observed in Figure 8 On the other hand, spectroscopy curves in Figure 9 show that polyamide backsheets that have been in the plant for 6 years (MPA1, MPA2_1 and MPA2_2) have changed with respect to the reference F I G U R E 8 Baselined spectra of photovoltaic (PV) modules with fluoropolymer backsheet PV module (MPAref) that has only been outdoors for less than a year.
In particular, peaks of modules that had been exposed for 6 years have notably decreased their magnitude relative to MPAref. These  Additionally, Figure 11 shows detached material at a magnifica-  were not suffering any abnormal degradation while polyamide backsheets did. Moreover, it could be observed that polyamide backsheets were all degrading but at slightly different rates.

| UV-Vis spectroscopy
Additionally, originally installed PV modules could be compared against PV modules from the original batch but that had been kept indoors and were installed as replacement of failing PV modules less than a year before they were tested. Although the newly installed PV modules were from the same batch and included both fluoropolymer and polyamide backsheets, none of them showed any degradation F I G U R E 1 1 Scanning electron microscopy of selected photovoltaic (PV) modules at Â50,000 augmentation F I G U R E 1 2 Surface gloss test. Fluoropolymer backsheets and newly installed polyamide backsheet have high reflectance while old polyamide backsheets have lost this reflectance that originally had, as sample MPAref shows.
signs at a visible level. However, when observed at a microscopic level, some material detachment could be already observed in the polyamide backsheets. This last finding is relevant as it shows that the bad state of the polyamide backsheets is not just a matter of time, but also exposure to normal operating conditions in a PV plant. More importantly, it has been observed that SEM microscopy is the best way among those used in this work to detect degradation of the backsheet at a very early stage.
When dealing with widespread and progressive degradation of any element in a PV plant, PV modules in this case, it is of great importance to be able to detect it on time so that they can be replaced before the problems grow exponentially and too much energy is lost; or before warranties expire. As such, one of the conclusions of this work is that ground faults represent a very good and early indicator of a degradation process going on and should be treated with great care. Nonetheless, these alarms may be due to many elements of the PV plant and it can be difficult to find the origin of the fault. As such, it is of great importance to have tools that are quick to implement in order to detect the root of the problem. In this paper, we have shown different techniques to assess the PV modules' backsheet degradation level, from visual inspection to different laboratory tests. In this case, it was especially relevant that, although recently installed polyamide backsheets had no visual signs of degradation, SEM microscopy images showed that the degradation process had already started. In this particular case, action should had been taken around 2014, when ground faults started to increase. At that time, visual inspections may not have been enough to detect the problem of the PV modules' backsheet, but SEM microscopy images would have already shown the degradation process before the problem escalated, that is, about 2 years before the energy lost due to ground faults started to become a problem and giving more reaction time for corrective measures.