Post‐mortem analysis of a commercial Copper Indium Gallium Diselenide (CIGS) photovoltaic module after potential induced degradation

An extensive post‐mortem analysis was conducted on a commercial copper‐(indium‐gallium)‐diselenide (CIGS) photovoltaic module that degraded after exposure to the high voltage stress of a standardized potential induced degradation (PID) test. We employed a custom‐developed coring technique to extract samples from the full‐size field module, which showed degraded and nondegraded areas (regarded as reference) in electroluminescence after the PID test. The resulting solar cell samples were compared based on their electrical properties and sodium profiles using a wide range of laboratory‐based analysis techniques including photoluminescence and lock‐in thermography imaging, current–voltage measurements, and glow discharge optical emission spectroscopy. The samples that were extracted from the degraded areas of the module showed lower photoluminescence intensity and had significantly lower open‐circuit voltage V(oc) and fill factor (FF) values in comparison with reference samples. An increased content of sodium within the absorber layer was also observed for these samples, linking sodium migration to PID. These observations at the module level are consistent with earlier reports on PID‐stressed CIGS cells and mini‐modules. This is to our knowledge the first reported study of a microscopic investigation on a real‐life full‐scale CIGS module after PID.

other. The potential difference leads to leakage current formation that can take various pathways across and through the module. 4 It can also induce sodium migration which is abundant in the glass sheets. Efficiency and power losses have been linked to this sodium migration from the glass to the solar cell. 5 Sodium accumulation within the solar cell affects its electrical properties.
The PID phenomenon can be observed in all PV technologies.
Copper-(indium-gallium)-diselenide (CIGS) PV modules have demonstrated higher resistance against PID in comparison with multicrystalline Si and a-Si, when mini-modules of each type were compared under the same testing conditions. 6 Severe PID failures in the field have only been reported for a specific CIGS manufacturer. However, considering the current objectives to increase the maximum system voltages to even higher values than ±1500 V, the degradation risks due to PID may also rise for CIGS PV modules. It is therefore important to study and understand the degradation mechanisms behind PID. The underlying mechanisms for PID largely differ for each technology; the mechanisms of shunting at stacking faults and the surface polarisation effect that are common in c-Si PV modules are not observed in CIGS modules. 7,8 For CIGS PV modules, p-n junction damage or transparent conductive oxide (TCO) corrosion due to sodium accumulation are proposed as root causes of the degradation. 6,[9][10][11] However, contradictory results have been reported. We have recently published a review article that presents an overview of the published studies of PID in CIGS PVs including reported field observations, testing methods, observations of sodium migration and leakage current, degradation mechanisms, and mitigation approaches. 12 As presented by the review article, a full understanding of PID is yet to be achieved.
One reason for incomplete understanding of PID is the lack of comparisons and poor correlations, particularly between field and laboratory studies. This is because of differences in test setups and test parameters in reported studies, in addition to varying module and cell design. The extend of analysis after PID testing also varies.
For example, field studies focus mainly on macroscopic investigations such as tracking of the electrical properties upon PID stressing. 4,13 Due to the packaging of the field modules with glass sheets and encapsulant, such studies can only report currentvoltage and leakage current measurements, which yields limited information on the nature of the defects and the root causes of PID.
On the other hand, in-depth microanalyses have only been possible with laboratory-made CIGS devices. 11,14-16 Sodium migration and accumulation within the different layers of CIGS solar cell have been reported, which helps to identify degradation mechanisms behind PID. 9,16,17 However, such studies were carried out on cells or mini-modules without the full encapsulation stack; therefore, the samples, as well as the exposure conditions, may not be representative of field modules. This creates a gap between the two areas of expertise and makes it difficult to correlate the studies and observations to reach common conclusions on PID.
In this study, we aim to bridge this gap by conducting a postmortem analysis at the microscopic scale via laboratory-based techniques on a commercial frameless CIGS field module that was PID tested following the IEC 62804-1 standard. The module was selected from a specific early production batch of a commercial supplier (2010) that exhibited enhanced PID in the field. Areas of interest for postmortem analysis, including both degraded and nondegraded areas of the full-size module, were defined based on the electroluminescence (EL) image of the module after the PID test. From these areas of interest, representative samples were extracted by coring and then unpackaged for an extensive post-mortem analysis. The analysis includes photoluminescence (PL) and illuminated lock-in thermography (ILIT) imaging, current-voltage (I-V) measurements, and glow discharge optical emission spectroscopy (GD-OES).

| The PID test
The IEC 62804-1 test procedure was followed which prescribes application of a bias voltage from an external power source between the grounded module frame and the solar cells through the shorted two connectors from the junction box. The commercial modules we investigate in this study are frameless; therefore, they were supported with metal bars which were then connected to the power source with metal clamps (Figure 1). A voltage of À1000 V was applied for 48 h in a climate chamber at 85 C and 85% relative humidity (RH). I-V measurements were conducted, and EL images were taken both before and after the PID test. Another commercial module from the same production batch was also tested under damp heat conditions (85 C and 85% RH) without electrical bias.
F I G U R E 1 Schematic of the potential induced degradation test setup. The test was conducted on a frameless glass/glass commercial copper-(indium-gallium)-diselenide photovoltaic (PV) module 2.2 | Coring and unpackaging Based on the EL image taken after the PID test, areas of interest for post-mortem analysis were selected. A coring method was custom developed for sample extraction from the defined areas. 18,19 A drilling machine with a hollow diamond drill put together with a water-cooling system was used for this application. The diameter of the drill for this study was chosen to be 3.5 cm, which yielded round samples with a diameter of 3.1 cm. The drilling was applied to the back of the module, which was placed in a tray with a plank support underneath. Coring was followed by unpackaging, where mechanical force was applied at an elevated temperature to separate the front glass from the encapsulant and then to peel off the encapsulant. GreatEyes LumiSolarcell setup was used for PL imaging with a silicon charged-coupled device (CCD) camera. The samples were illuminated with two red light emitting diode (LED) lamps that have a peak emission at 660 nm. All images displayed in this paper were taken with a lens aperture of f/2.8 with integration time of 20 s. A black paper mask was used to cover the glass exposed from the scribes of the cored edge, which can saturate the luminescence image. ILIT measurements were performed using a setup by InfraTec with an ImageIR 8300 infrared camera. Two infrared LED lamps, with a peak emission wavelength at $860 nm, were used to illuminate the samples. All An argon plasma was used for sputtering. The photons emitted were detected by a CCD array. The equipment includes a Grimm-type glow discharge source with an anode of 2.5-mm diameter and a nonconducting cathode plate.

| RESULTS AND DISCUSSIONS
3.1 | Post-mortem analysis of the module Figure 4 shows the I-V curves of the field module measured before and after the PID tests conducted in a climate chamber at 85 C and 85% RH, with a bias of À1000 V for 48 h. Table 1 summarises the change in electrical parameters upon PID stressing. The power has dropped by 50%. The power drop was caused by V oc and FF losses, which are 27% and 28%, respectively. On the contrary, the I sc loss was only 4%. This is consistent with most of the literature on PID degradation of CIGS, where power losses are also related to drastic losses of V oc and FF with no significant change in I sc and R sh . However, there have been studies in literature that reported negative effects on all I-V parameters, observing loss in I sc and shunting. 5,12,20,21 Table 1 also shows the changes in electrical properties for a module that was tested only in damp heat (DH) conditions without any bias application.
There was no power degradation; on the contrary, a slight increase in module performance was noted. A possible explanation is the effect of thermal annealing during the heat exposure. 22 Hence, the power losses in PID-tested module can be attributed to the additional bias application during the test.
The EL images of the field module taken before and after the PID tests are shown in Figure 5. Based on the EL image taken after the PID test, we made two observations: On the other hand, the formation of shunts in the centre of the module was also clearly visible in the EL image, which was not present before the PID test. These shunts are also worth investigating as they  the V oc , and eventually damaging the p-n junction. 10,11,17 As the sodium migration is from the substrate glass in our study, one way to resolve this could be to add a diffusion barrier layer between the glass substrate and the Mo layer. This issue and its possible mitigation routes are discussed in more detail in our recently published CIGS PID review paper. 12 3.3 | Post-mortem analysis of the extracted samples from the area with shunts In addition to samples that were extracted from the degraded area of the module, we also investigated a sample from the area with local shunts formed after the PID test. Figure 10 shows where the sample We further investigated whether the formation of shunts after the PID test was also related to sodium migration by carrying out a GD-OES measurement. According to the PL map of the sample, two areas were defined for the measurement, namely, Bright Area (green box) and Dark Area (red box), corresponding to the active part and the shunt on the sample ( Figure 11). The sodium profile of these regions together with that of a reference sample are displayed in Figure 11. No change in sodium content was observed for either area including the shunted region. We therefore cannot associate the formation of shunts within this area with sodium migration induced by the PID test. However, more statistics is required for a conclusive statement on this matter. We also suggest a study based on EL images of the whole module taken at different intervals during the PID test to further investigate the initial formation and the propagation of shunts.

| CONCLUSIONS
We examined a commercial CIGS PV module that was PID tested according to IEC 62804-1 standards (bias application of À1000 V at 85 C and 85% RH for 48 h). An EL image taken after the PID test Although more work is needed to fully understand the mechanisms at play, the present results connect earlier research on module PID to cell/mini-module PID, confirming that the phenomena observed on small samples also occur at the module level. This was made possible by coring, as this technique allowed us to conduct an in-depth postmortem microscopic analysis on a field module.

ACKNOWLEDGMENTS
We would like to acknowledge the Early Research Program 'Sustainability & Reliability for solar and other (opto-)electronic thin-film devices' (STAR) from TNO for funding.

DATA AVAILABILITY STATEMENT
The data that support the findings will be available in TNO Database at https://www.tno.nl/nl/ following an embargo from the date of publication to allow for commercialization of research findings.