Hubert Seigneur

Potential‐induced degradation (PID) remains a significant reliability concern for photovoltaic (PV) modules, arising when a voltage difference between the module frame and the solar cells drives unintended leakage current through the glass–encapsulant stack. Although PID ultimately manifests as PID‐s, PID‐p, or PID‐c, the underlying behavior of the leakage current—its magnitude and time dependence—requires clearer electrochemical interpretation. Traditional explanations attribute the initial transient current to bulk capacitive elements of the glass, encapsulant, and antireflection coatings, and the steady‐state current according to their effective ohmic resistance. More recent studies, however, indicate that electrochemical charge‐transfer processes at the encapsulant–metallization interface can play a dominant role in defining the leakage‐current path. This paper develops a unified electrochemical framework for modeling PID leakage current. First, an RC‐equivalent circuit is formulated by combining conventional RC elements with a Randles‐type interface to capture transient leakage current through double‐layer capacitance and faradaic processes at ionic–electronic boundaries. Second, the steady‐state current–voltage behavior is explained using a linearized Butler–Volmer relationship, showing that the measured ohmic response corresponds to the low‐overpotential limit of charge‐transfer kinetics. Analytical results demonstrate that, for typical module materials—3.2‐mm soda‐lime glass and 0.45‐mm encapsulant—the dominant modulators to PID leakage current are the glass surface resistance (under dry‐surface conditions), the glass bulk capacitance, and the encapsulant resistance (under wet‐surface conditions), with soda lime glass surface and EVA/POE encapsulant resistances primarily governing steady‐state current. The proposed electrochemical model is validated against measured leakage‐current data, showing good agreement in both the magnitude and the time‐dependent evolution of PID leakage current.

This study investigates the polarization‐type potential‐induced degradation (PID‐p) of bifacial glass/glass ‐type passivated emitter rear contact ( ‐PERC) modules in the field and their recovery behavior. Modules were installed with three mounting configurations providing different albedo conditions. System voltage ( 600, 1500, and +1500 V) was applied to the cell circuits, with respect to the grounded module frames. No degradation was observed for positively biased modules, but PID‐p was identified on the rear side when cells are negatively biased, with maximum power dropping during the first days and stabilizing at values up to 8% loss. Electroluminescence images revealed a variation of the cells' PID‐p susceptibility within a module. Three parameters were shown to impact the degradation rate: rear albedo light, voltage, and wetness conditions. Degraded modules were recovered either by (1) a positive bias (+1500 V), (2) outdoor illumination with the front side facing sun, (3) outdoor illumination with the rear side facing sun, or (4) dark storage. A recovery pattern was identified with – parameters decreasing to a local minimum before increasing to full recovery. The proposed mechanism is based on the band bending at the rear ‐type Si/AlO interface, going from inversion to depletion and accumulation states. Full recovery was achieved in 2–7 h for the modules recovered with the rear side facing sun, four to eight nights for the modules positively biased at night, and 10–20 days for the modules with the front side facing sun. Dark storage showed slower recovery rates as – parameters were not improving even after 1 month. The recovery rates were correlated with the net Coulombs transferred during the preceding PID stress: When more Coulombs are transferred during the degradation, the extent of degradation is greater, leading to slower recovery rates.

Bifacial photovoltaic (PV) modules have the advantage of using light reflected off of the ground to contribute to power production. Predicting the energy gain is challenging and requires complex models to do so accurately. Often, module degradation over time is neglected in models for the sake of simplicity or is underestimated. Comparing outdoor and indoor current–voltage (I–V) performance for bifacial modules is more challenging than for monofacial modules, as there are additional variables to consider such as rear albedo non-uniformity, cell mismatch, and their effects on temperature. This challenge is compounded when heterogeneous degradation modes occur, such as polarization-type potential-induced degradation (PID-p). To examine the effects of PID-p on I–V predictions using an empirical data-driven approach, 16 bifacial PERC modules are installed outdoors on racks with different albedo conditions. A subset is exposed to high-voltage biases of −1500 V or +1500 V. Outdoor data are traced at irradiance ranges of 150–250 W/m², 500–600 W/m², and 900–1000 W/m². These curves are corrected using control module temperature, wire resistivity, and module resistance measured indoors. We examine several methods to transform indoor I–V curves to accurately, and more simply than existing methods, approximate outdoor performance for bifacial modules without and with varying levels of PID-p degradation. This way, bifacial performance modeling can be more accessible and informed by fielded, degraded modules. Distributions of percent errors between indoor and outdoor performance parameters and Mean Absolute Percent Errors (MAPEs) are used to assess method quality. Results including low-irradiance data (150–250 W/m²) are discussed but are filtered for quantifying method quality as these data introduce substantial errors. The method with the most optimal tradeoff between low MAPE and analysis simplicity involves measuring the front side of a module indoors at an irradiance equal to plane-of-array irradiance plus the product of module bifaciality and albedo irradiance. This method gives MAPE values of 1–6.5% for non-degraded and 1.6–5.9% for PID-p degraded module performance.

As the photovoltaics (PV) industry grows in sophistication, so must the extent to which systems are characterized. UV Fluorescence (UVF) imaging is a valuable, easy-to-perform, high-throughput, nonintrusive technique for characterizing modules in the field and in the lab. However, UVF is still a relatively new technique, and many in the PV industry are still unaware of its potential. We provide a guideline for obtaining, processing, and interpreting UVF images. We have provided a list of considerations for imaging hardware and settings, a suggested pipeline for image processing, and details on a survey of features shown in UVF images. A new database with UVF images of 7190 modules and another database curated by BrightSpot Automation are publicly available.

This study examines the susceptibility of bifacial glass/glass passivated emitter and rear cell (PERC) modules to potential-induced degradation-polarization (PID-p) in the field. While there are several studies showing PID-p occurring on both front and back faces of bifacial PERC in accelerated tests, we address the yet unclarified behavior in fielded modules. We examine the effects of mounting configuration; specifically, comparing modules mounted near ground and in elevated ground rack configurations. Modules with the cell circuit in −1500 V system voltage configuration, whether mounted on racks about 30 cm above the ground or elevated 2 m high showed mean degradation of 4.5% to 6% in power under standard test conditions over about 2.5 weeks as measured from the front side of the module. This extent of degradation remained sustained for a duration of about 6 months analyzed. Average daytime temperatures of modules in the various mounting configurations were similar and therefore judged to be insufficient to be a primary influence for the modest PID-p rate differences that we observed among mounting configurations. Increased leakage current in the morning suggests morning dew was sustained longer on modules near the ground measured over six months which would be expected to increase the PID-p rate over the long term. However, the main difference seen between the modules on the various mountings during the initial period with up to 6% mean degradation by PID-p was the approximately two times the irradiance from albedo on the rear of modules mounted in elevated ground rack compared to those on the near ground rack. This difference in incident albedo led to a modestly reduced rate of the development of PID-p of the modules on the elevated ground rack. The difference is attributed to the dissipation of PID-p-causing electrical charge by the albedo incident on the module rear. The behavior could be modeled by a sigmoidal equation with consideration of the differences in the insolation on the module rear.

This work investigates several photovoltaic (PV) modules that have shown signs of metal contact corrosion due to field exposure in a hot and humid climate. This includes two multicrystalline silicon aluminum back surface field systems with 10 and 14 years of exposure and one monocrystalline silicon passivated emitter and rear cell system with four years of exposure. A comprehensive, multiscale characterization process is used to evaluate these PV modules in great detail. Current–voltage ($I\text{–}V$), Suns-$V_{\text{OC}}$ measurements, electroluminescence imaging, infrared imaging, and ultraviolet fluorescence imaging were performed, and locations of interest were cored and analyzed using cross-sectional scanning electron microscopy (SEM). A rigorous, quantitative analysis procedure for the cross-sectional SEM images is proposed and implemented. Careful characterization does reveal that some of these PV modules do indeed exhibit the same classic signs of acetic-acid-based corrosion of the glass frit that is present at the silver/silicon interface, which have been observed previously in PV modules exposed to damp heat in an environmental chamber.

In this paper, we use deep learning models to detect and classify various data objects within photovoltaics (PV) module datasheets stored in pdf or other image file formats. We seek to differentiate between regular text, table, header, footer, module drawing, document header, IV curve, PV curve, warranty graph, module image, and icon in order to automate the extraction of useful metadata from datasheets. Our work investigates whether existing computer vision models such as Detectron2 and YOLOv8x can be leveraged to accurately localize and classify the various data types with and without transfer learning. The models were trained on 57 datasheets with multiple pages or 169 pages in total. Detectron2 achieved an overall mAP50-95 of 0.78 without transfer learning (600 epochs and 16 hours) and an mAP50-95 of 0.54 (60 epochs and 2 hours) with transfer learning. YOLOv8x achieved an mAP50-95 of 0.89 without transfer learning (1000 epochs and 2 hours) and an mAP50-95 of 0.9 (20 epochs and several minutes).

The deliberate removal of photovoltaic modules from a string can occur for various reasons encompassing maintenance, measurements, theft, or failure, reducing that string length relative to others when replacement modules are not available and there are not any viable alternative makes and models that could be inserted. This phenomenon, delineated in our prior experimentally validated research, manifests two significant effects: (1) a shift in the ideal maximum power point and (2) the induction of potentially substantial reverse currents in the shortened strings at open-circuit voltage, V_OC. However, the scalability and asymptotic limits of these observed behaviors concerning array size remained undetermined. In this study, we elucidate the operational dynamics of such arrays by manipulating two mismatch-contributing variables in simulated arrays of up to 900 strings: the number of removed modules per string (indicative of the level of mismatch, ranging up to 5) and the quantity of shortened strings (1 to 60). Simulation outcomes underscore that mismatch severity impacts array operation more than the proportion of shortened strings. This research delves into the practical ramifications of operating with shortened strings, including implications for low-irradiance operation and the manifestation of deleterious reverse currents (>35 A in specific cases), emphasizing the need for careful array configuration for optimal performance and safety in these implementations.

Potential induced degradation (PID) is a reliability issue affecting photovoltaic (PV) modules, mainly when PV strings operate under high voltages in hot/humid conditions. Polarization‐type PID (PID‐p) has been known to decrease module performance quickly. PID‐p can be reduced or recovered under the light in some cases, but this effect, as expected, would be less pronounced on the rear side of bifacial PV modules receiving lower irradiance. As bifacial PV modules are projected to dominate the PV market within the next 10 years, it is crucial to understand the PID‐p issue in bifacial modules better. In this study, we performed indoor PID testing to induce PID‐p on 14 commercial bifacial p‐PERC modules with three different module constructions from three manufacturers. Four rounds (+ve and −ve polarities for front and rear sides) of PID testing are done at 25°C, 54% relative humidity (RH) for 168 h using the aluminum foil method. Each module side (front cell side and back cell side) is tested individually under both negative and positive voltage bias. The results show that the highest degradation of 32% in maximum power (Pmax) at standard test conditions (1000 W/m²) and 51% at low irradiance (200 W/m²) has been observed in some cases. Recovery under sunlight is also done, and outcomes show a near‐complete recovery in Pmax. This study presents an extensive experimental methodology and a detailed analysis to systematically and simultaneously/sequentially evaluate multiple construction types of bifacial modules to the PID‐p susceptibility and recovery.

The long-term reliability and durability of photovoltaic (PV) modules have been critical to the recent growth of solar power. The analysis of timeseries data and ultraviolet fluorescent (UVF) images provide insights into critical degradation modes that affect the long-term performance of PV modules. From the timeseries data, we automate the calculation of performance loss rates for 67 different systems located throughout the state of Florida. From UVF imaging, we utilize a pre-trained Mask-RCNN deep learning network for the segmentation of PV modules out of noisy images. In order to perform both of these tasks, a considerable amount of time has to be put into cleaning and reorganizing the input data. This is important both for human understanding of the data and machine action-ability to perform analysis. FAIR principles can be used to assist in this cleaning process to produce datasets that are of high quality for use in machine learning tasks. Here, we apply the FAIR principles to timeseries and image analysis as a showcase for FAIRified workflows for the automation and standardization of datasets for use in machine learning.