Solar cells allow the energy from the sun to be converted into electrical energy; this makes solar energy an environmentally friendly, sustainable alternative to fossil fuel energy sources. Solar cells are connected together in a photovoltaic (PV) module to provide the higher current, voltage and power outputs necessary for electrical applications. However, the performance of PV modules can limited by the degradation and defects. PV modules can be characterised using various opto-electronic techniques, each providing information about the performance of the module. The current-voltage (I-V) characteristic curve of a module being the most commonly used characterisation technique. The I-V curve is typically measured in outdoor, fully illuminated, conditions. This allows performance parameters such as short circuit current (ISC), open circuit voltage (VOC) and maximum power (PMAX) to be determined. However, it can be difficult to determine the root cause of the performance drop from the I-V curve alone. Electroluminescence (EL) is a module characterisation technique that allows defects and failures in PV modules to be successfully identified. This study investigates the characterisation of solar cells and photovoltaic modules using EL. EL occurs when a solar cell or module is forward biased and the injected electron-hole pairs recombine radiatively. The intensity of the emitted EL is related the applied voltage and the material properties. EL imaging is a useful characterisation technique in identifying module defects and failures. Defects such as micro-cracks, broken contact fingers and fractures are detected in EL images as well as material features such as grain boundaries. The common defects in crystalline silicon are catalogued and the possible causes are discussed. An experimental setup was developed in order to systematically take a high resolution EL image of every cell in the module and record the applied voltage and current. This produces a very detailed, clear, image of each cell with a pixel size in the micrometre range. This process is time consuming to acquire an EL image of an entire module so alternatively a different setup can be used and an EL image of a whole module can be captured in a single frame with an increased pixel size in the millimetre range. For EL imaging a silicon charge-coupled device (CCD) camera was used because it has very good spatial resolution however this sensor is only sensitive to wavelength in the range of 300-1200 nm. There is an overlap in wavelengths from about 900 to 1100 nm allowing the EL emitted from silicon solar cells to be detected. In conjunction with the high-resolution EL system an image processing program was developed to crop, adjust and align the images so only the relevant cell was included. This program also automatically detects certain defects that have a regular shape. Micro-cracks, broken fingers and striation rings are automatically identified. The program has an adjustable sensitivity to identify small or large defects. Defective cells are distinguished from undamaged cells by comparing the binary images to the ideal, undamaged cell. The current-voltage curves and the performance parameters of modules were compared with the EL images in order to discuss and identify power limiting defects. Features that remove significant portions of the cell from electrical contact such as micro-cracks are shown to have a larger effect of the performance of the module. Other features such as broken contact fingers, contact forming failures and striation rings do not significantly lower the performance of the module. Thus an understanding of how different features affect the module performance is important in order to correctly interpret the EL results. The intensity of the luminescence emitted is related to the applied voltage and the quantum efficiency of the cell material. The spectrum of the emitted luminescence was modelled and related to the recombination properties of the cell such as surface recombination velocity and minority carrier diffusion length/lifetime. In this study the emitted spectrum was modelled and the effects of recombination properties of the cell on the emitted spectrum were examined. The spectrum of the detected EL was modelled, dependent on the sensitivity of the camera, the transmission of the filters and the emitted photon flux. The integration of short-pass filters into the experimental setup in order to isolate short-wavelength luminescence was discussed. There is a proportional relationship between the intensity of the emitted EL and the local junction voltage. Resistive losses like series and shunt resistances lower the applied voltage and thus affect the EL image. The voltage dependence was assessed by comparing EL images taken at different applied biases. Analysis of the variation in EL intensity with voltage was successful in determining the origin of certain features in an EL image. Certain defects, those that are related to series resistance or shunting are highly voltage dependent. When a feature has little or no dependence on voltage then the defect could be in the laminate layers and not in the cell material. The results of this study allow for in-depth analysis of the defects found in PV modules using the high resolution EL imaging system and the image processing routine. The development of an image processing routine allows the interpretation of the EL image to be done automatically, resulting in a faster and more efficient process. By understanding the defects visible in the EL image, the test is more meaningful and allows the results to be used to predict module performance and potential failures.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:nmmu/vital:26866 |
| Date | January 2015 |
| Creators | Crozier, Jacqueline Louise |
| Publisher | Nelson Mandela Metropolitan University, Faculty of Science |
| Source Sets | South African National ETD Portal |
| Language | English |
| Detected Language | English |
| Type | Thesis, Doctoral, PhD |
| Format | xv, 140 leaves, pdf |
| Rights | Nelson Mandela Metropolitan University |
Page generated in 0.1568 seconds