Luminescence imaging is widely used to identify spatial defects and extract key electrical parameters of photovoltaic devices. To reliably identify defects, high‐quality images are desirable; ...however, acquiring such images implies a higher cost or lower throughput as they require better imaging systems or longer exposure times. This study proposes a deep learning‐based method to effectively diminish the noise in luminescence images, thereby enhancing their quality for inspection and analysis. The proposed method eliminates the requirement for extra hardware expenses or longer exposure times, making it a cost‐effective solution for image enhancement. This approach significantly improves image quality by >30% and >39% in terms of the peak signal‐to‐noise ratio and the structural similarity index, respectively, outperforming state‐of‐the‐art classical denoising algorithms.
A state‐of‐the‐art U‐net model has been developed to effectively denoise luminescence images, resulting in a clearer image on the right from the noisy image on the left. The method outperforms conventional denoising techniques and has the potential to reduce solar cell production costs by increasing throughput and reducing the need for expensive imaging systems.
We compare the recombination properties of a large number of grain boundaries in multicrystalline silicon wafers with different contamination levels and investigate their response to phosphorous ...gettering and hydrogenation. The recombination activity of a grain boundary is quantified in terms of the effective surface recombination velocity S GB based on photoluminescence imaging and 2-D modeling of the emitted photoluminescence signal. Our results show that varying impurity levels along the ingot significantly impact the grain boundary behavior. Grain boundaries from the middle of the ingot become more recombination active after either gettering or hydrogenation alone, whereas grain boundaries from the top and bottom of the ingot have a more varied response. Hydrogenation, in general, is much more effective on gettered grain boundaries compared with as-grown grain boundaries. A close inspection of their injection dependence reveals that while some grain boundaries exhibit little injection dependence before gettering, others show a relatively large injection dependence, with their S GB increasing as the injection level decreases. The former type tend not to be recombination active after both gettering and hydrogenation and are less likely to impact the final cell performance, in comparison with grain boundaries of the latter type.
To operate photovoltaic power plants at maximum capacity, it is desirable to identify cell or module failures in the field at the earliest possible stage. Currently used field inspection methods ...cannot detect many of the electronic defects that can be revealed with luminescence‐based techniques. In this work, photoluminescence images are acquired using the sun as the sole illumination source by separating the weak luminescence signal from the much stronger ambient sunlight signal. This is done by using an appropriate choice of optical filtering and modulation of the cells' bias between the normal operating point and open circuit condition. The switching is achieved by periodically changing the optical generation rate of at least one cell within the module. This changes the biasing condition of all other cells that are connected to the same bypass diode. This method has the advantage that it can deliver high quality images revealing electrical defects in individual cells and entire modules, without requiring any changes to the electrical connections of the photovoltaic system.
High resolution outdoor photoluminescence image of photovoltaic module was obtained under full sunlight using contactless modulation method (Figure A) that can identify various defects that lead to power loss. This image was compared to the indoor electroluminescence image (Figure B). The contactless luminescence modulation is achieved by periodically changing the optical generation rate of at least one cell within the module, which in turn changes the biasing condition of all other cells that are connected to the same bypass diode.
Monitoring the performance of solar modules in a photovoltaic system is critical in order to understand the health of the system. Existing methods for field inspection have limited capability of ...detecting various electronic defects that can, however, be identified with luminescence‐based methods. A contactless outdoor photoluminescence‐imaging based measurement method that uses the sun as the excitation source was presented in our earlier work. This paper extends our previous work and presents two unique applications to (a) identify and quantify local areas of high series resistance within the cells and (b) identify bypass diodes that have failed in open‐circuit. The paper also discusses specific technical considerations of this method. The main merit of this method is that it can be used when the module is under normal operation in the field, without requiring changing of the electrical wiring of the photovoltaic array.
We present a model for predicting the solar cell efficiency potential of multicrystalline silicon bricks prior to sawing. Three model inputs are required: bulk lifetime images from the side faces of ...the bricks, the cell manufacturing process, and its gettering action. The model is set up with numerical device and circuit simulations, but may afterwards be parameterized for inline application. In the example shown here, we chose literature data to quantify the increase in bulk lifetime caused by phosphorus gettering of impurities during cell manufacturing. Our proposed model enables manufacturers to (i) assess initial brick quality in relation to their specific cell production line, (ii) to exclude certain parts of the bricks from cell manufacturing, and (iii) to adjust cell manufacturing to initial material quality. The specific gettering efficiency and cell process can be fed into the model dynamically and need to be calibrated ideally for each material manufacturer and each cell production line. The model presented here can be extended to cast mono and dendritically grown bricks.
Photovoltaic (PV) electricity generation is the fastest growing energy source globally and also humanity's best opportunity to achieve the urgently required deep decarbonisation of the energy sector. ...As PV enters the terawatt scale, with millions of modules in a single PV power plant, quality testing of installed PV modules becomes indispensable to guarantee PV as a reliable long‐term source of electricity. We present a method that extends the use of photoluminescence (PL) imaging to field‐deployed solar modules in full sunlight. The method takes advantage of sunlight absorption in the Earth's atmosphere in a narrow spectral range around 1,135‐nm wavelength and recent developments in ultranarrow bandpass optical filter technology. The technical principles and experimental data are provided. This method lays the foundations for PL imaging, a powerful inspection method for the PV industry and research, to be applied to routine high‐volume inspection of fielded PV modules on large‐scale solar power plants.
We present a method that extends the use of photoluminescence (PL) imaging to field‐deployed solar modules in full sunlight. The method takes advantage of sunlight absorption in the Earth's atmosphere in a narrow spectral range around 1,135‐nm wavelength and recent developments in ultranarrow bandpass optical filter technology. This method lays the foundations for PL imaging, a powerful inspection method for the PV industry and research, to be applied to routine high‐volume inspection of fielded PV modules on large‐scale solar power plants.
End‐of‐line characterization of solar cells is necessary to filter out defective cells and bin cells to avoid power mismatch loss in photovoltaic modules. Current–voltage testers, used by almost any ...photovoltaic company and research laboratory, are costly to maintain and to adapt to recent and predicted morphological changes in solar cells: larger and thinner wafers, half or shingled cells, a wide range of busbar layouts, and more. In this study, we challenge this fundamental technique and propose to bin solar cells and detect defective cells based on a deep learning analysis of their electroluminescence images. The use of electroluminescence imaging addresses the above‐mentioned limitations of the current–voltage technique, as well as allowing faster measurements as it avoids any capacitance effects. By introducing LumiNet, a convolutional neural network end‐to‐end framework, solar cell efficiency bins can be accurately predicted from electroluminescence imaging with a mean error similar to that obtained by current–voltage measurements. The proposed framework is validated on several state‐of‐the‐art mono‐crystalline silicon solar cell structures. We show that photovoltaic modules fabricated using the proposed method would have similar mismatch loss as the traditional current–voltage binning. We then demonstrate the method on half‐cut silicon solar cells. Predicting the half‐cut cell efficiencies, from the deep learning framework, enables manufacturers to assess post‐cutting damages and reassess their binning strategy before module assembly. Furthermore, the deep learning framework is shown to work well even on datasets that have not been previously seen. The trained deep learning LumiNet models' structure and weight are shared with the community to accelerate the adaptation of deep learning for luminescence image analysis in the photovoltaic industry.
LumiNet, a deep learning two‐step framework, learns the efficiency‐relevant features from electroluminescence images and accurately predicts cell efficiencies. In practice, our pre‐trained LumiNet model can predict the efficiency full or half‐cut cells with different morphology and adapted for new tasks such as defect classification via fine‐tuning.