Publications Search Results

Now showing 1 - 10 of 205
  • Publication
    LeTID mitigation via an adapted firing process in p-type PERC cells from SMART cast-monocrystalline, Czochralski and high-performance multicrystalline silicon
    ( 2022)
    Maischner, F.
    ;
    Maus, S.
    ;
    Greulich, J.
    ;
    Lohmüller, S.
    ;
    Lohmüller, E.
    ;
    Saint-Cast, P.
    ;
    Ourinson, D.
    ;
    Vahlman, H.
    ;
    Hergert, K.
    ;
    Riepe, S.
    ;
    Glunz, S.
    ;
    Rein, S.
  • Publication
    The Empirical Digital Twin: Representation Learning on Solar Cell Images and Efficient Defect Detection with Human-in-the-Loop
    ( 2021)
    Kunze, P.
    ;
    Rein, S.
    ;
    Mueller, T.
    ;
    Hemsendorf, M.
    ;
    Ramspeck, K.
    ;
    Demant, M.
    Measurement images of solar cells provide information beyond current-voltage characteristics regarding process and material quality in a spatially resolved manner. However, this information is only partially used because algorithms search for human-defined defects and structures. These labels can be inaccurate and incomplete, a relevance in terms of electrical quality is not necessarily given. Thus, we propose a purely data-based approach to derive a comprehensive representation from the measured images that is meaningful in terms of electrical quality and show how it can be used for efficient defect detection. We call this representation the empirical digital twin. For its calculation, we design a convolutional neural network combining multiple measurement images by correlating them with quality variables. The digital twin is an intermediate representation of the network and summarizes quality-related defect signatures that are visible in the images. We show how this representation can be used to derive sorting criteria for quality inspection within an efficient human-in-the-loop approach detecting defects such as finger interruptions, shunts, etc. The human-in-the-loop method not only needs fewer training samples and thus fewer labels but also improved the 𝐹1-Score detection rate by about 2% on average.
  • Publication
    High-intensity illumination treatments against LeTID - Intensity and temperature dependence of stability and inline feasibility
    ( 2021)
    Vahlman, H.
    ;
    Roder, S.
    ;
    Nekarda, J.
    ;
    Rein, S.
    We study the mitigation of light- and elevated temperature-induced degradation (LeTID) with high-intensity illumination treatments, placing special emphasis on inline feasibility. After the treatments, we investigate the stability upon degradation conditions close to the recently suggested standard, which allows estimating the LeTID behavior during the operating lifetime of solar modules in the field. Subsequently, we map the stability at different treatment intensities and temperatures achievable with an air-cooled tool. We show that, when applying short treatment times, the stability improves with increasing treatment intensity and deteriorates steeply with rising temperature above an optimum region around 250 °C. However, these intensity- and temperature-dependent differences largely vanish when increasing the treatment time sufficiently. We also investigate the significance of darkness/illumination during the cooling ramp from the treatment temperature in view of the LeTID stability. We discuss our results based on suggested defect models of LeTID, and provide hypotheses of the origin of the instabilities observed at the high treatment temperatures. After identifying optimal treatments, we demonstrate that the energy yield loss due to LeTID reduces by over 60% after an inline-feasible process consisting of only 30 s of high-intensity illumination, combined with cooling of the samples from the process temperature under a lower-intensity illumination.
  • Publication
    Learning an Empirical Digital Twin from Measurement Images for a Comprehensive Quality Inspection of Solar Cells
    ( 2021)
    Kunze, P.
    ;
    Rein, S.
    ;
    Hemsendorf, M.
    ;
    Ramspeck, K.
    ;
    Demant, M.
    Measurement images of solar cells contain information about their material- and process-related quality beyond current-voltage characteristics. This information is currently only partially used because most algorithms look for human-defined image features or defects. Herein, a purely data-driven method is proposed to derive the essential image information in terms of the electrical quality within a comprehensive and meaningful representation. This representation is denoted as the empirical digital twin of the cell. Using it, solar cells can be classified according to their defects visible in the measurement images. For this purpose, a human-in-the-loop approach to efficiently develop a classification scheme is presented. Therefore, a convolutional neural network combining various measurement data of a sample by correlating them with quality parameters is designed. The digital twin is an intermediate representation of the network capturing the quality-relevant defect signatures from the images. Human experts can analyze this representation space to identify defect clusters that relate to different process errors, such as finger interruptions and shunts. How the representations are usable to derive sorting criteria for quality inspection is shown. Finally, how the empirical digital twin and the sorting scheme can be used for segmenting the defects without additional label effort is demonstrated.
  • Publication
    Early Stage Quality Assesment in Silicon Ingots from MDP Brick Characterization
    ( 2020)
    Kovvali, A.S.
    ;
    Demant, M.
    ;
    Rebba, B.
    ;
    Schüler, N.
    ;
    Haunschild, J.
    ;
    Rein, S.
    Feedback on the material quality of silicon ingots is highly beneficial in the photovoltaic production chain. It is crucial for crystal growers to improve the quality and optimize the crystallization process. Moreover, for solar cell manufacturers, knowing the quality beforehand helps to sort out the bad quality material thereby reducing the costs and enhancing the total yield. Therefore, rating material quality already on the brick level is highly valuable for the effective optimization of the value chain in both directions. In this paper, we propose a method to classify the silicon bricks based on their electrical quality. Due to our comprehensive data set and feature detection, the model is capable to predict the quality of even edge and corner bricks of the ingot. We introduce a novel feature extraction method to quantify quality-related features from spatially-resolved microwave-detected photoconductivity (MDP) brick measurements. Further, a machine-learning-based prediction model is developed to predict the open-circuit voltage (Voc) of solar cells from these features. A comparative analysis for brick quality estimation for inner and outer bricks of high-performance multi (HPM) and cast-mono (CM) silicon bricks is provided. The best mean absolute error in prediction achieved for HPM and CM materials is 3.1 mV and 4.8 mV, respectively.
  • Publication
    Machine Learning for Advanced Solar Cell Production. Adversarial Denoising, Sub-pixel Alignment and the Digital Twin
    ( 2020)
    Demant, M.
    ;
    Kurumundayil, L.
    ;
    Kunze, P.
    ;
    Woernhoer, A.
    ;
    Kovvali, A.
    ;
    Rein, S.
    Photovoltaic is a main pillar to achieve the transition towards a renewable energy supply. In order to continue the tremendous cost decrease of the last decades, novel cell technologies and production processes are implemented into mass production to improve cell efficiency. Raising their full potential requires novel techniques of quality assurance and data analysis. We present three use-cases along the value chain where machine learning techniques are investigated for quality inspection and process optimization: Adversarial learning to denoise wafer images, alignment of surface structuring processes via sub-pixel coordinate regression, and the development of a digital twin for wafers and solar cells for material and process analysis.
  • Publication
    Comparing Microwave Detected Photoconductance, Quasi Steady State Photoconductance and Photoluminiscence Imaging for Iron Analysis in Silicon
    ( 2020)
    Pengerla, M.
    ;
    Al-Hajjawi, S.
    ;
    Kuruganti, V.
    ;
    Haunschild, J.
    ;
    Schüler, N.
    ;
    Dornich, K.
    ;
    Rein, S.
    Interstitial iron (Fei) is one of the most prominent metallic impurities in crystalline silicon, as it is fast diffusive and highly recombination-active. Its accurate detection is crucial for quality control during solar cell production as iron contamination can significantly limit solar cell efficiency. This work gives a qualitative and quantitative comparison of iron characterization tools including QSSPC (quasi steady state photoconductance), PLI (Photoluminescence Imaging) and MDP (Microwave detected photo conductance). The detection limits, feasibility and accuracy of each tool for iron detection are investigated. In principle, despite of different injection regimes, the absolute iron concentration measured on the different characterising tools is in the same order of magnitude with very good qualitative and quantitative correlation. With the results obtained, the comparison of QSSPC, PLI and MDP showed a mean deviation of 20%.
  • Publication
    The Crystal Growth Explorer: Real-Time Navigable 3D Visualization of Silicon Grains and Defect Related Data in Cast-Mono and Multicrystalline Bricks
    ( 2020)
    Schönauer, J.
    ;
    Demant, M.
    ;
    Trötschler, T.
    ;
    Kovvali, A.S.
    ;
    Schremmer, H.
    ;
    Krenckel, P.
    ;
    Riepe, S.
    ;
    Rein, S.
    Development of novel materials for silicon solar cell production like SMART mono material with functional grain boundaries depends on the human insight and understanding of the structural and dynamic properties of the underlying crystallization process. In this work, we present an analysis and visualization process that allows for comfortable, flexible human exploration and interpretation of various crystallization-related data types ranging from per-wafer photographic images and defect measurements to segmented and 3D-reconstructed grain and defect data, by putting and combining them in a concrete, meaningful, vivid visualization context close to the spatial structure of the actual grown crystal. We address two urgent challenges for smart mono development by combining appropriate data analysis and inspection methods: (1st) analyzing crystal growth characteristics for non-destructive quality inspection and (2nd) investigating defect scenarios for identifying the best crystallization recipes.
  • Publication
    Efficient Deployment of Deep Neural Networks for Quality Inspection of Solar Cells using Smart Labeling
    ( 2020)
    Kunze, P.
    ;
    Greulich, J.
    ;
    Rein, S.
    ;
    Ramspeck, K.
    ;
    Hemsendorf, M.
    ;
    Vetter, A.
    ;
    Demant, M.
    Luminescence images of solar cells show material- and process-related defects in solar cells, which are relevant for monitoring, optimization and processing. Convolutional neural networks (CNNs) allow the reliable segmentation of these defects in images of the solar cells. Nevertheless, the training of CNNs requires a large amount of empirical data, in which the defects have to be labeled expensively by experts. We introduce a method allowing efficient training by using Smart Labels. We show how this technique can be used for process monitoring to detect systematic errors. This approach differs from previous methods, which rely on human heuristics in the form of feature engineering or learning-based methods with human-annotated defects. However, this previous approach has some limitations and risks. These include label mistakes due to overlapping defect structures, poorly reproducible annotations and varying label quality. Furthermore, existing algorithms have to be adapted to new cell lines or a new labeling process is required. We overcome these challenges by avoiding the use of human labels and instead perform the CNN training on the basis of spatially resolved reference measurements, which allows us to calculate spatially resolved labels in less than a second. This purely data-driven approach allows a fast training to quantify defects with physical relevance regarding dark saturation current density (0) and series resistance ( ). The trained CNN achieves a precision of 88% and a recall of 91% for 0 defects while for defects it attains a precision of 78% and a recall of 86%. The accelerated training process allows a fast deployment of deep learning models in the solar cell line.
  • Publication
    Evaluation of Inline High-Intensity Illumination Treatments against LeTID
    ( 2020)
    Vahlman, H.
    ;
    Roder, S.
    ;
    Krauß, K.
    ;
    Nekarda, J.
    ;
    Rein, S.
    High-intensity illumination treatments are a versatile method to reduce the performance loss of solar cells due to LeTID since they can be applied at any point of time between cell fabrication and module assembly. We evaluate the potential of these treatments using a true inline treatment tool and commercially available PERC solar cells. To test the stability of cell performance after the treatments, the cells were exposed to 0.15 suns and 75 °C at open circuit. These conditions are close to a recent testing standard suggestion and allow estimating the cell stability over the typical operating lifetime of solar modules. The results show that there is a window of process parameters which improves the stability of solar cell efficiency without compromising the efficiency immediately after the treatment, i.e. before the stability test. The treatments result in an estimated gain of ~ 3 % in the amount of energy produced by the cells during the operating lifetime of a solar module, corresponding to a reduction of LeTID-related energy yield losses by up to ~ 50 %. Importantly, this gain is achievable with belt speeds compatible with high throughput inline processing.