TY - JOUR A1 - Mohd, Zubair Akhtar A1 - Schmid, Maximilian A1 - Zippelius, Andreas A1 - Elger, Gordon T1 - Solder joint lifetime model using AI framework operating on FEA data JF - Engineering Failure Analysis N2 - The thermo-mechanical reliability of electronic systems is often limited by the crack growth within the solder joints. Addressing this issue requires careful consideration of the design of the package and solder pads. Finite Element Analysis (FEA) is widely used to predict crack growth and to model their lifetime. Traditionally, FEA post-processing methods rely on human expertise to select appropriate regions for evaluating plastic and creep strain at critical locations and correlating these values with experimental data using the Coffin-Manson equation, which predicts fatigue lifetime based on cyclic plastic strain. This study introduces a novel method for FEA post-processing of surface-mounted devices (SMD) on printed circuit boards (PCB) using artificial intelligence. The method transforms the FEA data into a 2D grid map of creep strain values and employs a Convolutional Neural Network (CNN) for automatic feature extraction. Afterwards, a fully connected layer correlates the extracted features with the experimental measured solder joint lifetime, effectively capturing nonlinear relationships. The study focuses on the development of the concept of crack formation in the solder interconnects of ceramic based high-power LED packages used in the automotive industry for headlights. The validated FEA model is based on an extensive data set of 1800 LED packages including seven different ceramic-based LED packages and five different solders. The design of the ceramic LED package covers two-pad and three-pad footprint for soldering and thin film and thick film metallized ceramic carriers. Results show a strong agreement (R2 Score is 99.867 %) between simulations and experimental data for ceramic LED packages. This automatic feature extraction from FEA data sets a new benchmark for improving solder reliability predictions, and it has proved to be better than established methods for lifetime prediction of solder joints. UR - https://doi.org/10.1016/j.engfailanal.2024.109032 Y1 - 2024 UR - https://doi.org/10.1016/j.engfailanal.2024.109032 UR - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:573-53803 SN - 1350-6307 VL - 2025 IS - 167, Part B PB - Elsevier CY - Oxford ER - TY - JOUR A1 - Mohd, Zubair Akhtar A1 - Schmid, Maximilian A1 - Elger, Gordon T1 - AI-driven point cloud framework for predicting solder joint reliability using 3D FEA data JF - Scientific Reports N2 - Crack propagation in solder joints remains a major challenge impacting the thermo-mechanical reliability of electronic devices, underscoring the importance of optimizing package and solder pad designs. Traditional Finite Element Analysis (FEA) techniques for predicting solder joint lifespan often rely on manual post-processing to identify high-risk regions for plastic strain accumulation. However, this manual process can fail to detect complex and subtle failure mechanisms and purely based on averaging the creep strain and correlating it to lifetime values collected from experiments using Coffin Manson equation. To address these limitations, this study presents an Artificial Intelligence (AI) framework designed for automated 3D FEA post-processing of surface-mounted devices (SMDs) assembled to Printed Circuit Board (PCB). This framework integrates 3D Convolutional Neural Networks (CNNs) and PointNet architectures to automatically extract complex spatial features from 3D FEA data. These learned features are then linked to experimentally measured solder joint lifetimes through fully connected neural network layers, allowing the model to capture complex and nonlinear failure behaviours. The research specifically targets crack development in solder joints of ceramic-based high-power LED packages used in automotive lighting systems. This dataset included variations in two-pad and three-pad configurations, as well as thin and thick film metallized ceramic substrates. Results from the study demonstrate that the PointNet model outperforms the 3D CNN, achieving a high correlation with experimental data (R2 = 99.91%). This AI-driven, automated feature extraction approach significantly improves the accuracy and provide the more reliable models for solder joint lifetime predictions, offering a substantial improvement over traditional method. UR - https://doi.org/10.1038/s41598-025-06902-0 Y1 - 2025 UR - https://doi.org/10.1038/s41598-025-06902-0 UR - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:573-60961 SN - 2045-2322 VL - 15 PB - Springer Nature CY - London ER - TY - INPR A1 - Mohd, Zubair Akhtar A1 - Schmid, Maximilian A1 - Elger, Gordon T1 - AI-Driven Point Cloud Framework for Predicting Solder Joint Reliability using 3D FEA Data T2 - Research Square N2 - Crack propagation in solder joints remains a critical challenge affecting the thermo-mechanical reliability of electronic devices, emphasizing the need for optimized package and solder pad designs. Traditional Finite Element Analysis (FEA) methods for predicting solder joint lifespan rely heavily on manual post-processing, where high-risk regions for plastic strain accumulation are identified. However, these approaches often overlook intricate failure mechanisms, as they primarily average creep strain and correlate it with experimental lifetime data using the Coffin-Manson equation, limiting their predictive accuracy. To overcome these limitations, this study introduces a novel AI-driven framework that automates 3D FEA post-processing for surface-mounted devices (SMDs) connected to printed circuit boards (PCBs). Unlike traditional methods, this framework leverages deep learning architectures—specifically, 3D Convolutional Neural Networks (CNNs) and PointNet—to extract complex spatial features directly from 3D FEA data, eliminating the need for manual interpretation. These learned features are then mapped to experimentally measured solder joint lifetimes through fully connected neural network layers, allowing the model to capture nonlinear failure behaviours that conventional methods fail to recognize. The research focuses on crack propagation in ceramic-based high-power LED packages used in automotive lighting systems, incorporating variations in two-pad and three-pad configurations, as well as thin and thick film metallized ceramic substrates with validated FEA models. Comparative analysis shows that PointNet significantly outperforms 3D CNNs, achieving an exceptionally high correlation with experimental data (R² = 99.99%). This AI-driven automated feature extraction and lifetime prediction approach marks a major advancement over traditional FEA-based methods, offering superior accuracy, reliability, and scalability for predicting solder joint reliability in microelectronics. UR - https://doi.org/10.21203/rs.3.rs-6173485/v1 Y1 - 2025 UR - https://doi.org/10.21203/rs.3.rs-6173485/v1 UR - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:573-58308 SN - 2693-5015 PB - Research Square CY - Durham ER - TY - JOUR A1 - Zippelius, Andreas A1 - Mohd, Zubair Akhtar A1 - Schmid, Maximilian A1 - Elger, Gordon T1 - Comparison of different Input data for the prediction of LED solder joints using Artificial Neural Networks JF - IEEE Transactions on Device and Materials Reliability N2 - Scarcity of raw data is a major issue for applying data driven methods to reliability prediction, so making the best use of what is available is critical. This paper studies how different aspects of measurement data can be used best. Specifically, the reliability of the solder joint of LED packages is predicted based on Transient Thermal Analysis and Scanning Acoustic Microscopy data from a large measurement campaign. The impact of using full temporal information vs measurements at individual datapoints is investigated as well as the benefit of including the SAM data, and different ways of presenting the TTA information, either as a full curve or as expert-selected features. The impact of formatting categorical information of solder and LED package type as one-hot encoding or using embeddings is considered. Finally, the performance for Pass/Fail predictions of the best identified model with a model architecture developed on a similar dataset is compared. We identified the most relevant sources of information for predicting the behavior and the best format for the data, which helps guide the choice for future model architectures. UR - https://doi.org/10.1109/TDMR.2025.3633876 Y1 - 2025 UR - https://doi.org/10.1109/TDMR.2025.3633876 SN - 1558-2574 PB - IEEE CY - New York ER - TY - JOUR A1 - Pandey, Amit A1 - Mohd, Zubair Akhtar A1 - Veettil, Nandana Kappuva A1 - Wunderle, Bernhard A1 - Elger, Gordon T1 - Quantitative Kernel estimation from traffic signs using slanted edge spatial frequency response as a sharpness metric JF - Scientific Reports N2 - Sharpness is a critical optical property of automotive cameras, measured by the spatial frequency response (SFR) within the end-of-line (EOL) test after manufacturing. This work presents a method to estimate the blurring kernel of an automotive camera, which could be the first step toward state monitoring of automotive cameras. To achieve this, Principal Component Analysis (PCA) was performed, using synthetic kernels generated by Zemax. The PCA model was built with approximately 1300 base kernels representing spatially variant point spread functions (PSFs). This model generates kernel samples during the estimation process. Synthetic images were created by convolving the synthetic kernels with reference traffic sign images and compared with real-life data captured by an automotive camera. These synthetic data were utilized for algorithm development, and later on, validation was performed on real-life data. The algorithm extracts two pixels regions of interest (ROIs) containing slanted edges from the blurred image and crops matching ROIs from a reference sharp image. Each candidate kernel was used to blur the reference ROIs, and the resulting SFR was compared with the blurred ROIs’ SFR. Differential evolution optimization minimizes the SFR difference, selecting the kernel that best matches the observed blur. The final kernel was evaluated against the true kernel for accuracy. The structural similarity index measure (SSIM) between the original and estimated blurred ROIs ranges from 0.808 to 0.945. For true vs. estimated kernels, SSIM varies from 0.92 to 0.98. Pearson correlation coefficients range from 0.84 to 0.99, Cosine similarity from 0.86 to 0.99, and mean squared error (MSE) from to . Validation on real-life camera images showed that the SSIM between the estimated and blurred ROI was >0.82, showing promising accuracy in kernel estimation, which could be used towards in-field monitoring of camera sharpness degradation. UR - https://doi.org/10.1038/s41598-026-40556-w Y1 - 2026 UR - https://doi.org/10.1038/s41598-026-40556-w UR - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:573-66915 SN - 2045-2322 VL - 16 IS - 1 PB - Springer Nature CY - London ER - TY - INPR A1 - Pandey, Amit A1 - Mohd, Zubair Akhtar A1 - Veettil, Nandana Kappuva A1 - Wunderle, Bernhard A1 - Elger, Gordon T1 - Quantitative Kernel Estimation from Traffic Signs using Slanted Edge Spatial Frequency Response as a Sharpness Metric T2 - Research Square N2 - The sharpness is a critical optical property of automotive cameras, measured by the Spatial Frequency Response (SFR) within the end of line (EOL) test after manufacturing. This work presents a method to estimate the blurring kernel of automotive camera for state monitoring. To achieve this, Principal Component Analysis (PCA) is performed, using synthetic kernels generated by Zemax. The PCA model is built with approximately 1300 base kernels representing spatially variant point spread functions (PSFs). This model generates kernel samples during the estimation process. Synthetic images are created by convolving the synthetic kernels with reference traffic sign images and compared with real-life data captured by an automotive camera. These synthetic data are utilized for algorithm development, and later on validation is performed on real-life data. The algorithm extracts two 45 x 45 pixels regions of interest (ROIs) containing slanted edges from the blurred image and crops matching ROIs from a reference sharp image. Each candidate kernel blurs the reference ROIs, and the resulting Spatial Frequency Response (SFR) is compared with the blurred ROIs’ SFR. Differential evolution optimization minimizes the SFR difference, selecting the kernel that best matches the observed blur. The final kernel is evaluated against the true kernel for accuracy. Structural similarity index measure (SSIM) between the original and estimated blurred ROIs ranges from 0.808 to 0.945. For true vs. estimated kernels, SSIM varies from 0.92 to 0.98. Pearson correlation coefficients range from 0.84 to 0.99, Cosine similarity from 0.86 to 0.98, and mean squared error (MSE) from 1.1 x 10-5 to 8.3 x 10-5. Validation on real-life camera images shows that the SSIM between estimated ROI is 0.82 indicating a sufficient level of accuracy in kernel estimation to detect potential degradation of the camera. UR - https://doi.org/10.21203/rs.3.rs-6725582/v1 Y1 - 2025 UR - https://doi.org/10.21203/rs.3.rs-6725582/v1 UR - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:573-62419 SN - 2693-5015 PB - Research Square CY - Durham ER -