Lead-free tin-based solder joints often have a single-grained structure with random orientation and highly anisotropic properties. These alloys are typically stiffer than lead-based solders, hence transfer more stress to printed circuit boards (PCBs) during thermal cycling. This may lead to cracking of the PCB laminate close to the solder joints, which could increase the PCB flexibility, alleviate strain on the solder joints, and thereby enhance the solder fatigue life. If this happens during accelerated thermal cycling it may result in overestimating the lifetime of solder joints in field conditions. In this study, the grain structure of SAC305 solder joints connecting ceramic resistors to PCBs was studied using polarized light microscopy and was found to be mostly single-grained. After thermal cycling, cracks were observed in the PCB under the solder joints. These cracks were likely formed at the early stages of thermal cycling prior to damage initiation in the solder. A finite element model incorporating temperature-dependant anisotropic thermal and mechanical properties of single-grained solder joints is developed to study these observations in detail. The model is able to predict the location of damage initiation in the PCB and the solder joints of ceramic resistors with reasonable accuracy. It also shows that the PCB cracks of even very small lengths may significantly reduce accumulated creep strain and creep work in the solder joints. The proposed model is also able to evaluate the influence of solder anisotropy on damage evolution in the neighbouring (opposite) solder joints of a ceramic resistor.

As the automotive industry shifts towards the electrification of drive trains, the efficiency of power electronics becomes more important. The use of silicon carbide (SiC) devices in power electronics has shown several benefits in efficiency, blocking voltage and high temperature operation. In addition, the ability of SiC to operate at higher frequencies due to lower switching losses can result in reduced weight and volume of the system, which also are important factors in vehicles. However, the reliability of packaged SiC devices is not yet fully assessed. Previous work has predicted that the different material properties of SiC compared to Si could have a large influence on the failure mechanisms and reliability. For example, the much higher elastic modulus of SiC compared to Si could increase strain on neighboring materials during power cycling. In this work, the failure mechanisms of packaged Si- and SiC-based power devices have been investigated following power cycling tests. The packaged devices were actively cycled in 4.5 s heating and 20 s cooling at ΔT = 60 - 80 K. A failure analysis using micro-focus X-ray and scanning acoustic microscopy (SAM) was carried out in order to determine the most important failure mechanisms. The results of the analysis indicate that the dominant failure mechanism is wire bond liftoff at the device chip for all of the SiC-based devices. Further analysis is required to determine the exact failure mechanisms of the analyzed Si-based devices. In addition, the SiC-based devices failed before the Si-based devices, which could be a result of the different properties of the SiC material.

The increasing complexity of electronics in systems used in safety critical applications, such as for example self-driving vehicles requires new methods to assure the hardware reliability of the electronic assemblies. Prognostics and Health Management (PHM) that uses a combination of data-driven and Physics-of-Failure models is a promising approach to avoid unexpected failures in the field. However, to enable PHM based partly on Physics-of-Failure models, sensor data that measures the relevant environment loads to which the electronics is subjected during its mission life are required. In this work, the feasibility to manufacture and use integrated sensors in the inner layers of a printed circuit board (PCB) as mission load indicators measuring impacts and vibrations has been investigated. A four-layered PCB was designed in which piezoelectric sensors based on polyvinylidenefluoride-co-trifluoroethylene (PVDF-TrFE) were printed on one of the laminate layers before the lamination process. Manufacturing of the PCB was followed by the assembly of components consisting of BGAs and QFN packages in a standard production reflow soldering process. Tests to ensure that the functionality of the sensor material was unaffected by the soldering process were performed. Results showed a yield of approximately 30 % of the sensors after the reflow soldering process. The yield was also dependent on sensor placement and possibly shape. Optimization of the sensor design and placement is expected to bring the yield to 50 % or better. The sensors responded as expected to impact tests. Delamination areas were present in the test PCBs, which requires further investigation. The delamination does not seem to be due to the presence of embedded sensors alone but rather the result of a combination of several factors. The conclusion of this work is that it is feasible to embed piezoelectric sensors in the layers of a PCB.

The increasing complexity of electronics in systems used in safety critical applications, such as self-driving vehicles, requires new methods to assure the hardware reliability of the electronic assemblies. Prognostics and health management (PHM) that uses a combination of data-driven and physics-of-failure models is a promising approach to avoid unexpected failures in the field. However, to enable PHM based partly on physics-of-failure models, sensor data that measure the relevant environment loads to which the electronics are subjected during its mission life are required. In this work, the feasibility to manufacture and use integrated sensors in the inner layers of a printed circuit board (PCB) as mission load indicators measuring impacts and vibrations has been investigated. A four-layered PCB was designed in which piezoelectric sensors based on polyvinylidenefluoride-co-trifluoroethylene (PVDF-TrFE) were printed on one of the laminate layers before the lamination process. Manufacturing of the PCB was followed by the assembly of components consisting of ball grid arrays (BGAs) and quad flat no-leads (QFN) packages in a standard production reflow soldering process. Tests to ensure that the functionality of the sensor material was unaffected by the soldering process were performed. Results showed a yield of approximately 30% of the sensors after the reflow soldering process. The yield was also dependent on sensor placement and possibly shape. Optimization of the sensor design and placement is expected to bring the yield to 50% or better. The sensors responded as expected to impact tests. Delamination areas were present in the test PCBs, which requires further investigation. The delamination does not seem to be due to the presence of embedded sensors alone but rather the result of a combination of several factors. The conclusion of this work is that it is feasible to embed piezoelectric sensors in the layers of a PCB.

Recording and prediction of the accumulated damage, which will eventually lead to the failure of power electronic modules, is an aspect of high importance for power electronic systems design and, in particular, for development of Prognostic and Health Management (PHM) schemes for in-field applications. To this end, this paper presents a simple and cost-effective prognostic method for predicting the remaining useful life (RUL) of TO-247 packaged silicon carbide (SiC) metal-oxide semiconductor field-effect transistors (MOSFETs) subjected to power cycling experiments. The model assumes that the major failure mode is bond-wire lift-off and uses a damage accumulation scheme based on Paris’ crack law. The only inputs to the model are historical data on the average junction temperature swing and the temperaturecompensated drain-source ON-state resistance at the peak temperature of the current cycle. Using only these two input values, the model is shown to predict RUL with surprising accuracy for the range of constant current loads determining cycling conditions under which the test data series have been acquired. This work is a first step in an ongoing project towards building more elaborate prognostic schemes for RUL-determination of SiC power MOSFETs in actual working conditions, using physics-informed neural networks (PINNs).

Lifetime predictions of lead-free solder joints remain a challenging task in assuring reliability of electronic packaging. Due to the complex interaction of a wide range of as-soldered microstructures, strong an isotropic properties and a failure mechanism that is not completely understood, life time predictions of SnAgCu solder joints remain a challenging task.In this work, we present experimental findings of transgranular crack propagation of SnAgCu solder joints in thermal cycling. These findings are in contrast with the established picture of a recrystallization-assisted intergranular crack propagation failure mechanism

Lead-free solder joints have been shown to increase the risk for crack formation in the PCB laminate under the solder pads. As such cracks propagate during thermal cycling, they decrease the strain imposed on the solderjoint by acting as strain relief. In accelerated thermal cycling, these joints have been found to remain virtually undamaged even after a very high number of cycles. If these cracks do not form or propagate to the same extent under milder cycling conditions, typical of service conditions, it may lead to an overestimation of the fatigue life of the solder joints in accelerated testing. In this work, the extent of strain relief and the influence of grain orientations on the initiation and propagation of these cracks are investigated through FE-modelling and compared to what has been experimentally observed for cross-sections of solder joints moulded in epoxy resin with added fluorescent agent and inspected using UV-light and electron backscatter diffraction. Due to the strong anisotropy of lead-free solder joints, the stress transferred to the laminate will vary significantly depending on grain orientation. The presence of these laminate cracks adds another layer of uncertainty to the already complex SnAgCu system, where the strong effects of anisotropy, the continuously evolving secondary precipitate coarsening and its interaction with the recrystallisation process govern the damage evolution. If these effects are not properly accounted for, the interpretation of thermal cycling or modelling and simulation results may be strongly misleading.

Canonical deep learning-based Remaining Useful Life (RUL) prediction relies on supervised learning methods which in turn requires large data sets of run-to-failure data to ensure model performance. In a large class of cases, run-to-failure data is difficult to collect in practice as it may be expensive and unsafe to operate assets until failure. As such, there is a need to leverage data that are not run-to-failure but may still contain some measurable, and thus learnable, degradation signal. In this paper, we propose utilizing self-supervised learning as a pretraining step to learn representations of the data which will enable efficient training on the downstream task of RUL prediction. The self-supervised learning task chosen is time series sequence ordering, a task that involves constructing tuples each consisting of $n$ sequences sampled from the time series and reordered with some probability $p$. Subsequently, a classifier is trained on the resulting binary classification task; distinguishing between correctly ordered and shuffled tuples. The classifier's weights are then transferred to the RUL-model and fine-tuned using run-to-failure data. We show that the proposed self-supervised learning scheme can retain performance when training on a fraction of the full data set. In addition, we show indications that self-supervised learning as a pretraining step can enhance the performance of the model even when training on the full run-to-failure data set. To conduct our experiments, we use a data set of simulated run-to-failure turbofan jet engines.

Robust and accurate prognostics models for estimation of remaining useful life (RUL) are becoming an increasingly important aspect of research in reliability and safety in modern electronic components and systems. In this work, a data driven approach to the prognostics problem is presented. In particular, machine learning models are trained to predict the RUL of wire-bonded silicon carbide (SiC) metal-oxide-semiconductor field-effect transistors (MOSFETs) subjected to power cycling until failure. During such power cycling, ON-state voltage and various temperature measurements are continuously collected. As the data set contains full run-to-failure trajectories, the issue of estimating RUL is naturally formulated in terms of supervised learning. Three neural network architectures were trained, evaluated, and compared on the RUL problem: a temporal convolutional neural network (TCN), a long short-term memory neural network (LSTM) and a convolutional gated recurrent neural network (Conv-GRU). While the results show that all networks perform well on held out testing data if the testing samples are of similar aging acceleration as the samples in the training data set, performance on out-of-distribution data is significantly lower. To this end, we discuss potential research directions to improve model performance in such scenarios.