The space- and other high reliability markets are continuously driven towards an increased use of deeply integrated electronics. The increasing demand for complexity and functionality results in the use of large package components with a high number of I/Os. In order to allow the use of components with high pin counts up to 1752, complex high-density interconnect (HDI) printed circuit board (PCB) technology is required. Reconciling the use of multiple laser-drilled microvia levels in a stacked configuration with the reliability requirements for space is challenging when using heritage dielectric materials. The use of ceramic filled low CTE material allows the manufacturing of complex HDI PCB technology with a high reliability.

The work presented in this paper is part of an ongoing European (Horizon 2020) “COMAP-4S” project on components and macro components packaging for space. The Project is coordinated by SAFRAN ELECTRONICS AND DEFENSE with partners ACB, TUB and NANOXPLORE. The most complex PCB technology targeted within the project is four levels of microvias, requiring the use of low CTE laminate material. The reliability of different microvia configurations from all four levels staggered to all stacked was evaluated using test methods as described in ESA’s ECSS-Q-ST-70-60C standard for qualification and procurement of printed circuit boards. The test results of various material-level reliability tests, interconnection stress testing (IST) and reflow simulation combined with rework and traditional thermal cycling are provided to demonstrate a high reliability of the different via configurations and overall PCB technology.

Every layer interconnect (ELIC) printed wiring boards (PWB) were sourced after D-coupon evaluation per IPC-TM-650 2.6.27B. U.S. and off-shore D-coupons were tested to assess reflow survivability, a prerequisite that demonstrates reliable copper interconnects. Tested chains mirrored configurations designed into the actual 12-layer ELIC PWBs including 11-stacks. The structures, test results, and reflow simulation models are discussed. Evidence of the weak microvia interface was presented to the IPC Technical Activities Executive Committee early 2017. Resistance-temperature measurements recorded during convection reflow assembly identified stacked microvia open-reconnects. Reflow tested daisy chains were cross-sectioned, viewed by optical microscopy, scanning electron microscopes (SEM), and focused ion beams (FIB). Copper fractures were located primarily between the target pad and electrolytic copper-fill. Stacked microvias which fractured during reflow imitated mechanical switches. Opens triggered generally above 215C, remained open, and then reconnected when cooled below 215C. Fractured microvias deceptively appear to function normally, are undetectable at ambient temperatures by in-circuit-tests (ICT), conventional air-to-air (ATA) temperature shock or cycle (TS/TC), and therefore introduce a hidden reliability concern when deployed.

Fractured microvias are simply physical contacts, not metallurgical bonds. Unlike stacked structures, staggered configurations formed using the same process withstood the reflow thermomechanical strain. Layer-to-layer staggered microvia paths provided a reliable high-density interconnect (HDI) alternative to stacked structures as long as there was room to fit the configuration. This provided a temporary window, a brief opportunity, to solve the weak stack dilemma that, unfortunately, was squandered and is now closed. Testing to date suggested U.S. suppliers, in general, are not prepared to supply ELIC constructions with greater than 3-stack microvias, let alone 11-stack microvias, that consistently survive reflow. Stacked microvias with weak copper interfaces are subject to reflow-induced fracture which, therefore, precludes their use in mission-critical, high-hazard environments. Consequently, the U.S. supply base is at a technology disadvantage plagued with issues sourcing ITAR compliant HDI PWBs necessary to address circuit congestion set in motion by increased silicon content and smaller footprints.

This paper presents two key aspects of reliability of ball grid array (BGA) technologies: one aspect for high-reliability and the other for commercial applications. For high reliability, it presents the effect of isothermal aging at the BGA part level to address burn-in requirements for die screening per Mil standard specification. Pull and shear behavior of BGAs with tin-lead and SAC305 solder balls were characterized for a number of components before and after two burn-in conditions (125℃ and 150℃ for 240 hr). The burn-in and fresh BGAs were assembled onto PCBs and then subject to thermal cycling (TC) between –40℃ and 105℃ to determine cycles-to-failures (CTFs) for comparison. Pull and shear strength test results of tin-lead and SAC305 solder balls were presented for CABGA208, CTBGA228, and CVBGA360 with 0.8, 0.5, and 0.4 mm pitches. CTFs in Weibull plots were presented for the SnPb CABGA208 for fresh and burn-in conditions.

The commercial aspect covered low temperature solder ball and solder paste for assembling of BGAs and other components. We evaluated two types of low temperature BiSn alloys with Cu/Ni or Ag additives and with melting temperatures between 139˚C solidus and 174˚C liquidus. The mixed hybrid configuration utilized SAC305 BGAs using SnBiAg alloy solder paste. The low temperature assemblies were subjected to two TCs (–40℃/105℃ and 0℃/105℃). CTF results were presented for CABGA208 with failure analysis performed by X-sectional evaluation.

Increasingly larger surface mount components are being developed in today’s SMT industry. With increasing footprints, maintaining acceptable warpage levels through reflow and/or real-world use is a growing challenge. Undoubtedly, efforts are made to mitigate warpage in both PCB and components. However, there are limits to these mitigation effects and they do not resolve sample to sample variation. Here the question is posed, what if 100% of components and corresponding PCB attach areas are measured for flatness prior to assembly? Could pick-and-place machines selectively pick a “best” matching component from those available, to place on the next PCB that comes down the line?

This paper and corresponding study only lays the foundation to answer these posed questions. It is hypothesized that matching shapes at room temperature based on minimizing gap between attaching surfaces is not the optimal way to make PCB to component matching decisions. Instead, it is suggested that predicting what these shapes will be at critical points in the reflow/reliability profile is the more critical shape matching to consider. In this study, a sampling of matching footprints of PCBs and components are measured under reflow temperatures via common full-field optical metrology techniques. Critical assembly temperatures are analyzed looking to optimize which component should go with which PCB by analyzing all possible combinations through software automation. Hypothetically, this data can then correlate back to room temperature shape combinations for the best overall surface mount reliability.

Raytheon Missiles and Defense, formed a L-Lead (Pb) to LF-Lead-free focus group to understand the metallurgical and manufacturing challenges to transition to a “pure” Pb-free metallurgical system for SMT surface mount technology CCA-circuit card assembly. To better prepare for this transition, a number of studies were initiated to evaluate baseline and next generation lead-free solder pastes. One such study used a temperature programmable heated or “hot-stage” installed on a current X-ray machine for real-time radiographic analysis under a simulated SMT reflow thermal profile. Representative PCB / CCA samples were solder printed with baseline and candidate LF solder paste alloys with and without BTC- bottom terminated components (QFN or MLF 100) devices, then transferred to the X-ray hot stage, which was programmed to run under the appropriate SMT Reflow thermal profile. During the hot stage reflow process, the devices and solder pastes were X-ray-video recorded with void density measurements made after reflow. Real-time radiographic imaging of hidden solder pastes under MLF 100s revealed solder joint and voiding formation mechanisms that differed from each of the various solder pastes evaluated. Radiographic results also demonstrated that solder paste wetting behavior correlated to both air + process void formation mechanisms. Moreover, a strong statistical correlation was observed between low void density versus high shock performance of the same baseline and LF solder paste alloy candidates evaluated in another parallel mechanical testing study.

Mainstream manufacturing testing strategies involve structural tests including optical inspection, structural defect finding, such as opens, shorts, missing and catastrophically defective components. This strategy, however, still leaves a great deal of potential faults, especially if the unit under test (UUT) is operating at high speed. Typically, high-speed testing, if performed at all, is left for a later stage of assembly. For example, while the structural tests can be performed on a panel containing several circuit boards, functional testing at high-speed is done only after the boards are separated, powered up and low-speed functional tests have passed. Moving high-speed testing up to an earlier test stage can save the costs of functional tests later. In many cases, if the high-speed controller is found to be faulty, it can be replaced during the structural test stage. This paper investigates ways that manufacturing defect analyzers (MDAs), in-circuit testers (ICTs), functional board testers (FBTs), and system level test (ST) automatic test equipment (ATE) can be augmented by a high-speed bus tester (HSIO) to provide at-speed tests in parallel with structural test. The article will discuss how this capability can be integrated into existing manufacturing test stages and examine the economic benefits of such an approach. It will also demonstrate the economic benefits of bringing high-speed test into the board test rather than perform those tests as part of the system level testing.

In recent years, Neural-Network based deep learning models has demonstrated high accuracy in object detection and classification in digital image processing. Manufacturing industry has successfully implemented prototypes and small-scale deployments to employ AI models for quality inspection. Proven that AI-assisted quality inspection can improve inspection accuracy, operation throughput and efficiency significantly through those prototypes and small-scale deployments. In past two years, two papers “A Framework for Large-Scale AI-Assisted Quality Inspection Implementation in Manufacturing Using Edge Computing” and “A Study of AI Models Benchmarking for Quality Inspection Implementation in Manufacturing Using Edge Computing” were presented at IPC APEX, which discussed the challenges in large-scale deployment of AI models for quality inspection operation, and the IT architectural decisions to fulfill the OT requirement and inference performance requirement at the edge.

This paper continues the discussion on the operational challenges at the edge and deep dives into data archiving and CI/CD (Continuous Integration/Continuous Delivery). It also discusses the technical challenges to meet the security requirement at the edge. A framework for data archiving and Edge CI/CD implementation is presented.

Corrosion is the most dominant failure mode in electronic products. In many cases, the failure seed is corrosion contamination already on the soldering leads before the assembly that propagates over time and is accelerated by humidity, temperature, and acidity in the environment. The corrosion degrades the board to failures later in the production post-assembly testing, and during the product's life cycle.

We present a method for mass real-time early detection of corrosion contamination on electronic components during the mounting pick-and-place process. The method is based on the correlation between the light reflectance from the soldering leads during their placement photography and the extent of the corrosion. Corroded leads have significantly rougher surface and pitting spots than pristine leads. As a result, they reflect light differently. The difference in their appearance can be detected by AI forensic analysis of the component’s pictures. An AI model correlating the leads finish with their corrosion content and progression level is presented, and its performance on mass scale data is analyzed.

We further present a real-life study on how corroded components were detected during the pick-and-place process only to fail during the ICT testing. The post-failure SEM/EDS and cross-section analysis confirm the AI failure predictions on multiple components with corrosion during full-scale production.

The presented method is deployed on multiple production lines inspecting all components without affecting throughput while flagging contaminated components that are unsafe. The accuracy of prediction is over 99.5% tested on over 2.5 billion components.