Semiconductor Design-for-Reliability (DfR) faces a critical knowledge bottleneck. Experimental parameters, failure criteria, and mechanism interpretations remain scattered across publications in inconsistent formats. Such fragmentation hinders systematic reuse and cross-study comparison. To address this challenge, we present an automated framework that extracts reliability data from unstructured technical documents and transforms it into standardized, auditable knowledge for experiment planning. Coordinated AI modules perform data extraction, lifetime and degradation modeling, physical constraint validation, and test condition recommendation. All intermediate results are stored in a shared database for traceability. Local deployment of all AI models ensures data confidentiality suitable for industrial use without reliance on external cloud services. Evaluation on a corpus of 50 IGBT power-cycling publications showed that the extraction module achieved an F1 score of 0.83, which outperformed rule-based methods by 15 percentage points. The largest gains were observed for semantically variable parameters such as failure criteria and Weibull coefficients. For experiment design recommendations, alignment with published test matrices reached up to 62.5%, compared to 0-25% for general-purpose AI models of comparable or greater scale. These results demonstrate that domain-structured knowledge integration, rather than AI model size alone, drives the performance improvement and offers a practical pathway for auditable, privacy-preserving automation in reliability engineering workflows.

The reliability of through-glass via (TGV) interconnects is critical for advanced semiconductor packaging. This work investigates microstructural and mechanical evolution in electroplated TGV–Cu subjected to long-term aging at 250 °C. TGV samples were fabricated via laser-induced etching and double-sided copper electroplating, then aged for up to 1008 h. Nanoindentation revealed region-dependent reductions in hardness (from 2.0–2.5 GPa to below 0.5 GPa) and modulus (from 110–130 GPa to 40–90 GPa), with surface-near regions most affected. The glass substrate maintained stable mechanical properties until microcracks formed after 1008 h. EBSD quantification showed grain-size enlargement from 0.46 µm to 1.86 µm and a concurrent decrease in dislocation density. Molecular dynamics simulations of 3, 4, 5 nm grains corroborated the inverse relationship between grain size and micro-mechanical properties. A hybrid Potts-phase field model further linked grain coarsening to stress relaxation and elastic-energy minimization, revealing that as grains grow, the overall von Mises stress in the structure decreases; high-modulus grains retain relatively higher local stresses, while low-modulus, low-stress grains exhibit faster growth rates. Electrical I–V measurements confirmed stable ohmic behavior, despite a drop in insulation resistance. These integrated experimental and computational insights provide theoretical guidance for optimizing TGV interposer design and ensuring long-term operational reliability in heterogeneous integration technologies. (Figure presented.)

Reliable 4H-SiC for high-power electronics and quantum photonics requires a quantitative understanding of how contact loading drives microstructure evolution and load-bearing/fracture response in epitaxial layers. Here, we integrate instrumented indentation, confocal micro-Raman residual-stress metrology, atomistic molecular dynamics (MD), and high-resolution TEM (HRTEM) to establish processing–microstructure–mechanical property linkages in chemical vapor deposition (CVD) 4H-SiC epilayers. At peak depths of 600–1050 nm, indentation promotes Palmqvist-type radial cracks and the apparent indentation toughness K_IC increases from 0.87 ± 0.08 to 1.20 ± 0.05 MPa m^1/2 with depth, consistent with plastic-zone growth and dislocation shielding. E₂(TO) Raman mapping quantifies an increase in residual stress from ∼302 ± 60 to ∼665 ± 72 MPa. It also shows that the incremental broadening of the FWHM becomes less pronounced beyond ∼750 nm, suggesting that the near-surface disorder indicator within the Raman probe volume approaches a quasi-steady level. MD captures a 4H → 3C phase transformation, amorphization beneath indenter ridges, and dislocation nucleation/growth, which HRTEM directly corroborates. The combined measurement–model–validation closed loop yields a depth-dependent relationship between residual-stress accumulation and apparent toughness, converting them into an actionable processing window: constraining penetration depth below ∼0.75 μm limits residual stress and near-surface disorder. These results provide physics-based guidance for machining and packaging of 4H-SiC epilayers and illustrate a transferable framework for brittle, anisotropic ceramics.

This paper investigates the surge reliability of commercial 1200V SiC MOSFETs through a combined approach of experimental testing and multiphysics simulation, elucidating the failure mechanisms under both step and repetitive surge current stress. The innovative integration of package-level electro-thermal coupling simulation with transient junction temperature estimation overcomes the limitations of conventional methodologies that rely solely on decapsulation analysis and Technology Computer Aided Design (TCAD) simulation. Experimental evaluations on six Devices Under Test (DUTs) with distinct structural configurations, employing surge testing and failure analysis techniques including C-mode Scanning Acoustic Microscopy (C-SAM), optical microscopy, and Scanning Electron Microscope (SEM), confirm that device failure primarily originates from gate-source short circuits caused by aluminum bonding wire melting. COMSOL multiphysics simulations further replicate the transient thermal characteristics of bonding wire regions, demonstrating rapid temperature escalation to the Aluminum melting point within 5-6 ms during surge events. A transient thermal resistance-based junction temperature characterization method is proposed, revealing an inverse proportionality between thermal resistance and chip area.

This study is motivated by a conceptual inconsistency in the physical interpretation of eight-chain hyperelastic theory, which arises from the combined effect of two distinct issues: the use of the marginal projection distribution p_z(|r_z|) as a surrogate for the full probability density of end-to-end distance p_r̄(r̄), and the subsequent reliance on a root mean square (RMS) approximation step in the micro–macro averaging of chain stretch. We first revisit this probabilistic mismatch by reformulating the probability density function of freely-jointed chains (FJCs) in terms of the squared end-to-end vector r², thereby restoring consistency on chain-level statistics. Building on this formulation, the micro–macro mapping averaging of chain conformational free energy is constructed directly in terms of r², leading to a one-step mean-field approximation that avoids RMS averaging. The modified probability transformation is examined by Monte Carlo sampling at the microscopic level. To account for interchain interactions, q-mean statistical description of micro tube confinement was incorporated, leading to the appearance of the general invariant I_q=λ₁^q+λ₂^q+λ₃^q. The resulting continuum constitutive model is assessed against multiaxial experimental data for several polymer networks, including vulcanized natural rubber, Entec Enflex S4035A thermoplastic elastomer, Tetra-PEG, and isoprene rubber vulcanizate. Comparisons with three existing hyperelastic strain energy formulations, the extended eight-chain, extended tube models, and the four-parameter ”comprehensive” model, demonstrate comparable phenomenological accuracy of the current model while providing a clearer and more consistent micro–macro physical interpretation of model parameters. A parametric study further illustrates how the dimensionless parameters n and q govern the shape of the macroscopic stress–strain responses. The present formulation provides a consistent theoretical basis within the scope of hyperelasticity and admits potential extensions toward more complex irreversible phenomena.

Sintered Cu nanoparticles (Cu NPs) are promising interconnection materials for high-temperature power electronics, yet how their authentic three-dimensional pore architecture governs microscale deformation remains unclear. Here, synchrotron nano-computed tomography (nano-CT) was combined with in-situ micropillar compression, explicit dynamic elastoplastic finite element analysis, and TEM/TKD characterization to interrogate sintered Cu NPs. The nano-CT voxel size was 45 nm, and the reconstructed volume corresponded to a cylinder 16 µm in diameter and 10 µm in height. The average sectional porosity was 12.44%, with a systematic discrepancy between two-dimensional and three-dimensional porosity quantification. During loading, the porosity decreased to 9.55% while the pore aspect ratio increased from 1.82–2.35. Finite element analysis further showed pronounced pore-adjacent stress/strain localization at the elastic–plastic transition, with local stress and equivalent plastic strain reaching 650 MPa and 1.7 × 10⁻², compared with 250 MPa and 1.1 × 10⁻³ in adjacent regions. The GND density increased by 95.9% at a compressive strain of 26%, linking pore-induced strain gradients to dislocation accumulation. These results quantitatively connect authentic three-dimensional pore architecture, local deformation localization, and dislocation-mediated strengthening in sintered Cu NPs. Highlights Synchrotron nano-CT (45 nm voxel size) reconstructed a 16 × 10 µm cylindrical volume of sintered Cu NPs and resolved the authentic 3D pore network. Sectional porosity was 12.44%, and 2D/3D quantification showed a systematic discrepancy, with porosity decreasing to 9.55% and pore aspect ratio increasing from 1.82 to 2.35 during compression. Pore-adjacent localization was quantified at the elastic–plastic transition, with local stress/PEEQ reaching 650 MPa and 1.7 × 10⁻² versus 250 MPa and 1.1 × 10⁻³ in adjacent regions. A 95.9% increase in GND density at 26% compressive strain links pore-induced strain gradients to dislocation accumulation and strain-gradient-driven strengthening.

Serving as a significant attachment component, Active Metal Brazed (AMB) substrates are prone to oxidation during storage and processing. The copper surface oxidation of AMB substrates seriously affects the reliability of high-power modules. The formation of Cu2O/CuO oxide layer on the surface degrades interfacial qualities in subsequent packaging processes, including sintering, wire-bonding, and transfer-molding. To investigate the mechanism behind, this paper constructs a physically grounded multiscale oxidation framework that explicitly bridges molecular dynamics with reactive force field (ReaxFF-MD) and mesoscale continuum modeling. For ReaxFF-MD simulations, it aims to resolve oxygen adsorption, dissociation, diffusion behavior, and Cu-O network formation on the Cu(100), Cu(110), and Cu(111) surfaces at temperatures of 300 K and 600 K. Diffusion coefficients and interface reaction kinetic parameters are then quantitatively extracted from MD simulations and transferred as a bridge in reaction-diffusion continuum model with flux damping governed by a characteristic structural thickness. The results demonstrate that the early oxidation process of the copper surface is modulated by structural evolution and orientation anisotropy. The proposed multiscale framework provides a physical mechanism basis for understanding the early oxidation mechanism of copper and its impact on the reliability of AMB packaging.

Ion implantation and subsequent annealing reshape the defect landscape and stress state of compound semiconductors, yet the temperature-dependent mechanisms in SiC remain incompletely understood. Here, we utilize molecular dynamics (MD) simulations and confocal micro-Raman measurements to resolve how implantation temperature and post-annealing regulate lattice disorder, amorphization kinetics, and residual-stress evolution in chemical vapor deposited (CVD) 4H-SiC. MD reveals surface-nucleated amorphization that propagates inward, whereas elevated implantation temperatures activate defect recombination pathways that suppress amorphous-layer formation. Raman signatures of optical-phonon shifts, linewidth broadening, and amorphization bands track the coupled evolution of lattice disorder and stress. Experimentally, increasing implantation temperature smooths the surface (S_a 0.133 → 0.101 nm) and reduces the amorphous-layer thickness (from ∼700 nm at 25°C to undetectable at 500°C), while driving more compressive residual stress (−57 → −132 MPa). Post-annealing largely restores phonon lifetimes and eliminates amorphization signatures, consistent with the recovery trends predicted by MD. These results delineate a thermal-treatment window that controls amorphization and residual stress in 4H-SiC, providing a transferable Raman-based methodology for nondestructive assessment of implantation-induced damage in compound semiconductors.

Pressureless sintered Ag pastes are promising die-attach materials for power electronics, yet practical sintering-profile optimization still relies heavily on trial-and-error, and the link from thermal kinetics to fracture-relevant microstructure and strength remains insufficiently established. In this work, a thermal-kinetics-guided workflow was developed to design paste-specific pressureless sintering profiles for two commercial Ag pastes (a spherical-particle paste and a flake-based paste) and to interpret the resulting strength–fracture response using SEM-based, heterogeneity-aware learning. Multi-heating-rate TGA was used to define the conversion fraction (α), while DSC was used to identify the dominant thermal-event window and extract the characteristic peak temperature for Arrhenius pairing. The TGA-defined conversion evolution was then modeled using a JMAK/Avrami form with Arrhenius temperature dependence to predict the isothermal holding time required to reach a target conversion at a selected dwell temperature. Relative to supplier-recommended profiles, the optimized profiles increased die-shear strength by 35% for the spherical-particle paste and by 206% for the flake-based paste, and promoted a fracture-mode transition from interfacial debonding toward mixed/cohesive fracture. Pearson/Ridge baselines and attention-based multiple-instance learning (MIL) linked strength and fracture-mode distribution to porosity/connectivity-related descriptors; paste-wise normalization mitigated paste-specific baselines and enabled MIL to reveal profile-induced microstructure–performance co-variation. Overall, this study establishes a practical workflow that couples thermal-kinetics-guided profile design with mechanical and fractographic validation and interpretable microstructure–property attribution, supporting mechanism-informed optimization of pressureless sintered Ag die-attach pastes.

Aiming at the reliability bottlenecks of power MOSFETs in such key application fields as new energy vehicles (NEVs), photovoltaic (PV) inverters, and industrial motor drives, this paper proposes an unsupervised anomaly detection method for power MOSFET degradation based on the Long Short-Term Memory Autoencoder (LSTM-AE). Without relying on labeled data or complex hardware modifications, the method can adaptively learn the health patterns and temporal features of devices, addressing the limitations of traditional methods such as difficulty in capturing long-term dependencies and poor noise immunity. Three different test conditions for power cycling tests were designed to collect operational data of power MOSFET devices. On-state resistance (RDS(on)) and body diode voltage drop(VSD) were selected as input features, and the Leave-One-Out Cross-Validation (LOOCV) method was adopted to verify the model performance. Experimental results demonstrate that the method can effectively capture weak degradation features during the stage of insignificant device performance changes, achieving an early warning prior to device failure. Under 3σ principle threshold, its detection performance outperforms the Multilayer Perceptron Autoencoder (MLP-AE) and traditional Mahalanobis Distance (MD) methods, with the optimal sequence size being 64. It achieves a high accuracy in identifying normal states and exhibits high recognition capability for severely degraded states, providing a robust, efficient, and practical solution for the health monitoring of power MOSFETs.

Accurate lifetime prediction of power semiconductor devices is critical for the reliability of power electronic systems. Power cycling tests generate only a small number of failure samples due to their high cost and long duration. This data scarcity makes it difficult to train reliable prediction models. At the same time, modern deep learning models require significant AI expertise to configure and tune, which creates a barrier for reliability engineers. To address these two problems, this paper proposes a domain-constrained multi-agent framework for automated remaining useful life (RUL) prediction of power semiconductor devices. The framework integrates three collaborative agents: a Stats Agent for statistical reliability assessment, a Physics Agent for physical plausibility validation, and a Model Agent for algorithm selection and hyperparameter tuning. A Central Controller manages the workflow and coordinates agent communication. The agents interact through a Propose-Critique-Refine (PCR) mechanism, in which the Model Agent proposes configurations and the other two agents critique them from statistical and physical perspectives. This iterative negotiation eliminates the need for manual parameter tuning. The framework is validated on the NASA IGBT accelerated aging dataset under Leave-One-Device-Out Cross-Validation. Results show that the framework achieves an R2 of 0.914 at a 50% observation ratio and remains above 0.79 even at 30%, demonstrating strong generalization under limited data conditions.

Two-dimensional materials (2DMs)-based devices exhibit aerospace potential due to their superior properties. However, the operational reliability of 2DMs-based devices in space environments is significantly influenced by charged-particle radiation, necessitating rigorous ground-based radiation tolerance assessments. Current research on radiation effects in 2DMs is primarily experimental, yet such methodologies are inherently time-consuming, resource-intensive, and limited in throughput. To address these challenges, computational modeling and simulation techniques are increasingly being integrated with experimental characterization to accelerate materials design and unravel underlying physical mechanisms. This review systematically evaluates the state-of-the-art multiscale computational frameworks for 2DMs research, focusing on recent advancements, technical challenges, and emerging opportunities. A novel integrative approach is proposed, combining density functional theory, molecular dynamics, Monte Carlo, finite element analysis, and machine learning techniques. Particular emphasis is placed on addressing challenges in multiscale modeling, including accurate representation of complex phenomena across spatial and temporal scales under extreme environmental conditions. Conversely, opportunities for enhancing predictive capabilities are highlighted, with implications for expediting materials discovery in electronics, photonics, energy storage, catalysis, and nanomechanical systems. This comprehensive survey provides a strategic roadmap for future research directions in multiscale computational modeling of 2DMs, emphasizing interdisciplinary methodologies that bridge atomistic simulations with macroscale engineering applications. The insights presented herein aim to advance the development of radiation-hardened 2DMs-based devices for next-generation aerospace systems.

As 2.5D/3D advanced packaging moves toward higher interconnect density, glass substrates are attractive for their electrical performance and dimensional stability. However, high aspect ratio, fine pitch through-glass vias (TGVs) are prone to thermo-mechanical failures under long-term thermal shock, and microscale reliability assessment is hindered by costly microstructural characterization and the lack of models capturing long-term microstructural evolution and property degradation. Here, fully copper-filled TGV-Cu structures were fabricated on Schott AF32 glass and subjected to thermal-shock cycling from - 40 ℃ to 125 ℃ (0, 700, 1400, and 2100 cycles). Results show a non-monotonic, location-dependent hardness evolution with pronounced heterogeneity, where the mid-via region exhibits stronger late-stage softening by nanoindentation. EBSD shows grain diameter increases from 0.61 μm to 0.91 μm, while GND density decreases by recovery and then re-accumulates preferentially near the via bottom, evidencing competition between grain-growth softening and dislocation hardening. Hybrid Potts-phase field simulations further predict continuous grain growth and stress relaxation (declining von Mises stress) inside the via; including the glass substrate and interface constraints raises the stress level but preserves the relaxation trend. Moreover, the phase field simulations results of equivalent plastic strain (PEEQ) reveal progressively intensified cyclic plastic localization near the top/bottom caps (Cu/glass intersections), providing mechanistic support for cracking localized at the via mouths.

The mechanical strength of sintered nanoparticles (NPs) limits their application in advanced electronics packaging. In this study, we explore the anisotropy in the microstructure and mechanical properties of sintered copper (Cu) NPs by combining experimental techniques with molecular dynamics (MD) simulations. We establish a clear relationship between processing conditions, microstructural evolution, and resulting properties in pressure-assisted sintering of Cu NPs. Our findings reveal that pressure-assisted sintering induces significant anisotropy in the microstructure, as evidenced by variations in areal relative density and the orientation distribution of necks formed during sintering. Specifically, along the direction of applied pressure, the microstructure exhibits reduced variation in areal relative density and a higher prevalence of necks with favorable orientations. The resulting anisotropic mechanical properties, with significantly higher strength along the pressure direction compared to other directions, are demonstrated through micro-cantilever bending tests and tensile simulations. This anisotropy is further explained by the combined effects of strain localization (influenced by areal relative density) and the failure modes of necks (determined by their orientation relative to the loading direction). This work provides valuable insights into the analysis of sintered NPs microstructures and offers guidance for optimizing the sintering process.

This study investigates the interface strength and fracture behavior of sintered copper (Cu) nanoparticles (NPs) for all-Cu integration in advanced microelectronics packaging. Micro-cantilever bending tests on three configurations (Cu NP-notched, interface-notched and un-notched micro-cantilevers) were analyzed using scanning electron microscopy (SEM), transmission electron microscopy (TEM), transmission Kikuchi diffraction (TKD) and cohesive zone model (CZM). The interface-notched micro-cantilevers demonstrate superior fracture resistance, with a stress intensity factor (K_Q) of 2.88±0.10 MPa m^1/2, compared to 2.12±0.11 MPa m^1/2 for Cu NP-notched micro-cantilevers. Simulation results, consistent with experimental results, reveal that Cu NP-notched micro-cantilevers exhibit lower fracture resistance due to porosity and stress concentrations, while interface-notched micro-cantilevers show enhanced strength, attributed to robust bonding and reduced void distribution. Un-notched micro-cantilevers display superior load-bearing capacity, with cracks bypassing the interface and propagating through porous regions. Moreover, in un-notched micro-cantilevers, a synergistic deformation mechanism is observed, where crack propagation through the sintered Cu NPs coexists with plastic slip deformation in the Cu substrate. These findings highlight the strong interfacial bonding and effective stress transfer at the Cu substrate-sintered Cu NP interface, validating the feasibility of direct sintering using Cu NPs without additional coatings.