Answers are grounded in this paper and cite their sources.
The continued evolution of semiconductor technology toward gate-all-around (GAA) architectures, vertically stacked memory, advanced BEOL interconnects, and heterogeneous integration has fundamentally transformed the role of Transmission Electron Microscopy (S/TEM) in device development and pathfinding use-cases. As critical dimensions approach the few-nanometer regime and device performance becomes increasingly sensitive to subtle variations in strain, chemistry, orientation, and electrostatic potential, conventional 2D-projection imaging and analysis alone is no longer sufficient. Successful semiconductor pathfinding applications demand integrated, quantitative, and depth-aware analytical techniques that combine structural, chemical, and crystallographic information within a unified metrology workflow. In this presentation, I will discuss how Thermo Scientific’s next-generation S/TEM platform—designed around multimodal integration and quantitative reproducibility—will help to transition from qualitative imaging towards engineering-grade, multi-dimensional semiconductor metrology. The emphasis is not only on improving existing techniques and exploring new ones, but also on coordinating the integration of routine aberration-corrected STEM imaging, high-efficiency STEM-EELS and STEM-EDS spectroscopies, and expanding 4D-STEM diffraction analysis (including center-of-mass analysis, strain mapping, and orientation mapping), through-focus iDPC imaging and emerging ptychographic and ePDF methodologies within unified acquisition and processing workflows. Chemical and bonding information remain foundational to semiconductor development. Over the last few years, significant advances in spectrometer and detector design, collection efficiency, and energy resolution have expanded the capabilities of STEM-EELS for device-relevant materials [1-4]. Developments associated with high-brightness cold field emission sources and optimized spectrometer optics with new functionalities and detection have enhanced signal-to-noise at low dose, enabling reliable analysis of ultra-thin dielectric stacks and beam-sensitive materials. Recent developments around Thermo Scientific’s Iliad EELS spectrometer and related high-stability mono-chromated systems have demonstrated improved detection of subtle bonding variations, opened access to the extreme-high energy loss world beyond 30keV, enhanced low-loss analysis for layer thickness and bandgap measurements, and improved robustness for routine semiconductor workflows [3,4]. These advances are particularly relevant for: Dopant profiling in nanosheet channels Valence state analysis in high-k and metal gate materials Barrier layer quantification in BEOL Interfacial chemistry in contact engineering When acquired using Electrostatic Beam Blanking and in full synchronization with high-solid-angle EDS, correlated EELS/EDS mapping significantly improves sample integrity and data reliability, while supporting quantitative interpretations across chemical and structural domains. The maturation of pixelated detectors and computational pipelines has transformed 4D-STEM from an exploratory technique into a robust semiconductor analysis tool [5-7]. By recording a full diffraction pattern at each probe position, quantitative extraction of crystallographic orientation, phase distribution, and full strain tensors becomes possible with nanometer-scale resolution. For GAA logic devices, 4D-STEM enables direct measurement of strain variations within stacked nanosheets, revealing heterogeneities that impact carrier mobility. In 3D NAND, vertical strain accumulation and lattice distortion can be mapped layer-by-layer. In BEOL interconnects, grain size distributions, twin density, and texture mapping correlate with electromigration reliability. Importantly, distortion-controlled STEM imaging and calibrated diffraction analysis provide the foundation for quantitative analysis across large fields of view while maintaining nanometer level spatial resolution. The integration of acquisition and processing software into unified workflows reduces reliance on ad-hoc scripting and enhances cross-site reproducibility—an increasingly critical requirement for semiconductor engineering organizations. Integrated differential phase contrast (iDPC) imaging has been confirmed as a powerful technique for linear electrostatic potential imaging with strong light-element sensitivity [8,9]. Compared to conventional annular dark-field contrast, iDPC improves interpretability in low-Z materials and thin dielectric stacks, and when combined with drift-controlled through-focus acquisition strategies, iDPC provides enhanced depth discrimination, supporting 3D analysis of stacked nanosheets, buried interfaces, and dielectric conformity down to 2-3nm scale. While not replacing full tomographic reconstruction and complex Multi-slice Electron Ptychography (MEP) reconstructions, through-focus iDPC offers quick access to a low-dose, low-complexity pathway toward depth-aware semiconductor metrology—particularly valuable for beam-sensitive device layers and fast time-to-data. Alternatively, MEP has made significant progress over the last couple of years, with improvements in detector performance, phase retrieval algorithms and computational efficiency [10,11]. For semiconductor applications, MEP offers improved transfer of low spatial frequencies and enhanced electrostatic potential mapping, with impressive demonstrations of three-dimensional reconstruction capability under appropriate conditions [12]. Similarly, electron pair distribution function (ePDF) and radial distribution function (RDF) analyses have matured as tools for characterizing short-range order in amorphous or partially ordered materials [13]. As disordered oxides, low-k dielectrics, and complex interlayers become more prevalent at advanced technology nodes, ePDF can provide complementary insight beyond conventional Bragg diffraction. From a practical standpoint, the most significant developments in semiconductor S/TEM over the last decade have not been any single technique, but the integration of multiple analytical modalities into calibrated, reproducible workflows. Semiconductor pathfinding and advanced failure analysis demand: - Correlation of structure, chemistry, strain, and orientation - Dose-optimized operation for sensitive materials - Quantitative calibration and distortion control - Software environments capable of managing large multimodal datasets By combining aberration-corrected imaging, high-efficiency spectroscopies, 4D-STEM analysis, through-focus iDPC, and emerging computational methods within coherent acquisition frameworks, S/TEM evolves from a projection-based visualization tool toward an integrated 3D metrology platform. As semiconductor architectures continue to scale and diversify, the role of S/TEM must similarly progress, supporting not only visualization at the atomic scale but also quantitative, reproducible, and depth-aware characterization aligned with engineering decision-making. The transition from isolated techniques to integrated multimodal analysis represents the defining direction for advanced Semiconductor Transmission Electron Microscopy in the coming decade.
No takes yet. Share an insight, caveat, or question.
Delille et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: