Analysis of As, P dopant distribution of NMOS transistor by FESTEM & EDS

Importance of the topic

The precise mapping of dopant atoms such as arsenic and phosphorus in NMOS transistors is vital for optimizing device performance and ensuring yield in semiconductor manufacturing. Low dopant concentrations and nanoscale features present significant analytical challenges. Advanced techniques that deliver high spatial resolution, sensitivity, and rapid feedback are critical for next-generation device development and process control.

Objectives and overview

This study aims to demonstrate the combined use of field emission scanning transmission electron microscopy (FESTEM) and energy dispersive X-ray spectroscopy (EDS) for quantitative dopant distribution analysis in a modern NMOS transistor. A further objective is to evaluate the COMPASS multivariate statistical analysis tool for automated phase separation and enhanced elemental mapping.

Methodology and instrumentation

High-angle annular dark field (HAADF) STEM imaging was paired with large-solid-angle silicon drift detector (SDD) EDS mapping. Key acquisition parameters included:

• Electron microscope: JEOL JEM-2800 FESTEM at 200 kV
• EDS detector: JEOL SDD, 100 mm2 active area, 0.95 sr collection angle
• Mapping resolution: 256 × 256 pixels at 4.8 million× magnification
• Acquisition time: ~51 minutes per map, count rate ~13 200 cps, dead time ~22 %

Main results and discussion

1. Conventional count-based maps show overlaps between dopant peaks (P K overlaps Pt M; As K overlaps Hf L), leading to ambiguous distribution.
2. Quantitative peak deconvolution (“Quant mapping”) removes overlapping contributions and reveals clear P and As distributions near device gate regions and defects.
3. Multivariate statistical analysis via the COMPASS tool extracts principal component spectra and creates phase maps automatically in seconds. This separates undoped silicon, P-doped Si, As-doped Si, and multiple secondary phases (e.g. Ni-Si-Pt, Al-Ti-N-Ar) with improved signal-to-noise, enabling trace-level identification.
4. Live-mapping capability allows real-time termination of acquisition once all relevant phases are identified, optimizing microscope use and operator efficiency.

Benefits and practical applications

• High‐accuracy dopant profiling in critical transistor regions enhances process control and yield.
• Automated phase separation reduces manual interpretation time and increases reproducibility.
• Improved sensitivity to trace elements supports analysis of shrinking device geometries.
• Real‐time mapping feedback streamlines research and quality control workflows.

Future trends and applications

Advances in detector technology and multivariate algorithms will further reduce acquisition times and increase mapping resolution. Integration with in situ experimentation and machine learning‐driven analysis pipelines is expected to enable adaptive data acquisition and predictive defect diagnostics. The methodology is extendable to emerging device architectures, novel materials, and 3D integration schemes.

Conclusion

The combined FESTEM‐EDS approach with large-angle SDD detectors enables rapid, high-resolution mapping of low‐level As and P dopants in state-of-the-art NMOS transistors. Quantitative deconvolution and COMPASS-driven multivariate analysis provide clear phase identification and trace element sensitivity. The live analysis capability improves throughput and supports advanced process development in semiconductor fabrication.

Reference

No explicit literature references were provided in the source document.