Application of multivariate statistical analysis to STEM X-ray spectral images: interfacial analysis in...

Area of Science:

• Materials Science
• Analytical Chemistry
• Microscopy

Background:

• Scanning transmission electron microscopy (STEM) coupled with energy-dispersive X-ray (EDX) spectroscopy generates complex spectral image data.
• Traditional analysis methods may not fully resolve subtle chemical variations or artifacts within these datasets.
• Advanced statistical techniques are needed to extract maximum information from high-dimensional spectral imaging.

Purpose of the Study:

• To apply multivariate statistical analysis, specifically multivariate curve resolution (MCR), to STEM-EDX spectral images.
• To demonstrate the capability of MCR in enhancing spectral contrast and identifying unknown components.
• To showcase MCR's utility in microelectronics failure analysis.

Main Methods:

• Application of multivariate curve resolution (MCR) to STEM-EDX spectral image data.
• Comparison of MCR results with conventional analysis techniques.
• Focus on microelectronics failure analysis case study.

Main Results:

• MCR successfully resolved an unexpected chemical contaminant within the spectral image.
• MCR identified a component related to foil thickness changes, likely an artifact from focused ion beam (FIB) preparation.
• The MCR approach provided a higher spectral contrast view compared to conventional analysis.

Conclusions:

• Multivariate curve resolution is a powerful tool for analyzing complex STEM-EDX spectral data.
• MCR enhances the detection of subtle chemical species and preparation-induced artifacts.
• This method significantly aids in detailed failure analysis of microelectronic devices.