Skip to main content
Upcoming Events:

Matls 701/702: Shayan Mousavi, PhD Candidate

Date & Time:
   Add All to Calendar
Location:

MS Teams - team code ffuvgr3

Event Contact:

materials@mcmaster.ca

AI-Assisted Electron Energy Loss Spectroscopy Signal Deconvolution

Overview

Electron energy loss spectroscopy (EELS) is an extremely powerful technique for exploring the chemical, electronic, and optical properties of materials at the sub-nanometer scale. Recent instrumental advancements have made near-meV spectroscopy, for exploring plasmonic and phononic properties of materials more accessible, and more popular than ever. However, the output signal, similar to any spectroscopy technique, is distorted with the optical transfer function of the instrument and high frequency noise of the electronics of the system. This problem is more challenging for near-meV EELS, also known as low-loss region, as the spectral features conveying information regarding plasmonic and phononic behaviors are buried under the strong tail of the zero-loss peak. Therefore, spectral deconvolution is highly crucial for data extraction and detailed signal analysis in the low-loss region.

In this seminar, application of a deep learning architecture in spectral deconvolution of low-loss EELS will be discussed. In this regard, a custom deep learning python software, EELSpecNet, will be introduced and its performance and application in low-loss EELS quantitative analysis will be reviewed.