Weighted least squares estimation of background in EELS imaging.

Area of Science:

• Materials Science
• Spectroscopy
• Analytical Chemistry

Background:

• Quantitative Electron Energy Loss Spectrometry (qEELS) relies on accurate background estimation.
• Variances in EELS data necessitate advanced statistical methods like weighted least squares.

Purpose of the Study:

• To investigate the efficacy of weighted least squares estimation for background modeling in qEELS.
• To analyze the impact of weighting on signal-to-noise ratio (SNR) and identify sources of error.

Main Methods:

• Application of weighted least squares (WLS) estimation for background fitting below core edges in EELS data.
• Analysis of the resulting SNR above the core edge.
• Quantification of bias introduced by logarithmic data transformation.

Main Results:

• Theoretical expectation of improved SNR with WLS was not met.
• Proper weighting was found to decrease, rather than increase, the above-edge SNR.
• Logarithmic transformation of EELS data introduces bias affecting SNR.

Conclusions:

• Standard WLS for background estimation in qEELS may not always enhance SNR.
• The bias from logarithmic data transformation is a critical factor influencing SNR in qEELS analysis.
• Further research is needed to optimize background estimation techniques in EELS.