Rapid factorization of single EEM for dissolved organic matter analysis

Area of Science:

• Analytical Chemistry
• Environmental Science
• Spectroscopy

Background:

• Fluorescence excitation-emission matrix (EEM) spectroscopy is vital for analyzing dissolved organic matter in aquatic environments.
• Traditional methods like parallel factor analysis (PARAFAC) require multi-sample datasets and manual interpretation, limiting real-time applications.

Purpose of the Study:

• To develop a rapid, automated approach for decomposing single EEM inputs into fluorescent components.
• To overcome the limitations of PARAFAC for real-time and in-situ applications.

Main Methods:

• Introduced empirical initialization non-negative matrix factorization (EI-NMF).
• EI-NMF involves chemical rank estimation using singular value decomposition (SVD), empirical initialization, and non-negative matrix factorization.
• Validated using simulated data and natural water samples.

Main Results:

• EI-NMF accurately determined chemical rank and recovered component spectra from simulated data (Tucker congruence coefficients >0.9).
• Decomposition of natural water samples showed excellent accuracy and chemical interpretability.
• The method achieved processing times under 0.1 seconds per EEM.

Conclusions:

• EI-NMF offers a computationally efficient framework for real-time decomposition of individual EEMs.
• This approach has significant potential for in situ monitoring of aquatic fluorescent components.
• EI-NMF provides accurate and chemically interpretable results from single EEM inputs.