Dimensionality-reduction techniques for complex mass spectrometric datasets: application to laboratory atmospheric...

Area of Science:

• Atmospheric Chemistry
• Environmental Science
• Chemical Kinetics

Background:

• Atmospheric oxidation of organic compounds creates complex product mixtures challenging for measurement and modeling.
• Mass spectrometry yields vast datasets, necessitating effective data reduction for chemical process understanding.
• Existing data reduction methods lack systematic investigation for atmospheric oxidation systems.

Purpose of the Study:

• To evaluate three dimensionality reduction approaches: Positive Matrix Factorization (PMF), Hierarchical Clustering Analysis (HCA), and Gamma Kinetics Parameterization (GKP).
• To assess the effectiveness of these methods in simplifying complex atmospheric oxidation product datasets.
• To determine the utility of these approaches for atmospheric chemistry modeling and mechanism development.

Main Methods:

• Evaluation using synthetic data from a three-generation oxidation system with known parameters.
• Analysis of measured products from OH-initiated oxidation of a substituted aromatic compound in a chamber experiment.
• Comparison of PMF, HCA, and GKP in grouping and characterizing complex chemical species.

Main Results:

• PMF captured temporal composition changes but did not group compounds by generation or chemical process.
• HCA identified major ion groups and behavioral patterns, preserving bulk chemical properties like carbon oxidation state.
• GKP successfully parameterized kinetic behavior, fitting species' time traces to determine generation and effective rate constants, reducing thousands of species to 10-30 groups.

Conclusions:

• GKP offers a quantitative link between chemical and kinetic characteristics, significantly reducing system complexity.
• The grouped species exhibit characteristic chemical compositions and kinetic behaviors, aiding in understanding oxidation systems.
• This approach provides a viable method for mechanism development and improving atmospheric chemistry models.