Classification and pattern recognition of acyclic octenes based on mass spectra

Area of Science:

• Analytical Chemistry
• Chemometrics
• Organic Chemistry

Background:

• Complex product mixtures arise from petrochemical processes like 1-butene transformation.
• Identifying specific isomers, such as acyclic octenes, within these mixtures is challenging.
• Gas chromatography-mass spectrometry (GC-MS) is a key technique for analyzing such complex samples.

Purpose of the Study:

• To develop and optimize classification models for acyclic octene isomers.
• To investigate the impact of spectral data preprocessing techniques on model performance.
• To apply the developed models to real-world GC-MS data from 1-butene dimerization.

Main Methods:

• Development of two Soft Independent Modeling of Class Analogy (SIMCA) models.
• Application of spectral transformations: autocorrelation and logarithmic intensity ratios.
• Investigation of data scaling methods, including square-root scaling.
• Analysis of GC-MS data from liquid-phase 1-butene dimerization over heterogeneous catalysts.

Main Results:

• Both spectral feature preprocessing and data scaling were vital for effective SIMCA model development.
• The optimized SIMCA models demonstrated successful classification of acyclic octene isomers.
• The models proved applicable to complex GC-MS data from catalytic 1-butene dimerization.

Conclusions:

• Chemometric approaches, specifically SIMCA, are powerful tools for analyzing complex hydrocarbon mixtures.
• Careful selection of spectral data preprocessing and scaling significantly enhances classification accuracy.
• This methodology aids in understanding and controlling petrochemical reaction pathways.