Knowledge-based genetic algorithm for resolving the near-infrared spectrum and understanding the water structures in aqueous solution

Abstract

The structure of water has been an interesting subject in various research fields due to its complexity and flexibility. Near-infrared (NIR) spectroscopy has been proven to be a powerful tool for analyzing the structure of water in aqueous solutions. However, low resolution of the spectrum and overlapping of the peaks make it difficult to obtain the spectral components for understanding the structure of water. In this work, a knowledge-based genetic algorithm was developed for Gaussian fitting of the NIR spectra of water measured at different temperatures. The number of the water structures was determined from the result of molecular dynamics simulation, and the spectral peak of the structures was simulated by a Gaussian function. Continuous wavelet transform was adopted to enhance the resolution and remove the background variation of the spectra. Through the variation of the water structures with temperature, the dissociation of the larger clusters into smaller ones was observed. Furthermore, an enhancement of the ordered (tetrahedral) water structures is indicated by the relative content in the glucose solutions of different concentrations, providing a proof for glucose making the water structure more ordered to explain the protective effect of glucose on the bio-molecules in bio-aqueous solution.