Exploring the physicochemical background of analytical NIRS

Near-infrared spectroscopy (NIRS) gains a remarkable significance in analytical applications nowadays. It is wide spreading in industrial, agricultural, pharmaceutical and other kinds of quality control/assurance contexts. However, the basic science in its background fell behind. Analytical NIRS frequently faces a fine spectral variability which most often remains extensively difficult for direct interpretation due to complexity of NIR spectra. The molecular and physicochemical factors responsible for the observed spectral variability are known only in general. For a given sample, however, the details remain ambiguous. For this reason a work-around in a form of statistical analysis was adapted for analytical routines, but the fact stands that NIR spectral data is being often used without full comprehension. In this sense NIRS is being forcibly used as black-box. Further development of applied NIRS could notably benefit from answering principle questions around its foundation. One of the urgent research questions is how NIR chemometric models react to the principal molecular properties of the analyzed sample, e.g. the intermolecular interactions. Recent advances in the anharmonic theory and technological advance in computing power allow for pushing the frontier of NIRS. The Applicants research so far has demonstrated that accurate spectra simulation of fairly complex molecules has become feasible. The scope of the project is to form a solid foundation for this emerging field by elucidating the correlations between spectral features and molecular structures for a wide selection of molecules. The anharmonic quantum mechanical calculations will form the primary tool for obtaining these objectives. Massively parallel simulation using a computing center will be the core part here. A complementary experimental work will be also an integral part of the project. This project is going to considerably advance the state of knowledge in the field of analytical NIRS, while also being highly beneficial for basic spectroscopy and physical chemistry. The proposed research is strictly novel and original as there were no similar attempts published so far other than preliminary publications originating from the applicants.

Near-infrared spectroscopy (NIRS) gains a remarkable significance in analytical applications in industrial, agricultural, pharmaceutical applications and quality control/assurance contexts. However, the basic science at the foundation of this analytical technique fell behind given difficult for interpretation spectra. The molecular and physicochemical factors responsible for the observed spectral variability are known only in general. For this reason, a work-around in a form of statistical analysis was adapted for analytical routines, limiting NIRS to "black-box" tool used without full comprehension. Further advances depend on fundamental research with one of urgent questions of how NIR chemometric models react to the principal molecular properties of the analyzed sample, e.g. the intermolecular interactions. The project formed a solid foundation for an emerging interdisciplinary field. In boundary-crossing research combining experimental methods of analytical spectroscopy, new generation of miniaturized sensors and novel tools of computational chemistry. The core research developed around anharmonic quantum mechanical simulations of overtones and combinations bands of model systems representing important types of samples for industry and life science. This project had strictly novel and original character, as no similar research directions could be pursued before. The solved problems and answered questions considerably advanced the state of knowledge in the field of analytical NIRS, while also being highly beneficial for basic spectroscopy and physical chemistry. The project found that the two important types of intermolecular interactions, hydrogen-bonding and pi-pi stacking, lead to different manifestations in NIR spectra. This physiochemical difference at the foundation layer clearly translates into the relevance of the associated variables in chemometric calibration models used widely in analytical applications and in the industry. It was found that the effect is subtle, but can be traced back in the regression models also in the important case, where the instrumental difference plays a role. Through this, better understanding of not only the fundamental phenomena (e.g. the physicochemical factors expressed in NIR spectra) can be understood, but also the effects that impact practical applications and real-life analytical scenarios. Consequently, the project firstly brought a significant impact on the basic knowledge with a decisive progress in fundamental understanding of NIR spectra. Secondly, practical applications of NIR spectroscopy benefit from the project, with new findings related to how the instrumental difference can be approached with the knowledge of the chemical information, as demonstrated in this project. The instrumental difference plays a critical role in the applications of novel miniaturized, portable NIR sensors - which are highly valued in various applications in science and industry, but still suffer from 'teething problems". Therefore, the project outcomes contribute sizeable to better utilization of the novel sensors, through essential fundamental research, which has become possible through the unique methods and cross-field research designed for this project.