Theoretical and empirical modeling of retention behavior of olanpazine in hidrophilic interaction liquid chromatography

Poslednju deceniju susrećemo se sa brojnim izazovima kada je u pitanju analitika različitih analita primenom tečne hromatografije hidrofilnih interakcija (eng. Hydrophilic Interaction Liquid Chromatography, HILIC). Uprkos stalnom porastu broja publikovanih radova i dalje se sa sigurnošću ne može predvideti ponašanje različitih analita u ovom sistemu hromatografije. Odgovor leži u složenosti retencionih mehanizama, te su i dalje neophodna naučna istraživanja koja omogućavaju da se ovaj hromatografski sistem detaljno prouči. Shodno tome, cilj ove doktorske disertacije bio je da se primene teorijski i empirijski modeli u cilju definisanja retencionog ponašanja strukturno sličnih analita (olanzapina i njegovih srodnih supstanci) u HILIC sistemu, kao i da se stečena znanja primene kroz odgovarajuću primenu nove metode. Ispitivana smeša analita olanzapina i njegovih osam srodnih supstanci po prvi put je ispitivana u HILIC sistemu u ovoj doktorskoj disertaciji. Pored toga, dve srodne supstance (2 i 8) predstavljaju supstance koje se po prvi put hromatografski opisuju i za koje ne postoje literaturni podaci. U okviru preliminarnih ispitivanja ispitane su četiri HILIC stacionarne faze (silika, diolna, cijano i cviterjonska) pod istim hromatografskim uslovima, čime je omogućeno donošenje preliminarnih zaključaka o uticaju fizičko-hemijskih karakteristika stacionarne faze, kao i analita na retenciono ponašanje u HILIC sistemu. U sledećoj fazi istraživanja, pristupilo se ispitivanju uticaja polarnije komponente mobilne faze (vodeni rastvor pufera), kako bi se dobile informacije o mehanizmima koji su uključeni u celokupan proces retencionog ponašanja. Izvršeno je fitovanje dobijenih rezultata retencionih vremena u teorijske (adsorpcioni, particioni, mešoviti i kvadratni model) i eksperimentalne (Neue model) matematičke modele. Za ocenu podobnosti matematičkih modela koristili su se koeficijent determinacije (R2) i unakrsno validirani koeficijent determinacije (Q2) kao statistički parametri koji opisuju kvalitet matematičkog modela...Over the last decade, we have been facing numerous challenges regarding the analysis of different substances in HILIC (Hydrophilic Interaction Liquid Chromatography) mode. Despite constant increase in the number of published scientific papers, retention mechanisms still cannot be predicted with great certainty. The answer can be found in the complexity of retention mechanisms, which creates the need for further scientific investigations in order to better understand the system. The goal of this doctoral thesis was to apply theoretical and empirical mathematical models in order to define retention behavior of the structurally similar substances (olanzapine and its related substances), which would help us use gained knowledge for the newly created model. This specific mixture of substances was investigated for the first time in HILIC in this doctoral thesis. Furthermore, two of the related substances (2 and 8) have never been investigated in any chromatography system before. Through preliminary investigations, four HILIC columns have been tested (silica, diol, cyano-propyl and zwitterionic) under the same chromatographic conditions, which allowed us to make conclusions about the influence of physicochemical characteristics of the columns and substances on the retention behavior in HILIC. In the next phase, for the purpose of investigating retention mechanisms, the influence of the more polar component of the mobile phase (aqueous buffer solution) on the retention was investigated. Fitting of the obtained results was conducted in the theoretical (adsorption, partition, mixed and quadratic) models and experimental (Neue) models. For the statistical evaluation of the models, coefficient of determination (R2) and coefficient of prediction (Q2) were used. Since the existing mathematical models did not have sufficient level of fit, in the next phase of the study, spline interpolation model was created. Spline model has never been used for the modeling of chromatographic behavior, prior to this study..