Semi-Empirical Molecular Orbital Methods and Ab Initio Calculations in Dye Chemistry: Computational Studies Towards the Design and Synthesis of Organic Pigments

Show full item record

Title: Semi-Empirical Molecular Orbital Methods and Ab Initio Calculations in Dye Chemistry: Computational Studies Towards the Design and Synthesis of Organic Pigments
Author: El-Shafei, Ahmed Mohamed
Advisors: CARL L. BUMGARDNER, Committee Co-Chair
Harold Freeman, Committee Co-Chair
DAVID HINKS, Committee Co-Chair
GERARDO MONTERO, Committee Member
Abstract: The main goal of this study was to determine the scope and limitations of some state-of-the-art methods in computational chemistry, including both molecular mechanics and semi-empirical and ab initio quantum mechanics, in the prediction of key properties of selected colorants and their intermediates. Specifically, prediction of equilibrium molecular geometry, wavelength of maximum absorption, mutagenicity via quantitative structure activity relationships (QSAR), photostability, and the photodegradation mechanism of some nonmutagenic organic pigments prepared from nonmutagenic benzidine analogs was studied. It is known that molecular geometry influences significantly electronic and thermodynamic properties of all compounds. Hence, the accurate prediction of geometries was assessed by the development of protocols using semi-empirical and ab initio methods, for use in comparisons with X-ray crystallographic data. In this regard, single crystals of ten compounds were grown and the associated X-ray structures solved. The effectiveness of each model and protocol was tested by comparison of predicted bond angles, bond lengths, intramolecular hydrogen bond distances, and torsion angles to X-ray data. Electronic properties such as wavelength of maximum absorption were calculated using PPP, ab initio and ZINDO and compared with experimental max data. In the case of semi-empirical calculations, different methods (AM1, PM3, CNDO, INDO, ZINDO, PPP) were employed. With regard to geometry optimization a combination of manual adjustments followed by MM2/PM3 calculations was superior to MM2/AM1 in the prediction of the equilibrium geometry of compounds 149-152. In addition, better results were obtained using PM3 versus AM1 when an optimized energy map was employed for these compounds. While PM3 was effective in predicting the equilibrium geometry of compounds 149-152, AM1 was superior to PM3 in predicting the equilibrium geometry of pigments 153-155. As to the prediction of electronic excitation energy, ZINDO was effective in predicting the wavelength of maximum absorption of pigments 153-155 and the equilibrium geometry and spin state of compounds 174a-b. In the case of ab initio calculations, local and non-local DFT methods were used to predict the equilibrium geometry of compounds 149-152, pigments 153-155 and C.I. Pigment Yellow 12. It was found that the resultant geometries were generally more accurate than those calculated using semi-empirical methods, presumably due to a superior wave function. Ab initio calculations, however, were not effective in identifying the spin state of compounds 174a-b. This suggests that spin calculations involving charged molecules are not yet feasible using ab initio. However, by comparing experimental magnetic susceptibility data for Fe3+-complexed dyes 174a-b with the calculated spin from ZINDO (INDO2/ROHF), it became clear that this is an appropriate method for calculating the spin state of metal complex dyes. In this regard, magnetic susceptibility measurements involving 1:2 Fe3+-complexed dyes 174a-b showed that Fe3+ exists in a high spin state, and INDO2/ROHF calculations showed that the Fe3+ high spin (sextet) state is more likely to exist than the low spin (doublet) state. Using the semi-empirical and ab initio calculated equilibrium geometries for pigments 153-155 and C.I. Pigment Yellow 12, the energy values of HOMO and LUMO states were calculated and these values correlated with experimental photostability data. The predicted HOMO energy values calculated using AM1 and local DFT correlated well with pigment photostability using AM1 and local DFT, while the HOMO-LUMO gap values obtained using a non-local DFT method correlated well with photostability. Thermodynamic descriptors were derived from the equilibrium geometry of training compounds comprising a series of benzidine analogs. In this aspect of the study, two statistical approaches, multi-linear regression and forward step-wise regression (FSR), were used to predict the mutagenicity of the training compounds and structurally related benzidines independently. FSR was found to be effective for predicting mutagenicity of the selected diamines. MLR showed that the key descriptors were surface volume, surface area and heat capacity.
Date: 2004-09-22
Degree: PhD
Discipline: Fiber and Polymer Science

Files in this item

Files Size Format View
etd.pdf 5.291Mb PDF View/Open

This item appears in the following Collection(s)

Show full item record