Applied electric field effect on light-scattering birefringence of dye-doped liquid crystal molecule and consistent neural network empirical physical formula construction for scattering intensities

• Published in: Journal of molecular liquids Vol. 163; no. 3; pp. 153 - 160
• Main Authors: Polat, Ömer, Yildiz, Nihat, San, Sait Eren
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 15.10.2011
• Subjects: Birefringence; Liquid crystal; Molecular structure; Neural network; Scattering intensity; Birefringence; Neural network; Liquid crystal; Molecular structure; Scattering intensity
• ISSN: 0167-7322; 1873-3166
• DOI: 10.1016/j.molliq.2011.08.012

In this paper, we realized two objectives. Firstly, birefringence of azo and anthraquinone dye-doped nematic liquid crystal (NLC) molecules was investigated by applied electric field dependent laser scattering intensities. The birefringence was essentially calculated from ordinary and extraordinary ray phase difference, which is determined from the measured intensities corresponding to parallel and perpendicular orientations of analyzer to polarizer. The birefringence was found to be dependent on both applied voltage and the kind of the doping dye. As the second objective, by nonlinear universal function approximator layered feedforward neural network (LFNN), we constructed explicit form of empirical physical formulas (EPFs) for experimentally measured dye-doped NLC nonlinear scattering intensities. Excellent LFNN test set predictions over yet-to-be measured experimental data proved that the constructed LFNN-EPFs estimated the measured intensities consistently. The correlation coefficients assessing the goodness of predictions were about r = 0.998for all cases. The LFNN-EPFs also extracted the intensity dependency on the kind of dye used. When theoretical and LFNN-EPFs intensities are compared, we conclude that given certain experimental conditions, theoretical and LFNN-EPFs predictions are in excellent agreement. In this sense, we can say that the physical laws embedded in the birefringence scattering data can be consistently extracted by LFNN. Therefore, judging from the consistent extraction of the molecular dependencies of pure and doped NLC intensities, we predict that the LFNN-EPFs can help to identify unknown molecular structural parameters in liquid crystal extracts. More concretely, by suitable mathematical operations such as differentiation, integration, minimization on these intensity LFNN-EPFs, some useful information into the charge distributions of the LC molecules can be gained. ► We investigate the birefringence in dye-doped nematic liquid crystal (NLC) molecules. ► To do this, we use NLC laser scattering intensities under applied electric field. ► We construct neural network empirical physical formulas (EPFs) for NLC intensities. ► Neural network EPFs can help to identify unknown liquid crystal molecular parameters.