Development of a particle swarm optimization based support vector regression model for titanium dioxide band gap characterization

Energy band gap of titanium dioxide (TiO2) semiconductor plays significant roles in many practical applications of the semiconductor and determines its appropriateness in technological and industrial applications such as UV absorption, pigment, photo-catalysis, pollution control systems and solar cells among others. Substitution of impurities into crystal lattice structure is the most commonly used method of tuning the band gap of TiO2 for specific application and eventually leads to lattice distortion. This work utilizes the distortion in the lattice structure to estimate the band gap of doped TiO2, for the first time, through hybridization of a particle swarm optimization algorithm (PSO) with a support vector regression (SVR) algorithm for developing a PSO-SVR model. The precision and accuracy of the developed PSO-SVR model was further justified by applying the model for estimating the effect of cobalt-sulfur co-doping, nickel-iodine co-doping, tungsten and indium doping on the band gap of TiO2 and excellent agreement with the experimentally reported values was achieved. Practical implementation of the proposed PSO-SVR model would further widen the applications of the semiconductor and reduce the experimental stress involved in band gap determination of TiO2.