Machine learning for predicting the average length of vertically aligned TiO2 nanotubes

Technological advances depend on the study of specific materials, such as TiO2 nanotubes that have a variety of applications in different industries due to their properties. These properties are directly related to the nanotubes, size, for example, with their length; hence, measuring this dimension accurately is important. Nowadays, length measurement is performed through semi-automatic functions on scanning electron microscopy images. Time-consuming image analysis, subjective and low-representative readings, and damaged samples are some disadvantages found in this process. This paper presents a proposal for predicting the average length of vertically aligned TiO2 nanotubes using machine learning and ellipsometry because they can overcome the disadvantages mentioned. Different models of measurements of light reflection intensity and ellipsometric parameters predicted the length. The results of a model that showed a low prediction error using linear support vector machines for regression are reported.