Principal Component Analysis as a Post-processing Tool in Micro- and Nano-photonics Measurements and Modelizations

The signals typically extracted from nano-devices are weak and can be strongly masked by noise, presenting a low signal-to-noise ratio. Computational electromagnetism and high resolution measuring techniques in Nano-Science an Nano-Technology produce a large amount of data that needs dedicated post-processing to enhance pure spatial, pure temporal, or mixed spatial-temporal structures embedded in the data. The Principal Component Analysis (PCA) has been proved in other fields as a very useful tool able to reveal inner correlations, and hidden structures in very large data sets. Among other applications, PCA has been used to characterize spatial-temporal patterns of noise in visible and infrared imaging devices. In this contribution, we offer an statistical technique, the PCA, that is well suited to extract useful information, even from faint signal embedded in noise but having a non negligible degree of correlation. PCA is blind in nature, i.e., it does not need any prior preparation of the analyzed data. However, the correct interpretation of the results requires some knowledge of the physics and technology supporting the studied phenomena. We show how PCA is useful for a variety of Nano-Science and Nano-Technology applications. The cases treated here involve a large amount of numeric data, and experimental techniques dealing with weak signals and strong noise structures.