Application of Multivariate Methods to the Analysis of the Fine Structure of Resonant Raman Scattering Spectra

JOSE ROBLEDO; H.J. SÁNCHEZ; JUAN JOSÉ LEANI; C.A. PEREZ; D. PEREZ

Bologna

Congreso; European Conference on X-Ray Spectrometry (EXRS). 2014; 2014

Alma Mater Studiorum University of Bologna

Multivariate Statistical Methods [1] for data analysis are commonly used in manymultidisciplinary areas of science, such as in physics and chemistry. In particular, Principal Component Analysis (one of many Multivariate Statistical Methods) can be applied to analyze the encrypted information contained in the noise of x-ray spectra. In this work, different Multivariate Methods (mainly Principal Component Analysis) are used to analyze Resonant Raman Scattering [2] spectra obtained from oxidized samples with the aim of discriminating oxidation states. Two types of samples were prepared. 1) Multilayer samples formed by two layers of one element in different oxidation states. These samples were measured in grazing incidenceconditions around the critical angle of total reflection. 2) In situ surface oxidation of a metal sample. These samples were irradiated in total reflection conditions at fixed angle using a local heater in order to produce the surface oxidation during the irradiation.Specifically, spectra were obtained from four samples: 1) a chromium multilayer sample consisting of two nanometric layers of CrO (external) and Cr2O3 (internal) at different incident angles, 2) a copper multilayer sample consisting of two nanometric layers of Cu2O(external) and CuO (internal) at different incident angles, 3) a nanometric layer of chromium deposited on a silicon wafer, and 4) a nanometric layer of copper deposited on a silicon wafer.Samples were measured in the Total Reflection Station of the XRF beamline of the Laboratorio Nacional de Luz Sincrotron (Campinas, Brasil) [3]. Monochromatic photons excited the samples so as to produce resonant Raman scattering using an energy dispersive setup. Several spectra were recorded for every stage of the measurements. By using Principal Component Analysis on the spectra measured it was possible to discriminate different oxides at different nanometric depth in the multilayer sample of chromium. This methodology also allowed to detect diffusion phenomena in the multilayer sample of copper, and to find out the oxidation time of the copper surface sample. In addition, some experimental problems that took place during the measurements of the oxidation of the chromium sample were noticed successfully. [1] R. A. Johnson and D. W. Wichern. Applied Multivariate Statistical Analysis. Prentice Hall, Inc., 1992.[2] Y. B. Bannett and I. Freund. Resonant X-Ray Raman Scattering. Phys. Rev. Lett., 34:372-376, 1975.∫[3] C. A. Pérez, M. Radke, H. J. Sánchez, H. Tolentino, R. T. Neuenshwander, W. Barg, M. Rubio, M. I. Silveira Bueno, I. M. Raimundo, and J. R. Rohwedder. X-Ray Spectroscopy, 28(320), 1999.