Guilherme da Fonseca, Bruno2023-08-302023-08-3020232023-08-30http://hdl.handle.net/1828/15314This dissertation discusses the applications of Raman spectroscopy for the characterization of carbon materials, through the investigation of heterogeneous hydrogenation, estimation of elemental and organic carbon from diesel emission particulate and automated classification. Carbon materials are present in several aspects of human life. Some illustrative examples come from technological use of graphite, diamond and glassy carbon and the impact of diesel emissions in human health. The general objective of the dissertation is to improve the characterization of carbon materials and show the potential and limitations of methods that combines Raman spectroscopy and chemometrics for carbon materials characterization. The literature is vast about the utility of Raman spectroscopy for characterizing carbon materials. However, some methods are either not clearly described in terms of reproducibility and implementation, or very specific to a particular type of material. Also, some studies describe their material based in a single spectrum instead of a spatial and statistical characterization. Chapter 2 demonstrates the usefulness of Raman maps for characterizing carbon films obtained chemically. The study investigated the heterogeneous hydrogenation in the samples and showed the effect of different parameters on the homogeneity of the material. Due to heterogeneous hydrogenation, some regions could be assigned to different types of carbon materials and could lead to different properties. This is highly relevant for potential use in industry as quality control tool. For the scientific community, this chapter showed the importance of the fluorescent background in the Raman spectrum for proper characterization of carbon materials. Organic carbon and elemental carbon are general classifications for particulate matter present in diesel emissions. These two species present potential harm to the health of workers, monitoring their presence is thus paramount for a safe workplace environment. The literature offers different approaches for estimating elemental carbon and the comparison with different techniques. Most of them ignore the organic fraction and don’t compare with the North American standard technique (NIOSH method). The fluorescent background and the correlation with hydrogenation of carbon materials that were observed in chapter 2 inspired the use of the background as source of information for estimating organic carbon content. The results from Raman and partial least squares in chapter 3 suggest the background as an important source of information for estimating organic carbon and total carbon. The predictions of total carbon agreed with the values obtained from the standard technique in the range between 60 and 600 µg. Although Raman spectroscopy is largely used for characterization of carbon materials, it is not rare to find misinterpretation of the spectra. Chapter 4 presents an automated prediction of carbon materials based on their Raman spectra and using principal component analysis followed by linear discriminant analysis. The Raman spectra used in for training the model were obtained from highly oriented pyrolytic graphite, glassy carbon, diamond-like carbon, hydrogenated graphite-like carbon and hydrogenated polymer-like carbon. The testing dataset was based on Raman spectra from the literature and from samples synthesized in the lab. The results showed accuracy of 97 % and some assignments found in the literature could be corrected by the model proposed in this chapter. Chapter 5 offers a summary of the research projects presented in this dissertation and discuss the possible future projects for developing the use of Raman spectroscopy and machine learning for carbon materialsenAvailable to the World Wide Webcarbon materialsdiamond-like carbonRaman spectroscopyChemometricsamorphous carbonmachine learningThesisB.G. daFonseca; S.S. Thind and A.G. Brolo, Raman maps reveal heterogeneous hydrogenation on carbon materials, J. Raman Spectrosc. 2021, 52, 516-524.