Use of Raman microscopy and band-target entropy minimization technique to differentiate physical mixture from chemical mixture in mixed metal oxides Effendi Widjaja * , Jeyagowry T. Sampanthar * , Xuan Ding Han, Eunice Goh Institute of Chemical and Engineering Sciences, Agency for Science, Technology and Research (A*STAR), 1 Pesek Road, Jurong Island, Singapore 627833, Singapore Available online 3 December 2007 Abstract Raman microscopy has been applied to characterize physical and chemical mixtures of mixed metal oxides. The obtained Raman mapping data were first subjected to singular value decomposition to obtain the right singular vectors, and the right singular vectors were then subjected to band- target entropy minimization (BTEM) to recover the pure component spectra of the observed species present in the sample. Subsequently, these resolved pure component spectral information was used to differentiate the physical and chemical mixtures. In addition, BTEM is also able to recover the pure component spectra of both unstable compound and its degradation product due to laser irradiation. In current study, the physical mixture of Mn 2 O 3 and Co 3 O 4 , and the chemical mixture of CoMn 2 O 4 spinel oxide were investigated. # 2007 Elsevier B.V. All rights reserved. Keywords: Raman microscopy; Band-target entropy minimization; Physical mixture; Chemical mixture; Mixed metal oxide; Raman bands 1. Introduction Raman spectroscopy, which is one of the various analytical instruments used to characterize solid catalysts, is used to obtain the characteristics of the surface molecular structures. Raman spectroscopy has been applied to characterize various bulk and supported single and mixed metal oxides [1–10]. In situ Raman spectroscopy has also been applied to investigate the structural changes of metal oxides during transformations [11,12]. With the development of analytical instrumentation, an optical microscope can be attached to a Raman spectrometer to build a Raman microscopy. It is now possible to construct a Raman map containing both spectral and spatial information from a sample under characterization. This mapping measure- ment is performed by scanning in a grid pattern over the surface of the sample via programmed movements of a microscope objective. Therefore, using Raman microscopy will enable a user to examine the Raman band structures which are associated with chemical and structural properties of the investigated materials as well as the spatial distribution of materials in two spatial dimensions. In current study, we attempted to develop an alternative method, which can be applied to differentiate between physical and chemical mixtures in the prepared mixed metal oxides. This method works by combining Raman mapping measurements with a multivariate data analysis technique, i.e. self-modeling curve resolution (SMCR) [13–16]. SMCR is a numerical technique specifically developed to resolve pure component spectra from a set of mixture spectra without any recourse to spectral library. The ultimate goal of SMCR is to determine the number of detected individual constituents, their identity, and their relative concentrations directly from the original data. When a Raman mapping data is analyzed, SMCR is also able to produce the spatial distribution of each component. Among all available SMCR techniques, band-target entropy minimization (BTEM) [17] is specifically developed to recover signals from trace components and to reconstruct the full-range pure component spectrum associated with a targeted spectral features. BTEM has been applied to a large variety of liquid-phase multi-component systems [18–20] as well as www.elsevier.com/locate/cattod Available online at www.sciencedirect.com Catalysis Today 131 (2008) 21–27 * Corresponding authors. Fax: +65 6316 6185. E-mail addresses: effendi_widjaja@ices.a-star.edu.sg (E. Widjaja), T_Jeyagowry@ices.a-star.edu.sg (J.T. Sampanthar). 0920-5861/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.cattod.2007.10.079