SrTiO3 (STO) is a widely investigated material which still attracts attention for fundamental research and applications as well. One the one hand it is generally considered as a textbook example of quantum paraelectric since it remains paraelectric down to 0K (1) . On the other hand point defects and in particular oxygen vacancies (Vo) play an important role for the transport properties (2). Raman spectroscopy is known to be an efficient tool for study vibrational modes and structural phase transitions (SPT). In addition it can provide information on point defects introduced in the lattice, by studying the perturbation (the change) of the spectra (3). Tenne et al (4) have reported on the influence of reduction on the Raman spectra of STO . They found two effects: the appearance of unexpected first-order peaks corresponding to optical phonons of the STO lattice and new lines attributed to local vibrational modes related to oxygen vacancies. The Raman spectrum of STO is very complicated and various processes can coexist even in the cubic phase (5,6) :strong and broad second order scattering and first order peaks with different origins such the deviation from stoichiometry, the presence of impurities or dopant ions, grains or strains .Furthermore, some additional lines appear below the antiferrodistortive (AFD) transition at 105 K, due to unit cell doubling and attributed to phonons at the zone boundary point R. All these processes have their thermal behavior. This renders difficult the analysis of the spectra and the separation between the different effects.As a consequence we have undertaken a new Raman investigation of Vo in STO, using principal components analysis (PCA) which is able to evidence the variability (even very small) between data due to any external parameter change. Thus we compare the temperature (T) dependence of the Raman spectra recorded in a as-grown stoichiometric crystal and a reduced crystal (with a O deficiency estimated to 0.0050) and we attempt to discern the own effect of Vo and the influence of temperature.Spectra recorded at the same temperature in the low T phase in both crystals are reported and compared in figure 1. Raman spectra appear rather complicated in both crystals and particularly in reduced sample. Only small differences occur between two spectra so that principal components analysis (PCA) was used to underline the changes related to Vo, The principle of PCA is to reduce the number of spectral variables using an orthogonal transformation and turn them into uncorrelated variables called principal components (PC) ranked in order of largest possible variance. Raman spectra have spectral variables which are reduced by PCA to scores and loadings. PCA was done on 16 Raman spectra grouping the two sets obtained on treated and untreated samples, from 100 to 1200 cm-1 . In Fig 2 are plotted the scores of the two first components PC1 and PC2 derived from PCA analysis. The spectra are ordered according to T change after projecting the scores of both samples on the axis PC1 while a clear separation is obtained between both samples along the axis PC2, emphasizing the difference related to oxygen vacancies consistently with the direct comparison of spectra (fig1). References: (1) K. A. Müller and H. Burkhard,.Phys. Rev. B 19, 3593 (1979). (2) Z. Wang, M. Cao, Z. Yao, Q. Zhang, Z. Song, W. Hu, Q. Xu, H. Hao, H. Liu and Z. Yu, J. Eur. Ceram. Soc. 34, 1755 (2014). (3) Marc D. Fontana and P. Bourson, Appl. Phys. Reviews 2, 040602 (2015). (4) D. A. Tenne, I. E. Gonenli, A. Soukiassian, et al Phys. Rev. B, 76, 024303 (2007). (5) J. Petzelt, T. Ostapchuk, I. Gregora, et al. Phys. Rev.B, 64, 184111 (2001) (6) T. Ostapchuk, J. Petzelt, V. Zelezny, et al. Phys. Rev. 66, 235406 (2002).
Oxygen vacancies in SrTiO3 highlighted by Raman spectroscopy and principal component analysis
F Cordero
2018
Abstract
SrTiO3 (STO) is a widely investigated material which still attracts attention for fundamental research and applications as well. One the one hand it is generally considered as a textbook example of quantum paraelectric since it remains paraelectric down to 0K (1) . On the other hand point defects and in particular oxygen vacancies (Vo) play an important role for the transport properties (2). Raman spectroscopy is known to be an efficient tool for study vibrational modes and structural phase transitions (SPT). In addition it can provide information on point defects introduced in the lattice, by studying the perturbation (the change) of the spectra (3). Tenne et al (4) have reported on the influence of reduction on the Raman spectra of STO . They found two effects: the appearance of unexpected first-order peaks corresponding to optical phonons of the STO lattice and new lines attributed to local vibrational modes related to oxygen vacancies. The Raman spectrum of STO is very complicated and various processes can coexist even in the cubic phase (5,6) :strong and broad second order scattering and first order peaks with different origins such the deviation from stoichiometry, the presence of impurities or dopant ions, grains or strains .Furthermore, some additional lines appear below the antiferrodistortive (AFD) transition at 105 K, due to unit cell doubling and attributed to phonons at the zone boundary point R. All these processes have their thermal behavior. This renders difficult the analysis of the spectra and the separation between the different effects.As a consequence we have undertaken a new Raman investigation of Vo in STO, using principal components analysis (PCA) which is able to evidence the variability (even very small) between data due to any external parameter change. Thus we compare the temperature (T) dependence of the Raman spectra recorded in a as-grown stoichiometric crystal and a reduced crystal (with a O deficiency estimated to 0.0050) and we attempt to discern the own effect of Vo and the influence of temperature.Spectra recorded at the same temperature in the low T phase in both crystals are reported and compared in figure 1. Raman spectra appear rather complicated in both crystals and particularly in reduced sample. Only small differences occur between two spectra so that principal components analysis (PCA) was used to underline the changes related to Vo, The principle of PCA is to reduce the number of spectral variables using an orthogonal transformation and turn them into uncorrelated variables called principal components (PC) ranked in order of largest possible variance. Raman spectra have spectral variables which are reduced by PCA to scores and loadings. PCA was done on 16 Raman spectra grouping the two sets obtained on treated and untreated samples, from 100 to 1200 cm-1 . In Fig 2 are plotted the scores of the two first components PC1 and PC2 derived from PCA analysis. The spectra are ordered according to T change after projecting the scores of both samples on the axis PC1 while a clear separation is obtained between both samples along the axis PC2, emphasizing the difference related to oxygen vacancies consistently with the direct comparison of spectra (fig1). References: (1) K. A. Müller and H. Burkhard,.Phys. Rev. B 19, 3593 (1979). (2) Z. Wang, M. Cao, Z. Yao, Q. Zhang, Z. Song, W. Hu, Q. Xu, H. Hao, H. Liu and Z. Yu, J. Eur. Ceram. Soc. 34, 1755 (2014). (3) Marc D. Fontana and P. Bourson, Appl. Phys. Reviews 2, 040602 (2015). (4) D. A. Tenne, I. E. Gonenli, A. Soukiassian, et al Phys. Rev. B, 76, 024303 (2007). (5) J. Petzelt, T. Ostapchuk, I. Gregora, et al. Phys. Rev.B, 64, 184111 (2001) (6) T. Ostapchuk, J. Petzelt, V. Zelezny, et al. Phys. Rev. 66, 235406 (2002).I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
