Our work is focused on advanced data analysis methods for the characterization of silicon nanocrystals (Si-NCs) by high angle annular dark field and electron energy loss spectroscopy (HAADF-EELS) in the aberration corrected and monochromated scanning transmission electron microscope (STEM). These Si-NCs are embedded in multilayer stacks where SiO2, SiC and Si3N4 are used as dielectric barriers. We have developed a collection of fast and powerful computational tools enabling nanometric spatial resolution imaging of the Si-NCs using sub-eV energy resolution EELS.
The starting point of our analysis is the generation of maps from properties measured on the low-loss EELS-SI, such as plasmon energy, EP, and relative thickness, t/λ. For each pixel, EP has been determined, revealing the spatial distribution of the Si-NCs and barrier dielectric material. This method is better suited than the examination of the HAADF images, because of the appearance of spurious features from the inhomogeneity of the sample, masking the Si-NC positions. Nevertheless, it is not possible to get a direct measurement of the pure contribution of the Si-NC to the spectra, as all measured data present at least a mixture of nanoparticle and matrix plasmon.
In spite of this difficulty, segmentation and identification of the different phases in the material have been achieved using mathematical morphology techniques. The result of this approach is that adjacent pixels in the EELS-SI that share a common property are identified. Higher signal-to-noise ratio spectra from the particle and dielectric regions have been generated, along with slices of the EELS-SI. Using these slices, we have made a detailed exploration of spectral factorization using multivariate analysis (MVA) algorithms. Among the tested MVA algorithms, we have learned that NMF and BLU succeed in disentangling the quantum-confined optoelectronic features from the Si-NCs from the matrix EELS signal. Less accessible properties, such as electron effective mass of the Si-NCs, have also been calculated and compared with the expected values. For this purpose, a thickness-normalized Kramers-Kronig analysis algorithm has been prepared.
In summary, maps of the spatially-resolved measured properties, such as EP and t/λ, have been produced for the three studied systems with different dielectric barriers. Moreover, the extraction of particular features by segmentation and MVA factorization of the EELS data has allowed recovering the contribution of the Si-NC to the spectrum for each sample. Finally, the models produced by the MVA algorithms have allowed a novel optoelectronic characterization of the studied embedded nanoparticle systems.