Abstract
A set of signal processing methods comprising fast Fourier transform interpolation, maximum entropy deconvolution and wavelet transformation has been successfully integrated to improve the equality of the extracted C K-edge spectra from electron spectroscopic imaging (ESI) series. Fast Fourier transform interpolation is used to improve the dispersion arising from discrete sampling of ESI series in the energy space. The maximum entropy method is used to dispel the convolution effect resulting from that ESI series acquired with a finite energy window. Wavelet transformation is applied to de-noise the extracted ESI spectrum. The post-processed ESI spectrum has quality as good as that of a probe-acquired spectrum and makes semi-quantitative analysis of the two-dimensional sp 2 /sp 3 ratio map in diamond-like carbon thin film possible. In general, this method is applicable for reconstructing good quality core-loss electron energy-loss spectra from a nanometre-sized area, so that it may be possible to quantitatively analyse two-dimensional information about electronic structure in materials with near nanometre resolution.