Nanoscale-level dielectric property image of low- k dielectric materials for copper metallization using energy-filtered TEM

The dielectric properties of low-k material have been characterized using image-spectrum technique via Kramers- Kronig analysis. Quantitative analysis of experimental image-spectrum has been improved using two new quantitative methods. FFT interpolation and maximum entropy deconvolution were successfully used to solve the two problems: under- sampling and loss of energy resolution in image-spectrum technique, respectively. In this study, carbonated SiO2 based low-k dielectric layer designed for copper metallization was used as a demo example. We show that the reconstructed image-spectrum obtained from ESI series images can be quantified with the same accuracy as conventional EELS spectrum. We also developed a new method to quantitatively determine dielectric constant for low-k materials. We have determined the thickness of the carbonated SiO2 based low-k material using extrapolated thickness method from the materials of known dielectric constants. The dielectric function map can be deduced from 2-dimensional reconstructed single scattering spectra with providing the information of thickness via Kramers-Kronig analysis. We proposed a four-dimensional data presentation for revealing the uniformity of the energy dependent property. The accuracy of our methods depends on the thickness determination and on the quality of the reconstructed spectra from the image series. Finally, the dielectric property of carbonated SiO2 based low-k material after annealed process was investigated using Kramers-Kronig analysis to found the relationship of dielectric constant and material density.