Lawrence Berkeley National Laboratory

High Resolution Transmission Electron Microscopy (HR-TEM) has been used as the ultimate method of thickness measurement for thin films. The appearance of phase contrast interference patterns in HR-TEM images has long been confused as the appearance of a crystal lattice by non-specialists. Relatively easy to interpret crystal lattice images are now directly observed with the introduction of annular dark field detectors for scanning TEM (STEM). With the recent development of reliable lattice image processing software that creates crystal structure images from phase contrast data, HR-TEM can also provide crystal lattice images. The resolution of both methods was steadily improved reaching now into the sub Angstrom region. Improvements in electron lens and image analysis software are increasing the spatial resolution of both methods. Optimum resolution for STEM requires that the probe beam be highly localized. In STEM, beam localization is enhanced by selection of the correct aperture. When STEM measurement is done using a highly localized probe beam, HR-TEM and STEM measurement of the thickness of silicon oxynitride films agree within experimental error. In this paper, the optimum conditions for HR-TEM and STEM measurement are discussed along with a method for repeatable film thickness determination. The impact of sample thickness is also discussed. The key result in this paper is the proposal of a reproducible method for film thickness determination.