SYSTEMATIC UNCERTAINTY IDENTIFIED IN MEASUREMENTS OF SILICON DIOXIDE FILM THICKNESSES BY X-RAY PHOTOELECTRON SPECTROSCOPY.

Silicon dioxide is currently used as a gate dielectric material in the semiconductor industry, and it is considered necessary to measure oxide film thicknesses between 1.0 nm and 2.5 nm with a relative uncertainty (3[sigma]) of [+ or -]4 %. Many techniques have been used for such thickness measurements but the results can disagree by more than a factor of two in this thickness range. These disagreements are due to different assumptions in the models for the various techniques and to the lack of adequate data.

X-ray photoelectron spectroscopy (XPS) is one of the techniques that have been used to measure silicon dioxide film thicknesses. The measured thicknesses depend on knowledge of the effective attenuation length (EAL) of the detected photoelectrons in SiO2 for the particular film thickness and measurement configuration. Unfortunately, experimental measurements of the EAL typically have varied by up to 50% and it has been necessary instead to use the electron inelastic mean free path (IMFP). The EAL differs from the IMFP due to the effects of elastic-electron scattering on photoelectron trajectories. The extent of this difference, a systematic uncertainty in the measurement of [SiO.sub.2] film thicknesses by XPS, has not been previously investigated.

NIST has collaborated with the Institute of Physical Chemistry in Warsaw, Poland, in a calculation of EAL values for [SiO.sub.2] films of varying thicknesses and for typical XPS measurement configurations. For common measurement conditions, the ratio of the EAL to the IMFP is approximately constant (within 3 %) for photoelectron emission angles up to 60[degrees]. The average value of this ratio is between 0.906 and 0.935 depending on the x-ray source, the film-thickness range, and the particular XPS configuration. The difference between the value of this ratio and unity, here about 8 %, is a measure of the systematic uncertainty in the XPS thickness measurement. For larger emission angles (often used to increase surface sensitivity), the EAL/IMFP ratio can appreciably exceed unity. These results can be used to make corrections for the systematic uncertainty due to elastic-electron scattering that are appropriate for the specific conditions. Similar calculations can be made for other thin-film materials where the systematic uncertainty in the XPS thickness measurement can be as much as 40 % for common conditions.