Comparison of the accuracies of two methods for the determination of the surface normal for X-ray photoelectron spectroscopy

Determining the true angle of detection of emitted electrons is a major factor for the accurate determination of film thicknesses in the range below 10 nm by x- ray photoelectron spectroscopy. To do this, the orientation of the surface normal in the instrument first requires determination. A comparison is made of two recent methods for the definition of the surface normal in x- ray photoelectron spectrometers. The first method, which we call the crystal axis method, involves the measurement of the Si 2p oxide and substrate intensities in the range +/- 5 degrees around the surface normal for a (100) Si wafer with approximately 1 nm of thermal oxide. The second method, which we call the overlayer thickness consistency method, involves measuring similar intensities at 10 degrees, 20 degrees, 30 degrees and 40 degrees from the surface normal, for an amorphous oxide on amorphous Si sample, with a view to obtaining a consistent thickness. It is shown that both can readily define the surface normal alignment better than 1 degrees but that the crystal axis method, for similar data quality and numbers of spectra, generates uncertainties a factor of approximately three times better than that of the overlayer consistency method and has fewer bias contributions. Possible improvements in the method are shown to be best by recording more data in the crystal axis method rather than, for example, by changing the materials.