XPS: binding energy calibration of electron spectrometers 4?assessment of effects for different x-ray sources, analyser resolutions, angles of emission and overall uncertainties

Abstract

A detailed analysis is made of the binding energy calibration of X-ray photoelectron spectrometers when using monochromated Al Kα x-rays or unmonochromated Al or Mg Kα x-rays. The binding energies of the peaks for Cu 2p3/2, Ag 3d5/2 and Au 4f7/2, as well as for the Ni Fermi edge, are measured at high resolution using monochromated Al Kα x-rays. The apparent binding energy shifts of the peaks are then calculated for this source, and also for the Al and Mg unmonochromated x-ray sources, using full synthetic Kα x-ray structures, as a function of Gaussian spectrometer energy resolutions in the range 0.2–1.5 eV. For all three x-ray sources, the relative binding energies for the Cu 2p3/2 and Au 4f7/2 peaks are contained within ±0.015 eV but the effects for Ag 3d5/2 are stronger and the containment range must be increased to ±0.026 eV. Further data and calculations are provided for surface core-level shifts and here it is found necessary to restrict emission angles to 56° for all the peak separations to be restricted to the above range of ±0.026 eV. Other instrumental effects may give rise to additional larger or smaller effects. Non-optimized settings for monochromators can show further shifts of up to ±0.2 eV. The uncertainties associated with the above calibration are then analyzed to show how the uncertainty at 95% confidence varies across the binding energy range. Example calculations show that seven repeats of both the Cu 2p3/2 and Au 4f7/2 binding energies may be used to define the peak repeatability and that one or two measurements can then be made for each calibration peak to define the calibration. The precise number of measurements to be used depends on the peak energy repeatability and the required confidence limits for the calibration. In practical situations, however, it is likely that the greatest uncertainty in the binding energy scale arises from the drift in the electronics between calibrations. © 1998 John Wiley & Sons, Ltd.