Intercomparison of algorithms for background correction in XPS

Abstract

In the present interlaboratory comparison, the consistency and validity of the Shirley, straight-line and Tougaard methods for determination of peak intensities from measured XPS spectra have been studied by analysis of data on Au and Ni taken at eight laboratories. The reproducibility of the individual measuring systems was determined to be 26%. The RMS scatter around their mean values of the ratio of the peak intensity to the Au 4d peak intensity was 6% for the Tougaard method, 9-10% for the Shirley method and 8-9% for the straight-line method. The increase in root mean square (RMS) scatter of data due to data being recorded in different laboratories amounts to 3-4%, independent of the method applied for background correction. The RMS deviation from theory of intensity ratios of peaks from Ni and Au to the Au 4d peak intensity is 14% for the Tougaard method, 33% for the Shirley method and 31% for the straight-line method. The absolute accuracy of the methods cannot be established to better than the accuracy on the theoretical peak intensity ratios and the uncertainty in the energy dependence of the analyser response function. It is, however, reasonable to expect the combined uncertainty of these factors to be at least 10-15%. The peak intensity ratios obtained with the Tougaard method then falls within the expected accuracy of the theoretical ratios. INTRODUCTIONThe present intercomparison work follows a previous extensive study' of the validity and consistency of methods for determination of XPS peak intensities. The notation and procedures for data analysis used in the present work are the same as in Ref. 1 and will only be summarized briefly here.Quantitative surface composition analysis by XPS relies on peak intensities. Peak areas are used as a measure of the peak intensity. Procedures to determine the peak area from a measured spectrum is complicated by the inelastic scattering events that the photoelectrons undergo as they leave the solid. This gives rise to an inelastic background intensity, which is subtracted to define the peak areas. Several methods have been used to define this background. Among these, the Shirley and the straight-line methods are most widely used. these methods, uncertainty arises from the problem of defining suitable start and end points for the evaluation of the background signal. More recently, a method that relies on a detailed description of the physical processes involved was proposed by T~ugaard.~.' This method has been applied by several groups, who have named the procedure the Tougaard method. To be consistent with this convention, we shall use here the same name for this background correction method. The complexity of the method is similar to that of the Shirley method. Because of shake-up processes, the XPS peak intensity extends over an energy range of 50-200 eV.5 To account for these shake-up processes, which vary with the sample composition, it is necessary to measure spectra over at least this energy range. For inhomogeneous samples, analysis of wide-scan spectra als...

Evaluation of uncertainties in X‐ray photoelectron spectroscopy intensities associated with different methods and procedures for background subtraction. I. Spectra for monochromatic Al X‐ray

“…29 (for case 2). in Figs 4,8, and 12, although some additional factor is needed to account for the extremely low value of R for participant 15 for case 2. All results from participant 23b and two results from participant 15 for case 2 were much higher than expected from the trends in Figs 27 and 31.…”

“…They found that peak intensities obtained using the Tougaard method for background subtraction agreed much better with the calculated intensities than those obtained with the linear and Shirley methods. An additional assessment of the three methods was reported by Jansson et al [8] who analyzed peak intensities obtained using the three methods from XPS spectra of Ni and Au measured in eight laboratories. Ratios of intensities of Ni and Au peaks to the Au 4d intensity were compared with ratios of the corresponding calculated intensities.…”

Standard test data for estimating peak-parameter errors in x-ray photoelectron spectroscopy. I. Peak binding energies

“…13. In addition to the individual factors in the design leading to signiÐcant di †erences in peak BE biases, the interactions among factors were also signiÐcant at the 5% signiÐcance level in many of the data sets.…”

Design principles of the wxEWA spectrum evaluation program for the photoelectron spectroscopy

“…The users' community tries to help the overall understanding of the evaluation needs through issuing standard spectral data [3,4] and making inter-laboratory comparisons [5] of the sample handling and data evaluation methods. Some data evaluation methods have been inherited from the nuclear spectroscopy [6], some others have been derived after facing field-specific problems [7,2].…”