The Chemical Significance of XPS BE Shifts: A Perspective

A recent publication from Paul Bagus (University of North Texas), Connie Nelin, and Dick Brundle [1] discusses the chemical significance of XPS BE shifts. Paul, Connie and Dick have made many outstanding contributions to the field of XPS, in particular by using computational chemistry approaches to model various XPS phenomena and spectral shapes - especially of transition metal species with complex multiplet splitting and satellite structures. 

Dr. Bagus describes this perspective below. This will be a good starting point for researchers interested in applying MO theory to XPS measurements. 

An all too common interpretation of the shifts of XPS BEs, Delta BE, for a given element in different compounds and in different environments is to relate the sign of the BE shift to the change in the effective charge, Q, of the core ionized atom. Thus, a shift to lower BE from sample 1 to sample 2 is interpreted as meaning that the atom in sample 2 has a smaller positive Q or a larger negative Q than the same atom in sample 1. Similarly, a shift to a larger BE is taken to mean that the atom in sample 2 has a larger Q. This paper shows that this simple interpretation of BE shifts is incomplete and that it is likely to be misleading.

While the effective charge Q does contribute to BE shifts, it is not the only physical or chemical mechanism that can contribute to XPS BE shifts. Two other mechanisms are the environment of the ionized atom that can lead to electrostatic potential that are different at different sites in a given sample and are different for different samples. Another mechanism is the degree of hybridization of an atom again at different sites and different compounds. An important objective of this perspective is to examine the mechanisms that lead to BE shifts. The chemical and physical content of these different mechanisms is first examined for a model system. With this model system, the different mechanisms can be separated and the magnitudes of the XPS BE shifts due to the different mechanisms can be understood directly in terms of the electronic charge distribution. Then five specific examples of XPS BEs measured for real systems are discussed and the observed BEs related to the physical and chemical mechanisms which are the origin of the BE shifts. The paper also considers the initial and final state contributions to the BE shifts and identifies when it is likely that initial state effects will dominate.

An important goal of the paper is that the principles and mechanisms for BE shifts can be applied, not only to the specific systems discussed in the paper but also to the understanding of the Delta BE for systems in general. It can lead readers to make suggestions for theoretical studies to help explain specific observations of BE shifts.

[1] P.S. Bagus, C.J. Nelin, C.R. Brundle, J. Vac. Sci. Technol. A 41 068501 (2023).

XPS Reference Pages

This site contains information gained from decades of X-ray photoelectron spectroscopy (XPS) analyses of an enormous variety of samples analyzed at Surface Science Western laboratories located at the University of Western Ontario. Originally this site was designed as a place for students and our clients to access valuable tips and information. It has since been opened to all those interested in the XPS technique. Summaries of literature data, relevant references and unpublished data taken of well characterized standard samples are presented. Also curve-fitting tips, instrument set-up tips (specifically for the Kratos AXIS Supra, Ultra and Nova), and CasaXPS tips pertaining to questions we normally get from our students and clients, and other odd bits of information are presented.

The fine print:
Surface Science Western and the University of Western Ontario does not warranty any of the information shown at this site. Any use of this data in scientific publications or other forms should include referencing to the originally published data referenced herein.

Asymmetric Peak Shapes

For conductive samples, such as metals and graphite, there is a distribution of unfilled one-electron levels (conduction electrons) that are available for shake-up like events following core electron photoemission. When this occurs, instead of a discrete structure like that seen for shake-up satellites, a tail on the higher binding energy side of the main peak – an asymmetric peak shape is evident[1]. An example of this is shown in Figure 1 for a sputter cleaned vanadium metal surface. It is clear from this figure that the asymmetric tail of the metal peak shape will overlap with higher oxidation state species. As such it is important that the total photoelectron yield contribution from the metal is captured during curve-fitting analysis. The use of standard spectra that is fit with mathematically derived asymmetric peak shapes allows for this.

Figure 1. Asymmetric peak shapes in the V 2p spectrum of an argon ion sputter cleaned surface of vanadium metal [2].

David Morgan at Cardiff University has recently published an excellent insight article [3] on asymmetric peak shapes in XPS.  This article goes into detail about the causes of asymmetry, curve-fitting of asymmetric peaks, implications of using hard X-ray sources (HAXPES), and asymmetry in other materials such as conductive metal oxides, graphitic materials, and polymeric materials. Well worth the read for a more in-depth look.

[1] D. Briggs, XPS: Basic Principles, Spectral Features and Qualitative Analysis, in: D. Briggs, J.T. Grant (Eds.), Surface Analysis by Auger and X-ray Photoelectron Spectroscopy, IM Publications, Chichester, 2003, pp. 31-56.
[2] M.C. Biesinger, L.W.M. Lau, A.R. Gerson, R.St.C. Smart, Resolving Surface Chemical States in XPS Analysis of First Row Transition Metals, Oxides and Hydroxides: Sc, Ti, V,Cu and Zn, Applied Surface Science, 257 (2010) 887-898.
[3] D.J. Morgan, XPS insights: Asymmetric peak shapes in XPS, Surface and Interface Analysis, 55 (2023) 567-571.

Using Adventitious Carbon for Charge Correcting

The C 1s spectrum for adventitious carbon can be fit by defining a peak constrained to be 1.5 eV above the main peak, of equal FWHM to the main peak (C-C, C-H). This higher binding energy peak is ascribed to alcohol and/or ester functionality (C-OH, C-O-C). Further high binding energy components can be added if required. For example: C=O at approximately 3 eV above the main peak and O-C=O at 3.8 to 4.3 eV above the main peak. One or both of these peaks may also have to be constrained to the FWHM of the main peak if they are poorly resolved.  Reference [1] and the table below outline standard starting fitting parameters for adventitious carbon. 
Adventitious carbon C 1s curve-fitting parameters [1].
Spectra from insulating samples can then be charge corrected by shifting all peaks to the adventitious C 1s spectral component (C-C, C-H) binding energy set to 284.8 eV. There is certainly error associated with this assignment. Swift [2] lists a number of studies showing errors ranging from ±0.1eV to ±0.4 eV.  “Newer” studies (late 1970's) range from ±0.1 to ±0.3 eV. “Older” studies (late 1960's to early 1970's) were in the ±0.4eV range - however, reproducibility and resolution of the spectrometers of the time may have played a role.  Barr's [3] work from 1995 states that error in using adventitious carbon is ±0.2 eV.  Our work [4] in 2002 also suggests error in the ±0.2eV to  ±0.3eV range.  Experience with numerous conducting samples (1995 to present) and a routinely calibrated instrument have shown that the C 1s signal generally ranges from 284.7 eV to as high as 285.2 eV [5].  Reference [1] presents a detailed assessment of the analysis of insulating samples from a multi-user facility from over a 5-year period that showed an adventitious C 1s (C-C, C-H) binding of 284.91 eV ±0.25eV.  

For organic systems, especially polymers, it is convenient to charge correct to the C-C, C-H signal set to 285.0 eV. This makes for easier comparison to the polymer handbook [6] which uses this number for charge correction.

[1] M.C. Biesinger, Appl. Surf. Sci, 597 (2022) 153681.
[2] T.L. Barr, S. Seal, J. Vac. Sci. Technol. A 13(3) (1995) 1239.
[3] P. Swift, Surf. Interface Anal. 4 (1982) 47.
[4] D.J. Miller, M.C. Biesinger, N.S. McIntyre, Surf. Interface Anal. 33 (2002) 299.
[5] M.C. Biesinger, unpublished results
[6] G. Beamson, D. Briggs, High Resolution XPS of Organic Polymers - The Scienta ESCA300 Database Wiley Interscience, 1992.

Graphitic/Graphene/Carbon Nanotube C 1s Curve-Fitting

Materials of a graphitic nature (e.g., graphite, graphene, carbon nanotubes etc.) will have a C 1s main peak, attributed to C=C, which can be used as a charge reference set to 284.5 eV. An average of values for graphite from 21 references from the NIST database [1] is 284.46 eV with a standard deviation of 0.14 eV. Note that the well characterized value of 284.5 eV for graphitic carbon is also a strong indicator that this value is not appropriate as a value to use for AdC charge referencing. While these types of samples are generally conductive and if they can be mounted in a manor (in electrical contact with the sample stage) to take advantage of this one should do so. However, many of these types of samples come as a small volume of powders or flakes which are very difficult to mount. Usually, we mount these on a double-sided adhesive which works well but electrically isolates the sample. Oxidation of these types of samples (e.g., graphene oxide) or their functionalization (e.g., functionalized CNTs) can result in them behaving less conductively or as a mixed conductive/insulating material.  Samples where these materials are mixed with other conducting or insulating compounds can also result in a mixed conductive/insulating sample. For most of these types of samples we now electrically isolate the sample and charge reference to C 1s at 284.5 eV for the graphitic (C=C) peak.[2]

Table 1 from [2] presents general fitting parameters for graphitic, graphene and carbon nanotube type materials. These starting fitting parameters include the main peak asymmetry (defined using an asymmetric Lorentzian (LA) line shape) and π to π* shake-up satellite from a pure graphite standard sample. These fitting parameters are similar to the approach taken by Morgan (Fig. 5, Table 2) [3],  Moeini et al. (Table 1) [4],  and Gengenbach et al.[5]  It is always best to run your own standard (pure graphite, graphene, CNT etc.) to get fitting parameters appropriate for your sample type, instrument and conditions used. Slight differences in the main peak asymmetry and differing shake-up satellite position, shape and intensities are possible for differing classes of graphitic materials. See for example from Morgan[3] where HOPG and nano-onion C 1s spectra show peak-shape differences, likely due to hydrogenation of the sample. However, with this caveat stated, the parameters used based on a graphite standard have worked very well for variety of samples (134) analyzed in the five-year data survey from [2]. Figure 1(A) presents the standard graphite spectrum used to obtain the parameters presented in Table 1. The spectra from Figure 1(B, C and D) show the use of these fitting parameters from Table 1 to effectively model a variety of graphitic component containing materials. 

Table 1. General fitting parameters for graphitic/graphene/carbon nanotube type materials. #Line-shape details for CasaXPS. Define asymmetric peak-shape in other software using pure graphite/graphene or CNT sample related your specimens. ##Gaussian/Lorentzian product formula, GL(30) is 30% Lorentzian 70% Gaussian.[2]

Figure 1.  Examples of curve-fitting of graphitic type systems using the parameters from Table 1.  A) pure graphite, B) carbon nanotube-based material modified in caustic solution, C) oxidized graphene and D) acid modified graphene and organic compound mixture.[2]

[1] C.D. Wagner, A.V. Naumkin, A. Kraut-Vass, J.W. Allison, C.J. Powell, J.R.Jr. Rumble, NIST Standard Reference Database 20, Version 3.4 (web version) (http:/ 2003.
[2] M.C. Biesinger, Appl. Surf. Sci. 597 (2022) 153681.
[3] D.J. Morgan, J. Carbon. Res. 7 (2021) 51.
[4] B. Moeini, M.R. Linford, N. Fairley, A. Barlow, P. Cumpson, D. Morgan, V. Fernandez, J. Baltrusaitis. Surf. Interface Anal. 54 (2022) 67.
[5] T.R. Gengenbach, G.H. Major, M.R. Linford, C.D. Easton, J. Vac. Sci. Technol. A, 39 (2021) 013204.