XPS insights: Sample degradation in X‐ray photoelectron spectroscopy

Standard X‐ray photoelectron spectroscopy (XPS) analysis is thought of by many as a non‐destructive form of analysis; however, both the interaction of the X‐ray photons and the subsequent electron cascade can cause significant changes to the analysed area. This XPS Insights paper gives a brief overview to this phenomenon, supported by specific examples and experimental advice to assess and minimise damage during analysis.


| INTRODUCTION: CAUSES AND SOURCES OF DAMAGE
Damage typically occurs from the interaction of primary or secondary radiation sources, such as photons and electrons, with a sample. The damage may not be limited to the surface, with the effect of the damage propagating throughout the bulk of the material due to the penetration depth of the X-ray photon and the subsequent cascade of high-energy photoelectrons and thermalisation processes. Figure 1 shows a simple representation of some of the damage that may occur; a more exhaustive list can be found in Baer et al. 1 The potential for damage in recorded spectra should not be ignored, and analysts should familiarise themselves with ISO Standard 18554:2016 -'Surface chemical analysis -Electron spectroscopies -Procedures for identifying, estimating and correcting for unintended degradation by X-rays in a material undergoing analysis by X-ray photoelectron spectroscopy' for a simple procedure to estimate and correct for degradation. Note, however, the standard does not address comparisons between sample types and neither does it address degradation mechanisms.
In the flowing sections of this XPS Insight, a brief overview of sample damage for different sample types during XPS analysis is presented and strategies for minimisation of spectral artefacts from such damage.

| Polymers, organics and 'soft matter'
Polymers tend to undergo degradation through loss of volatile compounds such as HCl in the case of chlorinated polymers, 2-10 or through crosslinking of unsaturated functions or bond scission. 11,12 In contrast, phenyl-based polymers such as polystyrene and PET tend to be stable under X-ray analysis which is attributed to the delocalised electrons acting as a 'sponge' and absorbing excess energy. 2 Beamson and Briggs introduced the polymer degradation index 2 as a measure of how different polymers/classes of polymer degrade over a period of 500 min by following the change in the X/C ratio, where X is typically a parameter characteristic of the polymer obtainable from the photoelectron spectrum, such as the area of Cl, F, O and so forth. This index, whilst useful in understanding degradation rates, is limited in that modern spectrometers, especially those with micro-focussed sources, have a significant impact on the rate of degradation. This effect has been shown by the current author, with the degradation rate for one micro-focussed XPS system doubling with a halving of the spot size, in agreement with that of Cazaux. [13][14][15] Such degradation can readily be seen in polyvinyl chloride (PVC), which is typically used as the exemplar for polymer degradation as it undergoes rapid dechlorination forming a network of polyenes. Figure 2 shows the change in Cl/C ratio for PVC as measured on two different monochromatic XPS systems and clearly illustrates the marked difference between a micro-focussed XPS system and that of varying spot size. Clearly, the shape of the plots between the two systems is very different, with the broad spot source having a linear profile, whilst the other has a more exponential decay, the linear region being confined to approximately the first 30 min.
Given the increased use of micro-focussed XPS systems, the Beamson and Briggs degradation index is perhaps somewhat dated, given the long degradation times specified and the specific instrument. Instead, a Photon Threshold Index (PTI) for a system is recommended whereby all polymers are scaled relative to the reduction of PVC measured on that particular instrument. 15,16 Nevertheless, irrespective of the system, given the propensity for degradation, rapid acquisition of the significant core levels for polymeric materials is preferrable to minimise spectral artefacts. It is logical to assume those which are non-conducting (the majority of materials) and those with weak bonds may be especially susceptible to reduction, as discussed by Cazaux. 14 Süzer has shown the reduction of different metal ions deposited on Au and Si wafers and related their reducibility to the electrochemical reduction potential of the metal ion, 17 whilst Auger decay mechanisms, Coulomb explosion or decomposition of ligands via photoreduction have all been identified as decomposition routes. 18 Irrespective of the decay mechanism, which is outside the scope of this insight note, the possibility that a sample may be susceptible to reduction during analysis should always be assumed. Figure 3 shows just some of the samples that can undergo reduction as measured on different XPS systems in the author's laboratory. A nonexhaustive list of other metal oxides and salts which have been reported to degrade under analysis include Au, 19 Pd, 20 Cu, [21][22][23] Re, 24  It is logical to assume also that rates of reduction may be different for nanoparticulate materials given their size and surface-to-volume ratio strongly affecting their surface properties and overlayer thicknesses. 37,38 F I G U R E 1 Types of damage that may occur from the interaction of the incoming X-ray photon and generated electron cascade in a material during X-ray photoelectron spectroscopy (XPS) analysis. Adapted from Baer et al. 1 F I G U R E 2 Plot of Cl/C ratio against time for physical vapor deposition (PVD) degradation on two different X-ray photoelectron spectroscopy (XPS) systems. Also shown is the data for a smaller spot size on the micro-focussed XPS system (data adapted from Morgan and Uthayasekaran 15 ).

| Thin films and self-assembled monolayers
Insulating thin film and related materials can be extremely susceptible to analysis-induced damage. Such damage arises from the electrons generated from the substrate being injected into the film overlayer, as discussed theoretically by Cazaux. 39 Such an effect has been demonstrated for PVA films on Si, Cu and Au substrates, with greater damage observed for the metallic substrates, which were correlated with photoelectron yield and substrate secondary electron emission 40 ; similar dependencies for SAMs on Si and Au substrates have been noted by Laibinis et al. 41 Again, in such analysis, the role of both X-ray and neutralising sources should be considered for possible enhancement of damage in addition to electron injection from the substrate.

| INSTRUMENTAL EFFECTS AND MINIMISING DAMAGE
Given the potentially damaging nature of XPS analysis, how do we minimise such damage?
First, it is important to note that all XPS systems operate differently and that the operating parameters will also vary and affect the quality of the data recorded. Therefore, it is beneficial for the system operator to understand factors that influence reduction, which can be broadly broken down in to (i) instrumental factors which the analyst has at least some control (e.g., X-ray power and sample mounting) or no control over (e.g., sample type and source type) and (ii) sample dependent factors which are touched upon below. To minimise damage, we should start with the view that the sample to analyse will degrade in some way and adjust our analysis methodology to mitigate artefacts in our spectra. The analyst should first ask the following questions: 1. Are there reducible metal states in the sample/is the sample polymeric?
2. Have they observed sensitivity of similar samples to the charge compensation source?
3. Have they observed increased reduction from micro-focussed Xrays?
Answers to such questions may preclude the use of a particular XPS system or, more likely, allow for adjustment of experimental procedures. For example, should we have an elemental state, such as Cu (II), which is known to reduce, the experimental procedure should be adapted so that it is collected as the first and last regions to check levels of reduction. Alternatively, the Cu (2p) region could be collected as a series of single scans, so each spectrum can be retrospectively checked for damage and the 'undamaged' spectra summed for quantification purposes.
The approach the analyst uses will be both sample and system dependent, but sometimes there is no escape from collecting data over shorter time frames and accepting a lower signal-to-noise ratio in the spectrum or collecting at higher pass energies (e.g., 40 eV rather than 20 eV) to increase the count rate.
Irrespective of these points, overall minimisation of X-ray exposure during analysis is preferrable. As a case in point, during the operation of the EPSRC national facility for photoelectron spectroscopy, we often have users new to surface analysis and regularly give some form of introduction to the technique, usually with samples in the system. Of course, the use of the visitors' samples for this discussion could lead to significant sample damage as the sample is continuously irradiated. Therefore, we tend to use what we consider to be stable, sacrificial samples, such as TiO 2 , Kapton or PET, depending on the users' samples, as exemplars to aid discussion.

| SUMMARY
This XPS Insight has sought to give a brief oversight on sample degradation during XPS analysis. Whilst usually considered in the analysis of polymers, it is often overlooked for inorganic materials leading to misleading or erroneous conclusions from XPS data. It is hoped that analysts will adopt and use the data and strategies presented herein to understand their own spectrometer and ultimately yield spectra with minimal artefacts, hence representative of the samples studied.