Ambient pressure photoelectron spectroscopy: Practical considerations and experimental frontiers

We first examine work by Yamamoto et al [60, 63] highlighting the value of directly comparing UHV and APXPS measurements of the same solid surface. In this case, the vapor phase is water, and its interaction with three different single crystal surfaces is investigated. Not only is the interaction of water with metal (oxide) surfaces relevant for a variety of technological and environmental chemical processes, but this work clearly demonstrates the utility of in situ XPS measurements with relation to more general concepts of chemical equilibria and wetting phenomena at such surfaces.

The authors first compare the interaction of water with the (1 1 1) and (1 1 0) crystalline facets of metallic Cu; see figure 2 for the relevant O 1s spectra taken at 1 Torr at room temperature, corresponding to a relative humidity of 5%. Figure 2(a) shows the O 1s spectrum of the (1 1 0) surface, with the gas phase water peak near 536 eV binding energy (BE), well-separated by 2–5 eV from the broader peaks of surface species. The wetting behavior of the Cu(1 1 1) and (1 1 0) surfaces is quite different, as shown in figure 2(b). Under the same conditions (295 K and ${{p}_{{{\text{H}}_{\text{2}}}\text{O}}}$  =  1 Torr; 5% relative humidity), no adsorption/dissociation of water is observed on the Cu(1 1 1) surface, while the Cu(1 1 0) is partially covered with a mixture of OH groups and adsorbed H₂O. This experiment gives a direct observation of the difference in activation barriers for water dissociation over the pure metal surfaces, which is higher for Cu(1 1 1) than for Cu(1 1 0).

Zoom In Zoom Out Reset image size

Further analysis of the Cu(1 1 0) surface coverage as a function of temperature (not shown here) reveals two types of OH groups—those which form hydrogen bonds to water molecules and those which do not (OH_pure). There is good agreement between the binding energies of OH_pure, H₂O, and mixed H₂O  +  OH species measured in APXPS and those measured in previous studies at UHV conditions. One notable difference, however, between this work and previous UHV investigations is the absence of adsorbed atomic oxygen in the APXPS study. This absence is simply explained by the different reaction equilibrium conditions that exist under UHV versus elevated pressure conditions for dissociation of water (see equation (1)). In UHV, formation of adsorbed atomic oxygen O_ads and water by reaction of two surface OH_pure sites is kinetically facile. In the presence of water vapor, the equilibrium is shifted in the opposite direction, favoring OH_pure surface groups over atomic O_ads [60, 63].

$$\begin{eqnarray}&&\text{O}{{\text{H}}_{\text{pure}}}+\text{O}{{\text{H}}_{\text{pure}}}\underset{{{p}_{{{\text{H}}_{\text{2}}}\text{O}}}}{\overset{\text{UHV}}{{\rightleftharpoons}}}{{\text{H}}_{\text{2}}}{{\text{O}}_{\left(\text{g}\right)}}+{{\text{O}}_{\text{ads}}}\end{eqnarray} \tag{ 1 }$$

The importance of adsorbed atomic oxygen in the dissociation of water at metal surfaces is further illustrated by the spectra show in figure 2(c). Here, a pre-treatment of the Cu(1 1 1) surface (exposure to 1  ×  10⁻⁶ Torr O₂ at 300 K, followed by annealing at 573 K [60]) leads to atomic O at the Cu(1 1 1) surface, which increases the hydrophilicity of the surface through a decrease of the dissociation barrier for water on O/Cu(1 1 1) compared to pure Cu(1 1 1) and generates a mixed H₂O  +  OH surface layer similar to that observed on Cu(1 1 0). This observation supports one of the main conclusions of this study: the formation of surface hydroxyls precedes the adsorption of water [63].

We now turn attention to the dissociation of water on a metal oxide single crystal surface, namely TiO₂(1 1 0) [63, 64]. These experiments highlight the importance of complementary UHV measurements of the clean sample surface prior to gas introduction. Initial UHV XPS measurements show that the clean TiO₂(1 1 0) surface always contains O-vacancy defects. Upon exposure to water vapor, the coverage of surface OH species is always twice that of the initial surface defect concentration. The O 1s spectrum of the clean surface is subtracted from spectra taken at higher water vapor pressures to calculate difference spectra (figure 3) which more clearly show the evolution of surface water and hydroxyl species as a function of temperature (the O 1s spectra would otherwise be dominated by the signal from TiO₂ lattice oxygen) [63].

Zoom In Zoom Out Reset image size

A combination of measurements collected at either the same temperature (isotherm) or ${{p}_{{{\text{H}}_{\text{2}}}\text{O}}}$ (isobar) provides valuable insight into the process of water dissociation at the TiO₂(1 1 0) surface. As mentioned previously, the coverage of OH on TiO₂ is determined by the surface defect concentration, and thus remains constant at all temperatures (figure 4(a)) and water vapor pressures (figure 4(b)). The coverage of water on the surface increases with increasing relative humidity until it is the same as that of OH (0.25 ML), after which the water coverage increases more rapidly. Note that figure 4(c) shows the coverage data as a function of relative humidity (RH), recalculated from the measured isotherm (figure 4(b)) and isobar (figure 4(a)) data sets; since both data sets overlap when plotted against relative humidity, it can be concluded that kinetic barriers in these experiments are negligible. The data in figure 4 were taken using two different sample holders, either equipped with a resistive heater or a Peltier cooler to measure the full relative humidity range. The Peltier cooler is used to measure at high relative humidity close to saturation, where a traditional cold finger cooling scheme would lead to condensation of water at the cooling lines and not the sample. In all data sets the OH coverage remains constant at 0.25 ML (twice the initial defect coverage) over the entire RH range, while the H₂O coverage increases, with the rate of water absorption sharply increasing above ~90% RH [60, 63].

Zoom In Zoom Out Reset image size

As mentioned before, the 'pressure gap' is the term used to describe the lingering discrepancy between the pressures available to APXPS and the pressures of many relevant technological applications. A recent work by Blomberg et al [51] on a model catalyst system serves as an exemplary case study in what we can learn when we approach this gap and as a reminder that we still have many obstacles to overcome to bridge it successfully.

Figure 5 shows an intensity plot of the O 1s/Pd 3p region as a function of temperature during CO oxidation over Pd(1 0 0) in a 1:1 mix of CO and O₂ at a total pressure of 0.5 Torr. The darker black and red regions indicate low counts (near the background level), while the brighter red and yellow areas indicate higher peak intensities, respectively (see online version for color figures). The continuous series of spectra, collected during a heating ramp followed by a cooling period, show several features: (1) at ~345 °C, the gas and surface species suddenly change from adsorbed CO with CO and O₂ in the gas phase below ~345 °C, to adsorbed O with CO₂ and O₂ in the gas phase above ~345 °C; (2) heating further above this critical 'activation temperature' does not significantly change the spectra; (3) the surface returns to its initial state upon cooling below ~225 °C. We can also see here an example where the core levels of two surface species (O1s of adsorbed CO and Pd 3p) overlap, but the line shapes are sufficiently different so that the two species can be de-convoluted during peak fitting (not shown here).

Zoom In Zoom Out Reset image size

Above 345 °C, the surface catalytic activity is sufficiently high to convert all CO molecules at the surface into CO₂; in addition, only gas phase CO₂ (i.e. the product) is observed near the sample surface. This result is a clear indication that under these experimental conditions, the reaction is mass-transfer limited and could imply that the Pd(1 0 0) surface with chemisorbed O is the active catalytic phase under these conditions. As the authors lay out in their introduction, however, the oxidation of Pd(1 0 0) under oxygen atmospheres is known to lead to surface or bulk oxidation of the Pd depending on the O₂ pressure (particularly at higher pressures approaching those that are technologically relevant for CO oxidation), and that the effect of CO on this oxidation process is still unclear. In these 1:1 CO:O₂ experiments, no evidence of Pd oxidation is found in either O 1s or Pd 3d spectra.

Experiments performed at higher total pressures (up to 1 Torr) at 1:1 CO:O₂ or at higher O₂ partial pressures (1:4 CO:O₂) do show evidence of Pd oxidation (figure 6). Nevertheless, the system still suffers from mass transfer limitations under these conditions, and a higher O₂ partial pressure results in a lower activation temperature. Extrapolation of these results to the likely termination at technological conditions suggests that a surface oxide dominates—a condition which is not captured here in the 1:1 stoichiometric experiments. In fact, the behavior of the solid surface in this work seems to change just at the threshold of pressures accessible to XPS, serving as a poignant reminder that the pressure gap is significant and should be approached with caution.

Zoom In Zoom Out Reset image size

The recent development of APXPS gas cells equipped with gas-impermeable single-layer graphene membranes [37] enables measurements at even higher pressures than those reported here, and exploration of the Pd/CO system with such a cell configuration could enable a direct measurement of the catalytically active species at more technologically-relevant gas pressures. However, there is the logistical requirement in this configuration that the graphene membrane acts as the catalyst support, which limits the sample to particles that can be deposited directly on the membrane and also introduces the possibility that the chemistry will be affected by catalyst/support interactions. While particles or polycrystalline samples may be closer to 'real' catalyst systems, it is not currently possible to measure unsupported or single-crystal surfaces using such a configuration.

Spectroscopic characterization of chemical and electronic states across multiple length scales remains a grand challenge for the surface scientist. Electrochemical systems, such as solid oxide electrochemical cells (SOECs), are a particularly interesting class of materials in this regard, as changes in both chemical and electronic states across short length scales can determine the material properties of interest. We next examine two studies demonstrating the utility of APXPS for operando probing of changes in both chemical and electronic states in such systems.

The thin-film SOEC design used in these works is shown schematically in figure 7. The design of the SOEC is such that the Au/ceria working electrodes and Pt counter electrodes are deposited as thin films on the same side of the yttria-stabilized zirconia (YSZ) solid electrolyte. In this way, all cell components are simultaneously exposed to the gas phase and are easily accessible for APXPS measurements.

Zoom In Zoom Out Reset image size

The distinct advantage of APXPS in studying electrochemical systems is the facile, contactless measurement of local potentials (even at elevated temperatures) across the different solid/solid interfaces in the SOEC—the working electrode is grounded to the APXPS analyzer, and thus the core level photoelectrons measured by APXPS show a shift in kinetic energy (ΔKE) that is related to the local potential (V_L) by ΔKE  =  eV_L. An example of this shift is shown in figure 8, where the apparent binding energy of the Ce 4d signal shifts in the electrochemically active spatial region as the applied bias is changed [68].

Zoom In Zoom Out Reset image size

Use of a 2D area detector during APXPS data acquisition allows for a snapshot of the distribution of local chemical states across the different component interfaces in the SOEC devices [68]. In the case of ceria, changes in the ratio of Ce³⁺/Ce⁴⁺ across the YSZ/ceria interface can be observed and quantified—the spatial region (in this case ~150 µm) over which the Ce oxidation states vary from their equilibrium ratio is indicative of the electrochemically active region. This is confirmed by simultaneous electrochemical current measurements, as well as the core-level binding energy shifts observed in APXPS measurements. Furthermore, chemical intermediates of the solid/vapor surface reactions can be identified and when correlated to the applied bias and local potential of the electrodes, important information about kinetic limitations of the surface reactions can be deduced. Figure 9 shows C 1s spectra collected at a ceria surface in the presence of CO₂ and with changes in the applied bias. Surface carbonate species at both oxidizing and reducing potentials indicates that this is an important surface intermediate in both electrochemical half-reactions. Additionally, differences in the steady-state concentrations of surface species at various applied biases is used to elucidate kinetic information about the electrochemical redox processes [67].

Zoom In Zoom Out Reset image size

These types of measurements are ultimately limited by the spatial resolution (~10 µm) of the detector. Improvements in detector hardware, as well as the option to employ scanning (as in scanning photoemission microscopy, SPEM [69]) or full field imaging (as in photoemission electron microscopy, PEEM [70]), are required to improve the spatial resolution, which is important for deconvolution and identification of surface species at discreet interfaces and across potential gradients in electrochemical systems. The first example of ambient-pressure SPEM was recently demonstrated [71].

The above examples show the utility of APXPS to investigate important phenomena in environmental science, catalysis, and electrochemistry with high surface and chemical, as well as potential, sensitivity. For the study of environmental processes, which are often governed by the presence of water vapor, one can postulate that the pressure gap has essentially been bridged since investigations at up to 100% RH at room temperature (equivalent to ~20 Torr water vapor partial pressure) are now routinely possible; the partial pressure of other important trace gases in the environment is much lower (CO₂ with ~0.4 Torr being the highest). Pressure gap considerations are much more prevalent still in the case of catalytic reactions, where off-line reactor measurements under technologically relevant conditions (>1 bar) and under the low pressure regime used in APXPS need to be performed. If critical reaction parameters, such as yield, selectivity, and conversion are similar in the low and high pressure regimes, this would be a necessary but not sufficient condition for the relevance of APXPS experimental findings for technological applications.

In the next section, we will consider current investigations on a unique class of interfaces with a vapor phase, namely that between a (in most cases) monolayer solid film and a bulk solid substrate. We will discuss this topic with particular attention to the 2D-layer growth mechanisms and the intercalation of gas molecules between the solid layers, which has garnered a lot of attention over the past years due to the opportunities for tuning the chemical properties of the essentially 2D space of the interface.

The growth mechanisms of 2D layers on metal surfaces may take different routes depending on metal substrate, precursor pressure, growth temperature and even the history of the substrate. For Ni layers and single crystals, the graphene growth becomes more complex due to the large solubility of carbon in the bulk Ni and the competing formation of surface carbides, Ni₂C [116]. In this specific case, it is not possible with post-growth characterization methods to determine the contribution from the dissolved carbon and from the CVD precursor. This problem was first approached in situ by Weatherup et al [117] who studied the growth of a graphene layer on polycrystalline Ni foil as a function of exposure to C₂H₂ with depth-resolved APXPS and in situ x-ray diffraction (XRD). Exposing a 600 °C hot Ni foil to an atmosphere of 10⁻⁵ Torr C₂H₂ led to the evolution of the C 1s core level peak as shown in figure 11.

Zoom In Zoom Out Reset image size

In figure 11, the C_A peak arises initially after C₂H₂ exposure and corresponds to adsorbed carbon from the decomposed hydrocarbon precursor. At increased exposure times, this carbon diffuses to the Ni subsurface forming a Ni–C alloy with carbon interstitially dissolved (C_Dis). The latter structure was confirmed by XRD (not shown) verifying that no bulk carbide was formed. After 16 s exposure, the C concentration has reached a level where graphene begins to grow, giving rise to the component C_Gr accompanied by the component C_B, which is attributed to sp³ carbon in defects and at edges at the graphene domains. The inset shows a C 1s spectrum after the precursor gas was evacuated subsequent to a 600 s exposure. The formation of the graphene clearly takes place at the isothermal growth temperature (figure 11(a)), and the C 1s spectrum after cooling (figure 11(b)) shows only a minor increase in the carbon concentration, which is attributed to precipitation from the Ni–C solid solution [117]. The contribution to the graphene growth from subsurface carbon was found to increase at higher growth temperatures and yielded more disordered graphene layers. A lower temperature isothermal growth scheme was thus suggested for achieving higher quality graphene layers. This is an important result for optimizing the growth procedures and could not have been obtained by studying the system with conventional XPS. A similar behavior was observed for graphene grown at various temperatures on a clean Ni(1 1 1) crystal as well as on crystals with various amounts of initial subsurface carbon impurities [118].

During CVD growth of a 2D layer, the typical pressures used vary between approximately 10⁻⁸–10⁻⁵ Torr. At this range, the gas cleanness plays a minor role compared to higher pressure studies. Instead, the drawback of the APXPS technique here is that the APXPS results provide a signature of the graphene quality that is averaged over the spatial range of the x-ray beam spot size and thus usually leaves out information regarding low-density defects and rotational disorder. To obtain detailed information on the atomic scale, we suggest the complementary use of STM or AFM, which can also be carried out at ambient conditions.

The role of subsurface carbon for the graphene-Ni(1 1 1) interface interactions was subsequently further explored using Monte Carlo simulations and APXPS [119]. When a Ni(1 1 1) single crystal sample was heated to 400 °C while exposed to C₂H₄ at various pressures [119], the main growth mechanism was shown to be in-plane transformation of surface carbides, Ni₂C. The formed graphene layer was stable at pressures up to approximately ~10⁻³ Torr. When increasing the pressure to ~10⁻¹ Torr, a large amount of carbon diffused through the graphene defects and boundaries into the Ni crystal. The increase of dissolved carbon led to a weakening of the graphene-Ni interactions, but also to the formation of several graphene layers, which was not observed for exposures at lower pressures at this temperature. In extension of this line of thought, it was shown by time-resolved APXPS (vacuum annealing) that a thin Ni layer placed on a solid carbon source, heated to 600 °C, produced an inhomogeneous, few-layer graphene on the upper Ni surface due to diffusion and precipitation of the carbon through the Ni layer [120]. A more homogenous monolayer graphene could be achieved when Al₂O₃ was introduced between the solid carbon source and the Ni layer [120]. Another approach for achieving a uniform graphene layer on polycrystalline Ni is to decorate the reactive Ni surface sites with Au, thus decreasing the available nucleation sites for the graphene [121]. The above are examples showing that the graphene-Ni interfacial interplay is complex and a small change in growth parameters may alter the final product tremendously.

Cu shows a huge potential for catalyzing graphene growth at industrial scales, and large-scaled growth by a roll-to-roll procedure is already applied [122]. Additionally, the interaction between graphene and Cu is much weaker than the above-mentioned case of the graphene/Ni (gr/Ni) system, allowing the possibility of decoupling the graphene from the Cu by intercalation and transferring it to another substrate [86, 123]. There exists a large body of publications characterizing the gr/Cu system by XPS, often where the samples are grown in a dedicated growth chamber and subsequently transferred through air to the XPS analysis chamber. For these ex situ XPS measurements, the reported XPS results differ, however, from those obtained by APXPS measurements of an in situ grown graphene layer. Kidambi et al [124] reported on the APXPS study of graphene on polycrystalline Cu foil. The low solubility of C in Cu limits the growth to occur mainly by deposited precursors. No carbide is observed and the overall growth mechanism appears to be simpler compared to the case of Ni. A somewhat large charge transfer from Cu to graphene (n-doping) leads to a shift of the sp² graphene signature in the C 1s core level towards higher binding energy compared to other systems. The resulting sp² component was thus found at 284.75 eV [124]. In contrast, the sp² carbon binding energy is at 284.4 eV in some ex situ XPS studies [125]. The difference appears to be due to intercalation of O₂ for air exposed samples, or when O₂ is present in the residual atmosphere in a UHV setup, which consequently weakens the graphene-Cu interactions and thus reduces the effect of the charge transfer, yielding a lower binding energy for the sp² carbon [124]. In such cases, where a contaminant is of high enough concentration to be detected, such as the O₂ intercalation layer, in situ APXPS during growth provides an advantage of pinpointing when during the growth process the contamination occurs.

The advantages and limitations of APXPS also apply to studying the growth of 2D layers other than graphene. The growth of hexagonal boron nitride (h-BN) on polycrystalline Cu was studied by APXPS by Kidambi et al [126]. The growth mechanism was found to be similar to that of graphene on Cu, i.e. isothermal direct growth but at ~975 °C. However, interestingly, boron was observed to dissolve into the Cu subsurface layers forming a precursor for a highly reactive surface when the system was exposed to air and subsequently annealed. When heating was applied to the air exposed sample, the intercalated oxygen reacted with the dissolved boron, creating volatile boron-oxide species which effectively destroyed the h-BN sheet.

Intercalation of species into substrate-2D layer interfaces has received much attention recently due to the prospects of gaining control of electronic, optical, mechanical, and chemical properties of the 2D layers. The mechanism for intercalation depends on the type of 2D layer, substrate, and intercalated species as discussed above, but generally proceeds through defects and grain boundaries. The 2D layer decouples from the substrate through the formation of an interface layer and the energy cost to generate the decoupled 2D sheet must be compensated by energy gained through the adsorption of the intercalated molecule or atom to the substrate.

Some fingerprints of intercalation are easily detected in XPS; for instance, the C 1s core level signature for graphene is usually shifted to lower binding energy upon intercalation, as described in the previous section for gr/Cu. Intercalation of h-BN has also shown a negative shift in binding energy for the N 1s and B 1s core levels. Additionally, the metal surface core level peaks are modified upon adsorption of intercalated molecules or atoms. In many cases it is beneficial to compare the adsorption structure below the 2D layer cover to that of the bare metal surface exposed to the same high-pressure gas using APXPS. A summary of all APXPS intercalation studies of 2D layers on transition metal surfaces, known by the authors to date, is summarized later in table 1. A few cases are discussed in more detail in the following.

In general, it is found in several of the studies discussed below that the adsorption energy of a molecule or atom, when adsorbed on a metal surface below a 2D layer, is decreased as compared to the adsorption on the bare metal surface. This is due to the energy cost required to decouple the 2D sheet from the metal surface, which destabilizes the adsorbate. The adsorption energy reduction is for example observed for H₂ and CO intercalating between graphene and h-BN on Pt(1 1 1) [127–130]. In the case of CO, this effect turns out to be beneficial for an efficient CO oxidation at the metal-2D layer interface.

In 2D layer/metal systems, the coverage of intercalated CO needs to meet a threshold in order to provide sufficient energy to favor the decoupling of the 2D layer. The threshold coverage thus scales with the 2D layer-substrate interaction at the interface and can, for highly interacting systems, be reached by exposure to a higher pressure. Such a threshold is easiest to explore by gradually increasing the pressure until intercalation is observed. For spectroscopic investigation which requires both high pressure and fast data acquisitions, APXPS is a fitting technique. CO was found to intercalate a full layer of graphene on Pt(1 1 1) at pressures above approximately ~0.01 Torr [127], while pressures above ~0.1 Torr CO were needed to intercalate a 0.9 monolayer (ML) graphene layer on Ir(1 1 1) [131]. The CO configuration at the gr/Ir(1 1 1) interface was found to be similar to that on a bare Ir(1 1 1) surface exposed to the same CO pressure [131]. When decreasing the partial CO pressure back to UHV conditions, the CO desorbed from the bare Ir(1 1 1) surface, leaving a stable but lower coverage structure at the surface. This UHV related structure, conversely, did not form in the presence of a graphene cover. The study thereby confirms the existence of a CO pressure gap induced by the interface, which may be utilized for future pressure switching applications [131]. For future studies, it would be interesting to investigate whether similar defect effects, such as the sheet opening upon attachment of hydroxyl groups to the graphene edges [103], are at play. Due to sensitivity limitations, such effects may be nearly impossible to detect using APXPS unless working with very small flakes of graphene where the concentration of edge-sites is increased. However, having a system of small graphene flakes could change the mechanism of intercalation entirely.

A comparison between CO intercalation below gr/Ni(1 1 1) and CO interaction with a Ni₂C surface carbide on Ni(1 1 1) was performed by Wei et al by APXPS [132]. A pressure-independent low coverage of adsorbed CO on Ni₂C/Ni(1 1 1) was found in the ranges 10⁻⁶–0.1 Torr. CO exposures to gr/Ni(1 1 1) led to a partial intercalation at p_CO  >  0.1 Torr and intercalation of a full CO layer at 5 Torr. The complete interlayer of CO led to a decoupling of the graphene from the Ni(1 1 1) surface. While the intercalated CO coverage at saturation was found to be similar to that of a bare Ni(1 1 1) surface, the graphene cover tended to suppress the binding of CO to Ni top sites in favor to the bridge sites. Partial removal of intercalated CO took place rapidly between 80 °C–100 °C. Removal of residual trapped CO could be achieved by annealing the sample to 200 °C [132].

The strongest graphene-metal interaction was observed for gr/Ru [133]. A gr/Ru(0 0 0 1) system was studied by APXPS while exposed to CO [134]. The graphene layer on Ru(0 0 0 1) is highly corrugated and the C 1s splits into two peaks separated by ~0.6 eV, corresponding to graphene areas strongly interacting with Ru (C1) and areas with weaker interaction (C2). The spectrum for the full ML graphene on Ru(0 0 0 1) is shown in figure 12 before CO exposure (bottom spectrum). The C 1s spectrum is complex due to overlapping Ru 3d peaks. Despite the complexity, a new C 1s component (C3) representing decoupled graphene is revealed after exposure to 1 Torr of CO. The system becomes fully intercalated after a prolonged (1 h) CO exposure at a pressure of 10 Torr as shown in the top spectrum of figure 12. Similar to the Ni case, the bridge sites become favorable adsorption sites for the intercalated CO.

Zoom In Zoom Out Reset image size

The partial pressure of CO required for intercalation increases with the interaction strength between the graphene and the transition metal substrate. Since the adsorption energy of CO does not differ significantly for the transition metals of interest, the varying CO pressure necessary for obtaining saturated coverage in different systems may be related to the strength of graphene-metal interaction as also discussed in [134]. The trend is summarized in figure 13.

Zoom In Zoom Out Reset image size

Similarly, intercalation of CO between h-BN and its metal substrate leads to a decoupling of the h-BN layer. CO intercalation was reported for h-BN on Rh(1 1 1) (h-BN/Rh(1 1 1)) [101] and h-BN/Pt(1 1 1) [129]. CO intercalation of gr/Rh(1 1 1) has not been reported in literature, but the interfacial interaction strength is similar to that of gr/Ru [135, 136]. While the graphene-substrate interaction goes as gr/Rh  >  gr/Pt, the interaction of h-BN/Pt(1 1 1) appears to be stronger than that of h-BN/Rh(1 1 1). This statement is based on the APXPS results of the intercalation of CO in the two cases: CO intercalates the full monolayer (ML) at the h-BN/Rh(1 1 1) interface at p_CO 0.01 Torr at room temperature, while p_CO  >  0.1 Torr is needed for intercalation of a full ML h-BN/Pt(1 1 1). The difference between the gr/CO/Pt(1 1 1) and h-BN/CO/Pt(1 1 1) interfaces is additionally reflected in the desorption kinetics, where CO desorbs already at room temperature from the latter system under UHV conditions, while elevated temperatures are needed to desorb CO from the gr/Pt interface [127, 129]. This difference between the behavior of CO intercalation under graphene and h-BN may be explained by the polar nature of the h-BN sheet, yielding a stronger interaction to the Pt than the nonpolar graphene [129]. Removal of CO and a recovery of the as-grown h-BN layer on Rh(1 1 1) occurs only when heating above 350 °C [101], indicating that the interaction of h-BN/Rh(1 1 1) is weaker than that of h-BN/Pt(1 1 1). The h-BN/Rh(1 1 1) system was also exposed to H₂, H₂O, O₂ and a mix of CO and O₂, but only intercalation of CO under the reported conditions was observed [101]. As a side note, H₂ was observed to intercalate under graphene on Ir(1 1 1) as measured by conventional XPS [137] and under graphene and h-BN on Pt(1 1 1) at 0.1 Torr measured by APXPS [130].

The interaction of h-BN with some transition metal surfaces (TMs) goes in the order of Ni  >  Ru  >  Cu  >  Rh as calculated in [138]. From the above discussion the h-BN on Pt(1 1 1) should be placed on the left side of Rh in the scheme. Thus, APXPS provides a direct probe for the qualitative interaction strength between a 2D layer and its substrate through ambient-pressure CO intercalation. It will be interesting to see if CO intercalation on h-BN/TMs systems follows the same qualitative trend as for gr/TMs.

Oxygen intercalation has been widely studied for graphene on metal substrates under UHV conditions [93, 139]. For oxygen intercalation to occur in the low-pressure range, a thermal activation above 225 °C for gr/Ir(1 1 1) [93] is needed. At room temperature, oxygen intercalation was however reported to occur on gr/Cu in ambient environments [140]. An additional parameter to consider when dealing with high pressures of oxygen in the presence of graphene is etching. This process happens at temperatures that depend on the pressure of the oxygen atmosphere, e.g. for graphene flakes on Cu foil: 500 °C at ${{p}_{{{\text{O}}_{\text{2}}}}}$ approximately ~4  ×  10⁻⁵ Torr, 450 °C at ${{p}_{{{\text{O}}_{\text{2}}}}}$ ~4  ×  10⁻² Torr and 280 °C at ${{p}_{{{\text{O}}_{\text{2}}}}}$ ~ 0.2 Torr [140]. Oxygen intercalation was studied by APXPS for graphene flakes on polycrystalline Cu [140] and graphene on Ru(0 0 0 1) [141, 142] and for h-BN on Ru(0 0 0 1) [142] and on Pt(1 1 1) [143]. For gr/Ru(0 0 0 1) the intercalation occurred when heating the system above 150 °C in an oxygen atmosphere of ${{p}_{{{\text{O}}_{\text{2}}}}}$ 0.5 Torr [141], while heating to 200 °C is needed at 0.1 Torr or above 300 °C at a much lower oxygen pressure of 5  ×  10⁻⁸ Torr for 0.5 ML graphene [142]. A bare Ru(0 0 0 1) oxidizes readily at room temperature when exposed to 0.5 Torr oxygen [141]. To reverse the oxidation of a clean Ru(0 0 0 1) substrate, temperatures of around 1000 °C are required. In contrast, oxygen desorbs from the gr/Ru(0 0 0 1) interface already at 420–470 °C [141]. This temperature again confirms that the adsorption energy of the intercalated species below the graphene cover is decreased by the additional energy cost required for decoupling the graphene. For a full ML h-BN on Pt(1 1 1), oxygen intercalation occurs at an oxygen pressure of 0.1 Torr when the sample is heated above 200 °C. Keeping this temperature and returning to UHV does not change the XPS signature for oxygen intercalated h-BN/Pt(1 1 1), as the desorption occurs when annealing the sample above 300 °C [143].

On a final remark, CO intercalation was reported to occur at 10⁻⁶ Torr when the sample was already intercalated by oxygen and kept at 200 °C, while approximately ~0.1 Torr was needed to intercalate an 'empty' h-BN/Pt(1 1 1) interface [143]. This co-intercalation resulted in the removal of intercalated oxygen by CO₂ formation below the h-BN cover, similar to what has been observed for graphene confinement induced reactions [127, 128].

The examples discussed in this section point towards the direction of many of the future investigations in this field of 2D-layer/solid interfaces, where co-adsorption and reaction in the confined space between a 2D material and a metal substrate may lead to new reaction pathways and products. APXPS is an excellent method to study the chemical and electronic properties of the substrate, 2D film, and the adsorbed species as the reaction occurs. A topic which is surprisingly not yet addressed using APXPS is the utilization of 2D materials as thin corrosion protective coatings for metals. For such studies of environmental applications, APXPS is an obvious technique for 'bridging the pressure gap' and accessing changes in the chemical state of the metal below the 2D coating as the system is exposed to harsh corrosive environments. As we have discussed in detail, the limited sensitivity of APXPS in detecting low concentrations of defect states and/or impurities is perhaps the most prominent disadvantage of application of this technique to such 2D layer systems, however this can be compensated for somewhat by judicious application of other complementary in situ surface characterization methods.

We now turn our attention to another condensed phase/vapor interface, namely the liquid/vapor interface, which poses its own challenges for surface science studies.

A liquid jet is formed by forcing a liquid through a small orifice using high pressures (~100 bar [150]). Liquid jets can be formed in the laminar regime but they spontaneously break apart into droplets after a critical length due to Plateau–Rayleigh instability. Because of this critical length, XPS measurements on a liquid jet have to be performed near the capillary exit, where the jet is still cylindrical and smooth.

The constantly refreshed sample volume (and surface) in a liquid jet provides major advantages. Typical jet velocities are in the range of tens of meters per second, and the jet surface is probed within a few mm after exiting the orifice/capillary. Thus, the surfaces are always clean as they are formed less than a millisecond before analysis. Secondly, charging and beam damage are negligible since the surface is only exposed to the x-ray beam for typically a few tens of microseconds. Jets have been used for the investigation of liquid surfaces by Siegbahn in the early days of APXPS, but only a single study using this method was published at that time [17].

In 1997 Faubel et al demonstrated that a liquid water jet of ~10 μm diameter can be produced in a vacuum chamber, maintaining a chamber pressure of better than 10⁻⁵ Torr at a vacuum pump speed of 3000 L s⁻¹ [150]. Their setup allowed ultraviolet photoelectron spectroscopy to be applied to technically challenging liquids such as water (high vapor pressure) and n-nonane (high vapor pressure and extremely low conductivity, σ  <  10⁻¹² Ω⁻¹ cm⁻¹). Liquid microjet setups have also been used by several other groups in the last decade for photoemission or x-ray absorption experiments [151–154].

One disadvantage of liquid jets is the need for a constant supply of the sample liquid (unless it is recirculated), which may be a problem if the liquid is expensive or scarce. Another issue, which is more prominent in non-conducting liquids, is the so-called streaming current. This phenomenon results because charge from the electrical double layer at the capillary/ liquid interface is swept away with the jet. It causes the jet to charge significantly and thus shifts the photoelectron peaks. Faubel et al observed this charge to be spatially uniform, only causing a fixed shift of the peaks and no broadening [150]. Liquid jet chambers are typically pumped to an extent that the pressure is lower than the vapor pressure of the liquid at room temperature. The pumping speed over a liquid jet determines the pressure around it. If the jet has enough time to equilibrate with this vapor, it reaches the equilibrium temperature at this pressure. The temperature of the jet, and thus the liquid under investigation, can be varied by changing the pumping speed and the distance at which the jet surface is measured after leaving the orifice [154].

In atmospheric science, droplet trains have been used to measure the gas uptake of liquid aerosols [155, 156]. A typical setup is very similar to a liquid jet device, only that the orifice is not static but can be vibrated at appropriate frequencies (typically several tens of kHz) to break up the liquid jet into droplets of well-defined size and droplet separation. In 2008, Starr et al demonstrated that a droplet train system can be used in an APXPS experiment to measure the surface propensity of methanol in an aqueous solution. Droplet trains have the advantage over liquid jets that they are stable over longer distances (up to 40 cm for p  <  40 Torr [156] and ~1 m for p  <  20 Torr [59]). This stability allows the exposure time of the liquid to the gas to be varied over a greater time (up to tens of milliseconds) than in the case of the liquid jet (sub-milliseconds). At such timescales, the thermal equilibration between the liquid surface and the vapor is complete, which means the temperature of the droplet surface can be manipulated by changing the background vapor pressure in the chamber [154].

Static droplets can be studied with standard XPS if the vapor pressure of the liquid in question is in the high vacuum range. Ionic liquids are an example for this case [157, 158]. Almost all other liquids have vapor pressures that are incompatible with UHV near room temperature. Static drops have several disadvantages, such as contaminant accumulation on the surface, charging (if the liquid and solid support are not conductive enough), and possibly beam damage. On the positive side, static droplets offer the advantage that liquid surfaces can be measured for hours or days if the equilibrium vapor pressure is maintained in the experimental cell, allowing measurements of slow vapor/liquid reactions. Static droplets can also be prepared in acoustic or optical traps [159, 160] where they are suspended without mechanical support and thus electrical connection, which will in most cases lead to severe charging problems. To date, no APXPS studies on static droplets have been reported.

Liquid layers can be condensed on solid substrates if the pressure of the vapor is near the saturation vapor pressure of its liquid at the temperature of the substrate and the substrate favors adsorption from the gas phase, i.e. in the case of water vapor, a hydrophilic substrate. Liquid water films with thicknesses exceeding a few layers can be grown at above ~90% relative humidity (RH) [161]. At and above 100% RH, condensation and thus layer growth proceeds indefinitely, leading to a constant spatial change of the position of the liquid surface which hampers the APXPS measurements.

Condensation of water on a soluble solid, like an alkali halide salt, creates a saturated solution above the deliquescence relative humidity, (e.g. 75% for NaCl), which can be studied with APXPS [162–166].

Liquid layers with thicknesses in the range of 10–20 nm can be prepared using the so-called 'dip-and-pull' [146] (or 'meniscus' [147]) method. In this method, a solid plate/foil is partly immersed into a liquid reservoir and pulled out slowly (at a rate of ~mm min⁻¹), forming a thin liquid layer. This nm-thick liquid layer typically extends a few cm above the macroscopic meniscus that forms near the liquid reservoir, while the macroscopic meniscus itself extends only a few mm above the liquid. This thin liquid layer is stable in the presence of the near-saturation vapor pressure of the liquid (in the case of water, RH ~100%).

In the pioneering APXPS work on liquids by Siegbahn et al, clean liquid surfaces were created by dynamic wetting of a moving surface, such as that of a wire, disk, or trundle [18, 167–169]. The wire setup had the advantage over the liquid jet technique at the time because the position of the liquid surface could be controlled better. This technique allowed scans over extended periods of time (many hours), which is necessary to study solutes since their concentrations are generally low. The main advantages of the disk and trundle setups is that they enable cooling of liquids under investigation (important for liquids with high vapor pressure), as well as changing the detection angle of electrons in a controlled way for depth-dependent measurements using a (fixed energy) laboratory x-ray source.

A special case for a liquid/vapor interface is that of the liquid-like layer on ice [170]. As on most solids, a premelted layer forms at the ice surface at temperatures close to the melting point. For the investigation of the liquid-like layer, a bulk ice sample has to be grown first, for instance by using a thermoelectric (Peltier) sample holder, as discussed in the previous section [58]. The ice sample has to be maintained at 100% RH, lest it will either continue to grow or evaporate—precise temperature and pressure control are paramount in these experiments.

The simplest method to prepare a thin layer of liquid on a solid is to increase the relative humidity in the analysis chamber to close to saturation (in the case of water, just below 100% RH), where a photoemission signal from the molecular water indicates the condensation of a thin water film (see figure 14). Depending on the thickness of the liquid layer, soft x-rays can probe the liquid/solid interface [163]. While water is the most frequently used liquid in this method, it is possible to use others. Thin films of solutions are prepared by the deliquescence, or rehydration, of salts or molecules that were ex situ deposited onto a substrate [163, 179]. However, the inability to control the concentration or pH of the thin liquid films is a limitation of this technique. Additionally, adventitious carbon, a ubiquitous impurity, often increases in concentration at high humidity. Electrochemical control of the solid/liquid interface requiring reference electrodes in bulk solution is incompatible with this method. Thus, this liquid/solid interface preparation method is limited to a few special cases, such as studying the speciation of ions at the interface of very thin solution layers [179].

A second method for generating a thin film on a substrate involves dipping the sample in a beaker of solution in the analysis chamber and slowly pulling the sample out of the liquid to extend the meniscus of the solution several millimeters along the sample surface, as discussed in the previous section. This technique, schematically depicted in figure 14, is the dip-and-pull or meniscus method [146, 148]. The beaker of solution allows for pH and concentration control. Placing counter and reference electrodes in the beaker offers electrochemical control; shifts in binding energies with changes in applied bias confirm that the thin film is in electrochemical contact with the bulk solution [146, 148]. Depending on the electrochemical reaction being studied, there may be limitations to species diffusion or interference from gas bubble formation. Solution films prepared with this method are between 10 and 30 nm thick, requiring tender x-rays to probe the liquid/solid interface [146, 177]. In addition to electrochemical experiments, pH and concentration dependent studies, hydrated biological systems, membranes, and corrosion processes are possible areas of inquiry for this method.

Passing a solution through a small capillary inside an analysis chamber can create a jet of liquid with a diameter with usually a few tens of µm (see figure 14). As mentioned previously, liquid jets are often used to study liquid/vapor interfaces, but by adding nanoparticles to the solution, the liquid/nanoparticle surface interface can be probed if a sufficient concentration of nanoparticles exists close to the liquid/vapor interface [151, 180]. Liquid jets offer the advantages of concentration and pH control of an impurity-free liquid (even free from adventitious carbon) [181]. Constant refreshing of the sample in the photon beam typically prevents beam damage. Though soft x-rays have been successfully used to probe the liquid/nanoparticle interface [151, 180, 182, 183], hard or tender x-rays would permit the study of nanoparticles that are farther from the liquid/vapor interface [181]. Recently, it has been shown that adding a pinhole to a synchrotron beamline can shape the beam to match the liquid jet diameter, thus reducing the overlapping gas phase signal of the solvent [184]. While no time-resolved studies of liquid jets have yet been conducted using APXPS, there is great potential in this area [181]. If a reaction is on the correct time scale, it could be initiated in the reservoir, gradually change the solution, and be followed with APXPS. One potential disadvantage of using liquid jets is the large amount of sample that is needed, especially nanoparticles that can be particularly expensive [181]. However, a continuous loop system that enables the trapped used solution to reenter the liquid jet reservoir would mitigate this issue.

In the final sample preparation method, the liquid/solid interface is formed at a thin membrane which isolates a liquid in a cell from the analysis chamber. In this strategy, the photons and photoelectrons travel through the membrane layer to probe the interface, illustrated in figure 17; this arrangement eliminates overlapping gas phase peaks, as there is no headspace above the liquid being probed by the x-rays. Solution concentrations and pH can both be accurately controlled. Depending on the cell configuration, the solvent could either be static or flowing behind the membrane, with electrochemical control possible in either situation. A robust, leak-tight liquid cell could be used in standard UHV instruments, greatly increasing the availability of the technique. Difficult samples, such as toxic, reactive, and/or radioactive materials, could be measured. Depending on the membrane strength, liquids and gas at much higher pressures could be accessed to bridge the pressure gap [185].

Zoom In Zoom Out Reset image size

The largest limitation of this method is the membrane material, which must be thin enough to be sufficiently transparent to photoelectrons yet thick enough to hold the pressure differential. Silica [186] and Nafion [187] require hard or tender x-rays [188] while graphene [41, 189] and graphene oxide [189] allow liquid cell studies with soft x-rays. Since the window material is also the solid component of the interface, measurements are currently limited to interface studies of solid materials that can be fabricated into a thin window or deposited onto the windows, such as electrodeposited species [41] or thin films [186]. Depositing nanoparticles or using a solution of nanoparticles to probe the solution/nanoparticle interface is also feasible [188].

Other challenges need to be overcome to further advance this sample preparation technique. Window fabrication (in the case of etched Si) or transfer to the cell (in the case of graphene or graphene oxide) can introduce impurities [185]. Especially for static solutions, beam damage can lead to bubble formation, causing interference with the photoelectron signal and possibly window bursting [185]. Kraus et al are developing a microchannel plate with many graphene windows to reduce the solution area exposed to the photon beam and prevent signal loss if a few windows burst [185]. Depending on the system studied, spectral features from the window material could interfere with peaks of interest, or the window material may participate in chemical reactions [186, 188]. Though this sample preparation technique is still the least developed, it holds great promise to expand APXPS to a new set of users without differentially pumped spectrometers and to a wide variety of traditionally difficult-to-probe samples.

By using a solution of nanoparticles in a liquid jet, the solid/liquid interface of the particles can be probed. Such a study was performed by Brown et al [183] to investigate directly the effect of pH on the protonation and deprotonation of silanol (Si–OH) groups. In general, there is a pH for each metal oxide in aqueous solution where the charges of acidic sites are balanced, known as the point of zero charge (pH_PZC). Above this pH, acidic sites are deprotonated (M-O⁻, where M  =  metal); below the pH_PZC, protonated ($\text{M-OH}_{\text{2}}^{+}$) acidic sites dominate the surface. Therefore, the pH of an aqueous solution changes the surface potential of a metal oxide.

The Si 2p spectra of an aqueous solution of 9 nm diameter SiO₂ nanoparticles at pH 10 and pH 0.3 are shown in figure 18. The large component in the spectra is assigned to the bulk Si atoms and neutral Si–OH surface groups. The change in surface potential between these two pH conditions results in a 300 meV shift of this main component. Deprotonation of the silanol groups generates a minority component at lower binding energies in basic conditions (see figure 18(a)). The protonation of the silanol groups at the acidic pH shifts the minor component to higher binding energies in figure 18(b). The minor component at both pH values is about 13% of the total intensity. These assignments were aided by density functional theory and are the first direct in situ spectroscopic evidence for deprotonated silanol groups. Based on the shift in the main peak and the known relationship between pH and surface potential [190], 30% of the silanol groups are calculated to be protonated under acidic conditions or deprotonated under basic conditions. This estimation of population is not available from ²⁹Si NMR data where overlapping peaks prevent species quantification [183].

Zoom In Zoom Out Reset image size

This example highlights a benefit of fast exchange of the liquid phase in the liquid jet, namely, the ability to maintain a constant pH and solution concentration at the point where photoelectrons are generated in the sample. Using this advantage, the surface potential of the nanoparticles is modified by a change in the solution pH. For nanoparticles in a liquid jet, it is not possible to directly apply an external potential to the particles. The next example correlates binding energy shifts in spectra with changes in applied potential to the sample.

Semiconductor/liquid interfaces behave similarly to a semiconductor/metal interface, where a charge separation region in the semiconductor originates from the electronic double layer in the liquid (i.e. band bending). The resulting electric field in the semiconductor influences the core level binding energies. APXPS with tender x-rays has recently been used to probe the energy at the interface between water and an electrode of a photoelectrochemical cell composed of a layer of TiO₂ on boron-doped p-type silicon substrate (p⁺-SiO₂/TiO₂ electrode) [148]. The substrate and the energy levels of each sample component are illustrated in figures 19(a) and (b), respectively. TiO₂ core level binding energies are dependent on the band bending that results from the liquid double layer, and the liquid binding energies shift with an applied potential, U.

Zoom In Zoom Out Reset image size

Using the dip-and-pull method shown in figure 14, a thin layer of aqueous electrolyte was grown on the p⁺-SiO₂/TiO₂ electrode, and core level spectra were collected while applying a potential ranging from  −1.2 V to  +0.4 V. The relative binding energy shift of the O 1s peak of the liquid H₂O (from the 1 M KOH electrolyte) and the O 1s and Ti 2p peaks of the TiO₂ are plotted in figure 20. As expected from figure 19(b), the water O 1s binding energy shifts with U. The binding energies of the TiO₂ have a more complex behavior that can be divided into four regions, as shown in figures 20 and 21. Applied potentials in region U₁, the most negative, result in no shift of the TiO₂ peaks because surface states of the conduction band are filled with electrons and pin the E_F to the conduction band minimum (see figure 21(a)). Standard semiconductor behavior occurs at potentials between  −1.0 and  −0.6 V (region U₂), where band bending shifts the TiO₂ peaks. Upon Fermi level pinning of the conduction band in Region U₃, the peak positions again remain constant. In Region U₄ the standard semiconductor behavior returns with shifts in the TiO₂ peaks. Figure 21 illustrates how the energy levels of the water double layer and the applied potential affect the E_F and conduction band of TiO₂.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The previous study exemplifies the application of APXPS to investigate fundamentals of energy level alignment at liquid/solid interfaces and demonstrates potential control of the thin liquid layer. The dip-and-pull method has also been recently applied to the study of nickel–iron oxyhydroxides that function as water oxidation electrocatalysts [191]. In this case, an important limitation of the dip-and-pull method with regard to certain electrochemical systems becomes apparent—it is possible to measure the chemical nature of these catalysts only at low current densities (i.e. far from some of the technologically relevant conditions). The geometry of the thin electrolyte film in the dip-and-pull method is such that ohmic losses through the liquid film are large and diffusion of ions into and out of the liquid film is slow. The result is a large potential drop and consumption of –OH ions in the liquid layer, the latter of which changes the local pH in the thin electrolyte film. Changes in the electrochemical cell design to decrease ohmic losses and facilitate ion transport, while still allowing for efficient collection of photoelectrons and escape of gas-phase reaction products, is a persistent challenge to investigating these electrocatalysts under technologically relevant conditions [191].

Regardless of the sample preparation method, photoelectrons from the narrow liquid/solid interface are emitted simultaneously with those from both bulk phases and are convoluted with depth-dependent signal attenuation. Depth profiling by changing the photon energy or emission angle only slightly enhances the interface signal, and quantitative information about the interface is difficult to obtain. Sub-nm resolution of chemical species at the liquid/solid interface has been recently obtained by combining standing-wave photoemission spectroscopy [192–194] with APXPS [147, 179]. In this technique, a substrate with a buried multilayer mirror (figure 22) generates a standing wave by interference of the incoming and reflected x-rays. The standing wave propagates perpendicular to the interface and acts as a ruler for the distribution of chemical species. Photoemission spectra are collected as the sample angle is scanned through the first order Bragg angle. The spectral intensity of the chemical species depends on the depth distribution above the multilayer mirror, and plotting this intensity versus angle generates plots known as rocking curves. Simulations of chemical speciation as a function of depth are used to fit the rocking curves and establish the depth distribution of all species below, in, and above the interface. This technique was pioneered examining solid/solid interfaces [192–194].

Zoom In Zoom Out Reset image size

An example of standing-wave ambient pressure photoemission spectroscopy (SWAPPS) measures the swelling of a Ni surface to a Ni oxyhydroxide in basic conditions under a  +0.6 V potential [147]. A layer of Ni (~8 nm) was deposited on the SiO₂ capping layer of the mirror substrate. The Ni 2p_3/2 spectrum of the prepared substrate in UHV ('dry' in figure 23) shows that the Ni is largely metallic. However, overlapping broad peaks in the O 1s spectrum in figure 23 indicate the presence of a thin oxide/hydroxide layer. The rocking curves of the metallic Ni 2p_3/2 and oxide O 1s intensity are plotted in figure 23. Iterative fits of the rocking curves (solid line in figure 24) reveal that 1.8 nm of the surface is oxidized, depicted in the diagram in figure 25.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In a water vapor pressure of 17 Torr (85% relative humidity, 'humid' spectra in figure 23), the Ni is still largely metallic with a thin oxide/hydroxide layer. The water vapor peak prominently appears in the O 1s spectrum at 536 eV (figure 23), and a contribution from adsorbed liquid water on the surface broadens the oxide/hydroxide signal. The sample was further oxidized by immersion in a 0.1 M KOH solution with cyclic voltammetry cycles between  +0.65 and –0.9 V. Using the meniscus method (figure 14), spectra were collected in the presence of a 20 nm layer of KOH at  +0.6 V bias ('immersed' in figure 23). Characteristic of Ni oxide/hydroxide/oxyhydroxide, the Ni 2p_3/2 binding energy increased to 856 eV. The liquid water signal is discernable in the O 1s spectrum in figure 23. Rocking curves of the Ni 3p and liquid O 1s signal are shown in figure 24 and were used to calculate the thicknesses of the sample components, shown in figure 25. The 8 nm Ni film expanded to a 27 nm oxide/hydroxide layer. With the standing wave used as an intrinsic ruler, the SWAPPS technique provides sub-nm description of species at solid/ liquid interfaces.

Despite the strengths of this technique, SWAPPS suffers from two distinct limitations. First, it is necessary to collect multiple core level spectra at multiple different sample angles. Combining this requirement with the higher energy x-rays required to probe the liquid/solid interface and the signal attenuation by the liquid and vapor phases results in longer acquisition times for each spectrum and thus very long total experiment times (on the order of several hours per sample condition) when compared to typical APXPS investigations. Maintaining stability of the liquid layer for such a long period of time is challenging, and the effects of beam-induced sample damage must be monitored closely. Secondly, a multi-layer x-ray mirror substrate is required for generation of the standing wave. The sample preparation steps (e.g. deposition, annealing, electrochemical cleaning) must not alter the substrate in such a way as to destroy the standing wave effect, which greatly limits the types of samples that can be studied using this method.

In summary, the discussion in this section shows the obstacles to the measurement of liquid/solid interfaces, mainly given by the necessity to prepare either very thin solid or liquid layers that permit detection of photoelectrons from the interfacial region. In addition, the enhancement of the interfacial signal over that originating from the adjacent bulk phases is a persistent challenge.