Chapter 8.8. Surfaces and interfaces

For flat surfaces, polarization-dependent NEXAFS and EXAFS analyses give fruitful information on the orientations and adsorption sites of surface chemical species, as discussed in Section 2 of Iwasawa et al. (2024). The adsorption sites of sulfur and carbon on nickel single-crystal surfaces have been identified by polarization-dependent EXAFS (Brennan et al., 1981; Ohta et al., 1986). Surface EXAFS has revealed the structure of the adsorption-induced surface reconstruction of p4gm(2×2) N on Ni(100) and (×2) S on Ni(111) (Kitajima et al., 1989; Wenzel et al., 1987). NEXAFS spectra at the C, N and O edges have provided valuable information about the structure and orientation with respect to the substrate of small molecules containing these atoms. In fact, two peaks appear in NEXAFS spectra that are assignable to σ* and π* orbitals (Stöhr, 1992). When the bonding direction and polarization direction are parallel to each other the σ* peak is enhanced, while the π* peak is enhanced when the interatomic bond is perpendicular to the X-ray polarization. Another application of NEXAFS obtains structural information from the photon energy of spectral features. The bond length and σ* resonance position are correlated theoretically as follows (Natoli, 1983), where Δ_σ is the σ* resonance position relative to the vacuum level, R is the interatomic distance in the adsorbate and V₀ is an inner potential. Sette and coworkers proposed a simpler relation in which equation (1) is approximated by a linear function over a limited range (Sette et al., 1984a,b),

NEXAFS is also applied to the analysis of polymeric materials, especially copolymer films, which contain a domain structure. Therefore, spatially resolved NEXAFS, such as scanning transmission X-ray microscopy (STXM), provides the distribution and location of the polymer domains of thin films (Ade et al., 1997). Meanwhile, NEXAFS spectra measured using photoemission electron microscopy (PEEM) can be applied to analysis of the surface structure of thick solids (Swiech et al., 1997). The structure of densely packed hexanethiolate self-assembled monolayers (SAMs) on Cu(100) has been studied by XAFS spectroscopy. The sulfur adsorption site was determined to occupy the fourfold hollow site of the unreconstructed Cu(100) surface, with a nearest-neighbour S–S distance of 3.6 Å. The mismatch between the interchain distance (4.5 Å) and the copper lattice is increased by the internal degree of freedom of the S—C bond (Kondoh et al., 2001). The three-dimensional structures of surface organometallic molecules containing copper, nickel, gold, platinum and other metals may be determined by polarization-dependent total reflection fluorescence XAFS (PTRF-XAFS), which allows high-sensitivity XAFS analysis of around 0.01 monolayer (ML) of such heavy elements on a surface (Koike et al., 2006). Metal–thiophene and metal–thiol interactions have systematically been studied using PTRF-XAFS, and general rules for stabilizing the metal species on the surface have been obtained (Asakura et al., 2013; Chun et al., 2007; Takakusagi, Chun et al., 2013; Takakusagi et al., 2016; Takakusagi, Nojima et al., 2013; Lu et al., 2021).

Total reflection EXAFS has been applied to the analysis of electrochemically deposited metal films on electrodes. Owing to its high penetration depth, one can carry out in situ EXAFS studies in the presence of water under electrochemical conditions. Iodine on Pt(111), lead on Au(111) and copper and platinum on Au(111) have been studied under in situ total reflection conditions (Blum et al., 1986; Gordon et al., 1986; Melroy et al., 1988; Samant et al., 1987; Yuan et al., 2018; Wakisaka et al., 2020, 2021). Polarization-dependent EXAFS has identified the active species in the hydrogen-evolution reaction catalysed by platinum species on a viologen-modified silicon surface as molecularly dispersed platinum complexes (Masuda et al., 2012).

Time-resolved NEXAFS measurements of 2D flat surfaces using the dispersive mode are explained in Section 3 of Iwasawa et al. (2024). In this section, some examples of the application of time-resolved dispersive NEXAFS to mechanistic studies of catalytic surface reactions are described. Because the dispersive NEXAFS technique requires chemical uniformity of the irradiated area on a 2D flat surface, single-crystal surfaces have often been used as model catalysts.

The first example is a kinetic study concerning water formation on Pt(111) using dispersive NEXAFS measurements and kinetic Monte Carlo simulations (Nagasaka et al., 2005, 2008). Fig. 1 shows coverage changes during the water-formation reaction on an oxygen-precovered Pt(111) surface obtained from spectral features of the dispersive NEXAFS measurements, in which spectral contributions from O, OH and H₂O can be distinguished. The initial induction period and the behaviour of the OH intermediate can be well reproduced by Monte Carlo simulations based on an autocatalytic cycle of elementary steps and water diffusion to the reaction front assisted by proton transfer in H₂O + OH networks formed on the Pt(111) surface under the reaction conditions.

Figure 1 Coverage changes during the water-formation reaction on Pt(111) obtained from dispersive NEXAFS measurements (Nagasaka et al., 2005).

The second example is the spectroscopic identification of a reaction precursor produced during the course of NO reduction on Rh(111) (Nakai et al., 2009). Although the reaction pathway N_ad + NO_ad → N₂O_g was initially assumed for this system, the anomalous observed reaction kinetics could not be explained by this pathway. Fig. 2 shows a dispersive NEXAFS spectrum obtained for a nitrogen-precovered Rh(111) surface under exposure to gaseous NO. A new spectral component appears between the components associated with N_ad and NO_ad. This new component is attributed to an NO dimer (NO)₂. The NO dimer is formed from gaseous NO and acts as a precursor in the reaction pathway N_ad + (NO)₂ → N₂O_g + NO_ad, as illustrated at the bottom of Fig. 2. This precursor-mediated reaction mechanism reconciles with the anomaly in the observed kinetics.

Figure 2 Dispersive NEXAFS spectrum for N-covered Rh(111) under NO-reduction reaction conditions (top) and a schematic illustration of the precursor-mediated reaction mechanism (bottom) (Nakai et al., 2009).

In the past few decades, the polymer electrolyte fuel cell (PEFC) has attracted much attention as one of the most efficient clean energy-generation systems, bringing low or even zero emissions into reality, and is suitable for automotive applications owing to the high power density at low temperatures. The application of time-resolved and spatially resolved XAFS to PEFCs has been reported for cathode electrocatalysts. The structural kinetics of Pt/C and Pt₃M/C (M = Co and Ni) during PEFC voltage-cycling processes have been systematically investigated by time-resolved XAFS, as shown in Fig. 3 (Ishiguro et al., 2014, 2016; Ozawa et al., 2018; Tan et al., 2019; Ichihashi et al., 2020; Samjeské et al., 2020; Wan et al., 2021). The rate constants for the redox reactions at the surface of the cathode electrocatalysts were successfully determined by in situ time-resolved XANES and EXAFS analyses, and the effects of alloying with platinum to improve catalyst durability were discussed from a mechanistic viewpoint.

Figure 3 (a) A series of in situ time-resolved Pt L3-edge EXAFS Fourier transforms for the voltage-cycling process from 0.4 to 1.0 V. (b) Parameters obtained from analysis of the in situ time-resolved XAFS spectra of Pt3Co/C cathode electrocatalysts (Ishiguro et al., 2014). Red, 0.4→1.0 V; blue, 1.0→0.4 V.

Three-dimensional visualization of cathode catalyst layers in practical membrane electrode assemblies (MEAs) has been investigated by X-ray computed laminography–XAFS (CL-XAFS) and computed tomography–XAFS (CT-XAFS) (Saida et al., 2012; Matsui et al., 2017, 2019, 2020; Higashi et al., 2020). Fig. 4 shows an X-ray CL image of a cathode catalyst layer in a used MEA after an accelerated durability test (Saida et al., 2012). The reconstructed CL image observed using an X-ray energy below the Pt L₃ edge clearly shows the morphology of the cathode catalyst layer (Fig. 4a). The three-dimensional distribution of the Pt cathode electrocatalyst in the cathode catalyst layer was successfully visualized by calculation of the Pt L₃ edge jump from CL-XAFS images (Fig. 4b). Spatially resolved XAFS is a promising technique to visualize the distribution and chemical states of each element in a heterogeneous sample in a nondestructive manner.

Figure 4 (a) A reconstructed computed laminography image of a cathode catalyst layer in a used MEA and (b) the three-dimensional distribution of the Pt cathode electrocatalyst in the cathode catalyst layer in (a) (Saida et al., 2012; Tada et al., 2015).

Further improvements in the oxygen-reduction reaction (ORR) activity and durability of cathode electrocatalysts, which will reduce the cost of PEFC stacking, require in situ/operando techniques that can directly characterize cathode electrocatalysts in the MEA of PEFCs and illuminate fundamental issues hindering the development of next-generation PEFCs (Iwasawa et al., 2016; Nagamatsu et al., 2016; Tada et al., 2015). Fig. 5 shows the surface structures of Pd–Pt nanoparticles in a Pd(core)–Pt(1 ML shell)/C cathode catalyst at 0.4 and 1.4 V determined by in situ XAFS analysis. The 4.5 nm-sized Pd(core)–Pt(1 ML shell) nanoparticles were regarded as consisting of ten layers because the interplanar distances of the assumed Pt(111) and Pd(111) planes were 2.26 and 2.24 Å, respectively. Hence, the coordination numbers (CNs) of Pt–Pt [CN(Pt–Pt)] and Pt–Pd [CN(Pt–Pd)] around the Pt atom expected from the Pd(core)–Pt(1 ML shell) should be 6 and 3, respectively, assuming an f.c.c.(111) arrangement. The observed CNs of Pt–Pt and Pt–Pd at 0.4 V were 6.6 (±1.3) and 3.3 (±1.1), respectively, i.e. almost the same as the expected values. The bond distances R(Pt—Pt) and R(Pt—Pd) were determined to be 2.71–2.72 Å and were similar to each other, indicating a compressive strain of the Pt shell layer. By increasing the potential from 0.4 to 1.4 V, the Pt shell was oxidized to form Pt—O bonds at 2.00 (±0.07) Å with a CN(Pt–O) of 1.8 (±6.4). CN(Pt–Pt) and CN(Pt–Pd) were determined to be 5.2 (±5.4) and 2.3 (±3.4), respectively, similar to the values of 6 and 3 expected from the Pd(core)–Pt(1 ML shell) structure. These structural parameters indicate retention of the core shell structure and the simple adsorption of O atoms on the Pt 1 ML surface. However, the observed CN values were a little smaller than the expected values and the error ranges in the CN and R values were rather large. Hence, the Pt 1 ML shell may have become a disordered monolayer by making strong Pt—O bonds. Decreasing the potential from 1.4 to 0.4 V reproduced the initial structural parameters at 0.4 V (Nagamatsu et al., 2016).

Figure 5 (a) Ex situ STEM image (×150 000) and EDS line profiles (Pt and Pd) and (b) potential-dependent structural transformations of a Pd(core)–Pt(1 ML shell)/C catalyst in an MEA at 0.4 and 1.4 V (versus the reversible hydrogen electrode) determined by in situ XAFS (Nagamatsu et al., 2016).