Chapter 3.22. Surfaces

X-ray absorption fine-structure (XAFS) analysis techniques provide averaged information on the structures and electronic states of both the surface and the bulk of a sample. XAFS information is also averaged temporally and spatially during the measurements. Nevertheless, XAFS can provide surface information on a variety of samples such as supported catalysts, nanomaterials, flat crystal planes, thin films etc. by analysing the transmitted X-rays, fluorescence X-rays, total reflection fluorescence X-rays, emitted electrons from the surface and so on. Since X-rays can penetrate deeply into materials, XAFS is in principle a bulk detection technique. The electron-yield method can increase the surface sensitivity: the escape depth of Auger electrons from the material is a few nanometres so that the surface sensitivity of XAFS is much increased (type I method; Citrin et al., 1978; Stöhr, 1978; Stöhr et al., 1978), which was originally proposed theoretically (Lee, 1976). However, since Auger electrons are interrupted by the photoelectron peaks, a partial electron-yield method can be used in which a retarding bias is applied to the mesh in front of the electron detector to reject the low-energy secondary electrons, increase the signal-to-background ratio and reduce the interference of the photoelectrons (Stöhr et al., 1978). To obtain spectra with high signal-to-background ratio, the fluorescence-yield method can be applied when the same elements are not present in the bulk (Fischer et al., 1986; Stöhr et al., 1985). The photostimulated desorbed ion-yield technique has the potential to provide a real topmost surface structure (Jaeger et al., 1981), although it is rarely used because of its poor dependence on the absorption coefficient and the low desorption probability.

The other way (type II) to obtain surface XAFS signals is to apply the XAFS technique to nanoparticles, which have a large surface area, and thus their surface fractions are large compared with bulk atoms (Lytle et al., 1977). Nanoparticles are used as catalytic materials, so EXAFS is being applied to the characterization of catalyst structures in an exponential manner. Therefore, we will review time-resolved, spatially resolved and in situ XAFS methods at a later stage.

The last, but not the least, method (type III) to increase the surface sensitivity is to use a glancing-angle incidence or a total reflection phenomenon (Martens et al., 1978). It is more effective for the surface EXAFS of high-Z elements, where high-energy X-rays can deeply penetrate into the bulk to increase the background scattering of X-rays from the bulk. The reflected beam contains XAFS information after the Kramers–Kronig transformation (Martens & Rabe, 1980). In combination with the fluorescence method, one may obtain EXAFS spectra of adsorbates with less than one monolayer (Chun et al., 2007; Heald et al., 1984). The fluorescence detection method allows us to observe the structure under reaction conditions (Asakura et al., 1997; Tanizawa et al., 2003). When the substrate is a low-Z element such as carbon, the critical angle is less than 1 mrad and it is hard to adjust it precisely. In this case the back-illuminated method is effective, in which the low-Z substrate is used as the window (Uehara et al., 2014).

For the characterization of flat surfaces, the polarization-dependent technique given by equations (1) and (2) for EXAFS oscillations can be used (Asakura, 2012),for K and L₁ edges and for L₂ and L₃ edges, where χ(k), χ_i(k), ɛ, r_i and θ_i are the total EXAFS oscillation, the EXAFS oscillation of the ith bond, the unit polarization vector and the ith bond unit vector and the angle between them, respectively. Analysis of the polarization dependence gives the orientation and adsorption sites of the adsorbates (Stöhr, 1988, 1992).

The next section describes developments in surface near-edge X-ray absorption fine structure (NEXAFS) under reaction conditions to follow reaction processes on a flat substrate surface.

XAFS in the near-edge region is called NEXAFS. NEXAFS of molecules containing low-Z elements (C, N and O) shows the characteristic peaks assigned to π* and σ* orbitals. The separation and intensity ratio of the two peaks give the bond length and information on the the orientation of the adsorbate, respectively (Stöhr, 1992).

NEXAFS also provides information on the amount of adsorbates as well as their geometric and electronic structures. Conventional partial electron-yield NEXAFS measurements typically require 5–10 min to acquire one spectrum. If the data-acquisition rate is sufficiently high, it can be applicable to kinetics studies of surface dynamic processes such as surface chemical reactions. The combination of wavelength-dispersed X-rays and a position-sensitive electron energy analyzer enables an NEXAFS spectrum to be obtained in one shot without scanning of the monochromator (Amemiya et al., 2002). Fig. 1 shows a schematic drawing of dispersive mode measurements of Auger electron-yield NEXAFS (dispersive NEXAFS). A flat 2D surface is irradiated by wavelength-dispersed X-rays and the Auger electrons emitted from each position on the surface are separately collected on a 2D detector, where the x axis corresponds to the position on the surface (i.e. the photon energy) and the y axis corresponds to the electron kinetic energy. Integrating the Auger electron counts along the y axis as a function of position (photon energy) yields a NEXAFS spectrum. Fig. 2(a) shows an actual electron image obtained for an oxygen-covered surface in the dispersive mode (Amemiya et al., 2011). O KLL Auger electrons are imaged as a vertical bright line at a certain position (photon energy) where O 1s electrons are excited with a large cross section. Photoelectrons also appear as diagonal straight lines, indicating that the photon energy varies linearly with the position on the surface. The real-time dispersive NEXAFS is obtained like a movie. Fig. 2(b) shows the progress of surface chemical reactions (CO oxidation at an oxygen-covered metal surface under exposure to gaseous CO) every 33 ms. The data-acquisition time has recently been further reduced to 300 µs by using a fast CCD camera. This fast NEXAFS technique allows us to follow very fast kinetics of irreversible surface chemical reactions with a submilli­second time resolution. If the pump-and-probe technique can be used in dispersive NEXAFS measurements for repeatable surface chemical reactions, the time resolution is much improved; in principle it reaches the time width of a single pulse of synchrotron radiation (typically several tens of picoseconds).

Figure 1 Schematic drawing of dispersive NEXAFS.

Figure 2 (a) Electron image obtained by the dispersive mode. (b) Time-resolved dispersive NEXAFS spectrum of a metal surface during a CO oxidation reaction.

Recent surface XAFS developments in time-resolved catalyst research on nanoparticles are described in this section (type II in Section 1) to identify dynamic structure change on real catalyst surfaces under working conditions (Sekizawa et al., 2013).

Quick XAFS (QXAFS) and energy-dispersive XAFS (DXAFS) are typical methods to achieve time-resolved XAFS measurements, as shown in Fig. 3. QXAFS can be measured using a similar optical setup to conventional XAFS, but the monochromator (flat silicon crystals) is quickly moved to achieve quick measurements of XAFS spectra. The time resolution is determined by the mechanical scanning speed of the monochromator. On the other hand, dispersive XAFS is measured by the use of an elliptically bent silicon crystal as a polychromator, and energy-dispersed X-rays through a sample set at one focusing point of the ellipse are detected using a position-sensitive detector. The time resolution of DXAFS usually depends on the response of the position-sensitive detector and the quality of the DXAFS spectra for analysis.

Figure 3 Setups for quick XAFS, dispersive XAFS, scanning nano-XAFS and computed tomography XAFS (Sekizawa et al., 2013).

The recent development of spatially resolved XAFS methods provides a significant opportunity for analysis of the structures and electronic states of real solid catalysts composed of inhomogeneous surface structures.

Spatially resolved XAFS is attained by two methods: the scanning-nanobeam and the computed tomography methods. Scanning nano-XAFS is a method to obtain two-dimensional spatially resolved XAFS imaging. An X-ray beam passing through a slit is focused to decrease the beam size by a set of Kirkpatrick–Baez (KB) mirrors, which is composed of two perpendicularly faced and elliptically bent mirrors schematically shown in Fig. 3. A sample mounting on a piezo stage is placed at the focusing point of the X-ray nanobeam and two-dimensionally scanned by the stage, recording fluorescent X-rays from the sample. X-ray beams of the order of 100 nm in the hard X-ray region are obtained using the KB mirrors. Computed tomography (CT) is a technique to reconstruct the three-dimensional image of a sample. A sample is perpendicularly rotated to an incident X-ray, and transmission images at each rotating angle are detected, as depicted in Fig. 3. A series of transmission images at different rotation angles are computationally reconstructed to obtain a three-dimensional image of the sample. The measurements of transmitted images at different X-ray energies in the XAFS region can provide CT-XAFS imaging, including three-dimensional morphology information and XAFS spectroscopic information. To image a flat thin sample, limited-angle CT-XAFS and computed laminography XAFS (CL-XAFS) have been developed (Saida et al., 2012).

One of the main advantages of XAFS techniques is the in situ analysis of various catalysts, nanomaterials and surfaces under working conditions, which cannot be performed using other analysis methods. Following the first in situ XAFS observation of structural change in a supported molybdenum dimer catalyst involving an intermediate structure during the course of the catalytic oxidation of ethanol using an in situ XAFS cell, as shown in Fig. 4, in 1985 (Iwasawa et al., 1985), in which the molybdenum dimers are located at the catalyst surface, in situ XAFS techniques are now well recognized as unique and powerful methods for the structural and electronic characterization of supported catalysts, functional nanomaterials and active surfaces under working/operating conditions and more explicitly for analysing dynamic functions such as catalysis (often called operando; Iwasawa, 1987, 1996; Iwasawa et al., 2016; Nagamatsu et al., 2016; Tada et al., 2015; Weckhuysen, 2003). In situ XAFS spectra at atmospheric to high pressures can be measured in transmission and fluorescence modes by means of in situ (static and steady-state) XAFS, time-resolved/real-time XAFS (QXAFS and DXAFS) and 2D and 3D spatially resolved/imaging XAFS, depending on the research objective and the time scale of the target functions (Iwasawa et al., 2016; Nagamatsu et al., 2016; Tada et al., 2015).

Figure 4 An in situ XAFS cell for XAFS measurements in transmission mode during catalyst fabrication and catalytic reaction processes (Iwasawa et al., 1985).