Anomalous small-angle X-ray scattering (ASAXS) is a technique capable of extracting element-specific structural information from multi-element materials, and is considered useful in the structure analysis of metal alloy catalyst particles for fuel cells. ASAXS utilizes the resonant term of the atomic scattering factor of the element of interest. The resonant term has a real part (f1) and an imaginary part (f2), and f1 mainly contributes to the anomalous scattering phenomenon. The amplitude of f1 varies with X-ray energy, and decreases sharply around the absorption edges of the element. The decrease is as large as -18 electrons in the case of platinum (Pt) LIII-edge. ASAXS experiments are possible only in synchrotron radiation facilities where X-ray energies are freely tunable.

ASAXS experiments require X-ray recordings at 3 or more different X-ray energies to mathematically isolate the element-specific component. Although the energy dependence of f1 seems substantial at around the absorption edge, its contribution to the total scattering intensity from a multi-element system is subtle, and high-precision measurements are required to obtain correct structural information.

In actual measurements, experimental errors can arise from many sources, even in present-day third-generation synchrotron radiation facilities. They include fluctuations of the X-ray beam intensities and positions, the calibration error of X-ray energies, and time-dependent changes of sample properties. Here we evaluate the effect of such errors by using model calculations. The model consists of a mixture of 3-nm Pt particles and 5-nm gold (Au) nanoparticles.

The scattering profiles of this mixture were calculated for X-ray energies of 11.504, 11.555 and 11.560 keV (Curve A in Figure 1. The LIII edge is at 11.564 keV). The scattering amplitude of the Pt particles was calculated by using the f1 and f2 values from the Cromer-Liberman tables (Brennan & Cowan, Rev. Sci. Instr., 63, 1992), using equations described by Sanada et al. (J. Am. Chem. Soc., 135, 2013). The scattering from Au particles (Curve B) was assumed to be independent of energy. The extracted scattering profile for Pt was identical to that calculated for a sample with Pt particles alone (Curve D, compare with C). Next, the scattering intensities at a specific energy were deliberately changed by a few percent before extracting the Pt contribution. For intensities at 11.56 keV, even a reduction of as slight as 0.2% resulted in catastrophic consequences (Curve F). With a 5% increase of intensity, the generated Pt profile superficially looked normal, but in reality the scattering curve retained the features of the Pt-Au mixture, meaning that the Pt-specific component was not successfully extracted (Curve D).

These results indicates that the accuracy requirements in ASAXS measurements are very stringent, and every effort should be made to eliminate factors that can cause errors and uncertainties.

In the meeting, data recorded from actual catalyst samples will also be presented, including those from MIRAI (fuel-cell vehicle manufactured by Toyota).

This work was performed under the NEDO FC-Platform project.