The grazing incidence X-ray experiments are powerful tools in surface study because the penetration of X-rays becomes extremely shallow around the optical total reflection condition1). While smooth and flat surface of solid or liquid are favorable target for analysis, in the case of thin films, it is necessary to consider the interference effect; the X-ray electric field is strongly modulated due to multiple reflection at each interface2). Angular profile of the fluorescent yield gives information on elemental depth distribution in principle, and one can analyze with no difficulty when the electric field is the evanescent wave3) (uniform material, no interference) or the standing wave4) (periodic multilayer, extreme case of the interference effect). In the present study, more general interference effect in a non-periodically layered thin film has been studied from a view of analytical application.
The schematics of the arrangement for grazing incidence X-ray fluorescence measurements is shown in the inset of Fig.1. The beam size was adjusted so as to obtain appropriate incident intensity, and was typically 3 mm wide and 0.05 mm high. Two ionization chambers was used for measuring reflectivity. Fluorescent X-rays were measured by a Si(Li) detector, which was set at 90 deg. to the incident X-rays with an inclination of about 10 deg from the horizontal plane. The samples used was Cu[100A]/Ag[230A]/ Au[500A]/Si thin films prepared by the RF sputtering technique. In order to study the interface, the surface and each interface are labeled with extremely small quantities of iron[3A], chromium[6A] and titanium [18A], respectively.
Whole X-ray fluorescence spectra were collected by scanning the incident angle. Since the incident X-ray energy is 8 keV, fluorescence from major elements are suppressed; the copper K shell is not excited, though small peaks of silver L and gold M lines were observed at higher angles. Figure 1 shows the angular dependence of integrated intensities of the fluorescence signals of trace iron, chromium and titanium, which are considered to represent the internal X-ray intensity at the surface, the 1st interface (Cu/Ag), and the 2nd interface (Ag/Au), respectively. Each fluorescent signal from each interface peaks at a different angle. Even when any interference effects is not expected, the elements located at the different depth would peak at different angle. This obviously reflects the gradual increase of the penetration depth according to the angular change. In the present case, however, due to the interference effect, the 1st interface is significantly enhanced at 7.8 mrad, and likewise the 2nd interface is emphasized at 9.2 mrad. These results have been analyzed5) and it has been found to be well interpreted by the calculation based on the Parratt's theory1).
One may notice that when a certain interface is enhanced, the X-ray intensity at the neighboring interface is decreased. This indicates that the interference effect could be applied to the distinction of neighboring interfaces, especially when the element of interest is localized at only either of them. The authors gratefully thank Dr. T. Saito of the National Research Institute for Metals and Prof. D. K. Bowen of Warwick University for their continuous encouragement and helpful discussion to this study. We also thank Mr. C. Bidmead of the University of Warwick for preparing samples.
1) L.G.Parratt, Phys. Rev., 95, 359 (1954).
2) K.Sakurai and A.Iida, Adv. in X-Ray Anal. 35: 813 (1993).
3) A.Iida, K.Sakurai, and Y.Gohshi, Nucl. Instrum. & Methods, A246, 736 (1985).
4) T.W.Barbee Jr. and W.K.Warburton, Mat. Lett. 3, 17 (1984).
5) K.Sakurai and A.Iida, to be appeared in Adv. in X-Ray Anal (1996).
Figure 1 Interference effect observed in angular dependence of reflectivity and fluorescent X-ray intensity from surface (Fe) and interfaces (Cr and Ti) of a Cu/Ag/Au/Si thin film. Schematic drawings of the experimental arrangement are shown in the inset.
