Method of x-ray goniometry of fluorescence radiation for analysis of surface layers and films on a substrate

The article presents a method for experimentally determining the fundamental parameter of interaction between a broadband x-ray spectrum and a material-the ratio of the effective attenuation coefficient μef of the primary spectrum and the attenuation coefficient μi of the fluorescence line of the i-th element. The determination of the fundamental parameter is necessary to justify the magnitude of the uncertainty of the concentration values found from the intensity of the fluorescence lines. The method is based on the known dependence of the intensity of x-ray fluorescence on the angles of entry φ and exit ψ of radiation from the surface of the sample. At a constant scattering angle 2θ = φ + ψ, a series of measurements of the x-ray spectrum are made with different values of the angle ψk and the integrated fluorescence intensity I(ψk) is measured. Varying the angle ψk in the range from 0° to 90° allows, when measuring a film on a substrate, to move from the "saturated layer" model to the thin film model with all intermediate stages. By fitting the experimental dependence I(ψk), the values μiμef and μef·D are determined, where D is the surface density. The objects of the study were single-component standards as well as zirconium films and nickel foils on single-crystalline silicon substrates. Spectral measurements were performed on an EDXRF spectrometer with SDD X-100 (Amptek). The ψk angle was set in 1° increments by rotating the sample on an HGZ (Carl Zeiss) goniometer; the spectrum accumulation time was 600 s. For single-component standards of zirconium, nickel iron, and titanium, the μiμef values were determined with a relative error of 3%. The influence of the broadband spectrum shape on the μiμef ratio and the λef (effective wavelength of the primary broadband spectrum) value was established: (1) a decrease in the voltage on the tube from 30 to 25 kV leads to an increase in λef for zirconium from 0.535 to 0.656 Å, for nickel from 0.694 to 1.004 Å, and for iron from 1.135 to 1.408 Å; (2) the influence of the Cu-Kα artifact line on the μiμef ratio was revealed: for iron and nickel, which have absorption jumps on opposite sides of this line, the μiμef values differ by more than two times due to the effective excitation of iron fluorescence by the Cu-Kα line. For nickel foils with a thickness of 6-12 μm, the values of μiμef and surface density were measured without the use of standards. The μiμef values of the foils are close to the nickel standard, and the surface density corresponds to the reference values. The relative measurement error does not exceed 5%. For thin zirconium films (0.2-6 μm), it was not possible to detect the theoretically predicted increase in λef with decreasing film thickness. The found values of surface density correspond to the reference values within 10% without the use of standards. The values of the parameters μiμef and μef·D can be determined for a multicomponent film, serve as a criterion for selecting film standards of similar composition and thickness, and be used to justify the uncertainty of measurement results by calibrations.