Issue link: https://iconnect007.uberflip.com/i/239245
determining phosphorus content in en plating using xrf spectroscopy continues ter system. Displayed are the modeled spectra for a 5 µm-thick nickel/phosphorus coating on iron, with the phosphorus content varying between 0% and 15%. One notices a clear dependence of the Ni and Fe intensity with respect to the phosphorus content. The physical reason is the absorption effect of the element phosphorus on both the Fe-K and the Ni-K radiation. So, there is a well-defined correlation between the thickness of the nickel/phosphorus coating and its phosphorus content as the unknown measurement variables and the intensities of Fe-K and Ni-K radiation that can be measured with a proportional counter tube. For a constant thickness, more phosphorus reduces the Ni-K intensity relative to the Fe-K intensity, because the absorption by phosphorus is less for iron than nickel. The evaluation software WinFTM [2] processes this information and computes the thickness and the composition of the coating from the measured spectrum. Table 1 shows the results obtained from the measurement of a flat NiP/Fe reference sample. The measuring application was not calibrated (standards-free analysis). The small deviation from the nominal value indicates that the model underly- ing the evaluation is quite good. Even more important is the good precision of 0.25% for the phosphorus concentration. However, it is apparent that the total measurement uncertainty increases due to systematic uncertainty (Section 3). An error due to possible distance variations must be considered as part of the random measurement uncertainty. A shift in the measurement distance results in a change in intensity of all spectral components and in particular, in erroneous %P readings Figure 3 illustrates this based on the specimen from Table 1. In general the uncertainty of the measuring distance setting z is better than 50 μm, such that, the resulting uncertainty in determining the concentration is about 0.2% phosphorus. If one further considers that a positioning uncertainty also occurs during calibration, then this error source must be estimated at about 0.3% phosphorus. This is the same magnitude as the precision itself. Other sources of error (influence of curvature and tilt of the specimen surface) during the indirect measurement of NiP/Fe can also contribute to the total error. Figure 2: Computed spectra for a 5 μm-thick nickel/phosphorus coating on iron, applicable for the measurement conditions of a Fischerscope® X-ray XULM, 50 kV, Ni-filter. The detector is an Xe-filled proportional counter tube. It is not possible to evaluate the P-K peak at 2 keV with this type of detector. Table 1: Measurement documentation for the standards-free thickness and composition determination of a NiP/Fe reference sample (14.3 μm, 9.3% P); Measurement under repeatability conditions. January 2014 • The PCB Magazine 41