Evaluating Consensus in Experimental K-ratios from over 40 WDS and EDS Measurement Systems

The k-ratio compares the X-ray intensity of a particular emission line between two materials. To convert raw X-ray intensity emitted from a sample to elemental concentration, the k-ratio is measured for the sample relative to a standard and corrected for matrix effects. The k-ratio is theoretically independent of instrument or detector type, beam conditions, or other variables associated with the measurement experiment. Thus the k-ratio measured between a pair of high-purity, stoichiometric crystals should be a universal constant, and community measurements of the same materials should converge on consensus values [1]. To test the universality of experimental k-ratios we acquired raw X-ray intensities from pairs of synthetic single crystals using multiple different labs and measurement systems. A total of 29 participating labs from academic, federal, and commercial institutions representing 10 countries received samples for analysis of experimental k-ratios. Participants acquired Wavelength Dispersive Spectrometry (WDS) data from 22 Electron Probe Microanalysis (EPMA) instruments and Energy Dispersive Spectroscopy (EDS) data from 18 EDS detectors mounted on EPMA or Scanning Electron Microscope (SEM) instruments. Analytical k-ratios were measured for Mg-Kα, Al-Kα, and O-Kα from three high-purity synthetic oxides: MgO, Al2O3, and MgAl2O4. Each participating lab received a polished, carbon-coated mount containing an aliquot from the same synthetic crystal with specific instructions to acquire raw X-ray intensities following identical measurement procedures. Data were acquired on both Cameca and JEOL EPMA instruments ranging in age from >35 to <1 years since installation. When available, k-ratios for Al and Mg were measured and aggregated from multiple spectrometers measured with both large- (LTAP/TAPL/TAPH) and normal- (TAP) sized diffraction crystals. Standardized procedures involved acquisition of wavescans to determine background offset positions and collection of raw X-ray counts acquired under specified beam conditions, detector settings, and counting times. Analytical k-ratios acquired with WDS from 22 EPMA instruments are shown in Fig. 1. The k-ratios show overall excellent agreement, with most labs reporting k-ratios that fall within 1% of the sample mean. Two laboratories reported k-ratios that differ by >2% from the mean value. One of these outlier laboratories (FIGMAS 1-02) identified a malfunction in instrument performance that affected Mg analysis, and following necessary repairs submitted a corrected dataset, as indicated in Fig. 1. Instructions for acquisition of WDS k-ratios involved aggregating k-ratios from the same type of diffraction crystal measured by multiple spectrometers. However, this could potentially obscure errors with individual spectrometers if k-ratios are not in agreement. Careful comparison of k-ratio data from the same diffraction crystal on multiple spectrometers can provide valuable information for evaluating instrument performance. Analytical k-ratios were acquired from 18 EDS detectors mounted on both SEM and EPMA instruments. Data were acquired using EDS detectors from five different manufacturers (Thermo, Oxford, Bruker, JEOL, Amptek) mounted with take-off angles ranging from 29-40 degrees. EDS k-ratios were normalized to a take-off angle of 40 degrees for direct comparison. Raw X-ray intensity determined from an EDS spectrum is a function of the peak integration method, and there can be slight differences in EDS k-ratios depending on which method of background correction is used. EDS-k-ratios reported here are calculated using the net X-ray intensity calculated by analysis software, thus it is expected that some of the uncertainty in EDS k-ratios could be attributed to differences in peak fitting algorithms. Determination of a universal k-ratio from EDS data will require off-line processing of raw EDS spectra using a consistent approach for peak fitting. Results of this round robin experiment indicate that the microanalysis community is capable of converging on consensus k-ratio values using measurements from a variety of instruments and detectors. The consensus results show good agreement with theoretical k-ratios determined with PENELOPE, DTSA II, and BadgerFilm simulations. We advocate for developing a universal standard mount containing a pair of materials which form the fundamental k-ratio for each element and a global community database of k-ratios to advance the accuracy and traceability of quantitative X-ray microanalysis. K-ratios for Mg, Al, and O measured with Wavelength Dispersive Spectrometry (WDS) from 22 different instruments. WDS data were acquired at 15 kV and 20 nA. The solid line shows the mean value of all participants. The dashed and dash-dotted lines show +/- 2% and +/- 5% from the sample average, respectively. The grey dashed line shows the theoretical k-ratio calculated from BadgerFilm simulation. K-ratios for Mg, Al, and O measured with Energy Dispersive Spectroscopy (EDS) from 18 different detectors. EDS data were acquired at 15 kV with approximately 20% dead time. All EDS k-ratios were normalized for take-off angle to enable direct comparison of results. The solid line shows the mean value of all participants. The dashed lines show +/- 2% from the sample average. The dot-dashed lines show +/- 5% from the sample average. The grey dashed line shows the theoretical k-ratio calculated from BadgerFilm simulation.