Estimation of uncertainty in electron probe microanalysis: iron determination in manuscripts, a case study

Abstract.

A case study of ISO GUM-based estimation of measurement uncertainty of quantitative surface elemental analysis is presented. The analytical task was the measurement of iron content in the ink writing on the surface of an 18^th century manuscript by electron probe microanalysis using a scanning electron microscope equipped with an energy-dispersive X-ray spectrometer (SEM/EDS). The problems that arise in uncertainty estimation of quantitative surface analysis are outlined (defining the measurand, preparing calibration standards, non-uniformity of the surface, etc) and ways of overcoming them are suggested. The average iron content on the ink-covered surface of the sample manuscript was found to be 0.12 ± 0.04 µg mm⁻² (at confidence level 95.5%). Paper sheets with ink lines of known iron contents were used as calibration standards. The main source of uncertainty was the variability of the parallel measurement results from different locations of the sample surface (mostly due to the variations in the surface structure of paper), which contributed 78% of the total uncertainty. It is concluded that EPMA using a SEM/EDS is suitable for at least semi-quantitative determination of iron in the writing of ink-written manuscripts. Several specific issues in uncertainty analysis are pointed out that need further investigation.