A new absorption correction based on a normalized lonization distribution model for electron microprobe analysis and its application for oxygen analysis

Abstract

This paper described a new absorption correction for electron microprobe analysis and its application for oxygen analysis in inorganic compounds. The proposed absorption correction was described as $$f(\chi ) = A[exp(v^2 )erfc(v) - B[exp(u^2 )erfc(u)],$$ whereA=[exp(c 2)(1+erf(c))−B)]−1,B=6.65×10−2, $$u = \chi \cdot \overline {\varrho z} /6.64,v = \chi \cdot \overline {\varrho z} /1.33 - c$$ andc=0.1995. This equation was based on the following normalized ionization distribution model $$\phi (w) = 0.664 \cdot \exp \{ - 0.665^2 (w - 0.3)^2 \} - 0.2\exp ( - 3.23^2 w^2 ).$$ . The above equation was established using φ(w) data which were determined by Bishop using the Monte Carlo method with a Cu target at 30 kV and an overvoltage ratio of 10. The proposed absorption correction was examined on 310 microanalysis data (oxygen: 160, heavier element: 150) and gave greater accuracy than that of the Philibert corrections. In the analysis of oxygen, over 68% of the data corrected by the proposed absorption correction and Duncumb-Reed atomic number correction laid within 5% of the true value. The RMS error of oxygen analysis obtained using the proposed absorption correction was 5.0% (with Duncumb-Reed Atomic number correction), and that of other elements was 5.5% (with the Pool-Thomas atomic number correction). These values were superior to that of simple Philibert (oxygen 9.6%, heavier elements 6.2%) or full Philibert absorption corrections (oxygen 8.5%, heavier elements 8.3%).