For pt.I see ibid., vol.18, p.1233 (1985). A new absorption correction is developed which is shown to work well for quantitative analysis of all systems including light elements. It is based upon representing the X-ray depth distributions as a quadrilateral shape defined by the mean depth of X-ray generation and the position and relative height of the peak in the X-ray depth profile. Equations for describing these parameters in terms of target material, incident electron energy, etc. are derived by reference to tracer and Monte Carlo determinations of X-ray depth distributions.

X-ray distributions in aluminium, gold and carbon have been measured using the tracer technique. The results have been compared with calculations using the simplified Monte Carlo method of Love et al. (1977). It is concluded that the Monte Carlo model predicts too high a peak to the X-ray distribution and that the discrepancy is worse at lower overvoltage ratios. Calculated surface ionisation functions Phi (0) are, however, closer to experimental values and the general shape of the Phi ( rho z) curve is considered satisfactory. The more rigorous model of Myklebust et al. (1976) also overestimates the height of the peak in the Phi ( rho z) curve and, whilst it performs rather better than the simplified model for gold, it gives even more unrealistic data on carbon.

For pt.III see ibid., vol.18, p.1269 (1985). X-ray distributions in targets inclined to the electron beam have been established using both tracer measurements and Monte Carlo calculations. The data are used to extend the Love-Scott II correction for quantitative electron-probe microanalysis so that it can also deal satisfactorily with tilted specimens.

No abstract

For pt.II see ibid., vol.18, p.1245 (1985). Four recently proposed correction procedures for quantitative electron-probe microanalysis have been evaluated using a fairly comprehensive collection of microanalysis results. The Love-Scott II model, which incorporates the new 'quadrilateral' absorption correction, gives the best performance for all specimens included in the database. The methods proposed by Bastin (1984) and by Tanuma (1983) are reasonable for heavier element analysis but are less satisfactory when applied to the lighter elements (Z<10). The Love-Scott I model, with its simpler 'square' Phi ( rho z) profile, is not recommended for use when absorption effects are large, say 50%, although it can deal satisfactorily with heavy element analyses.