The determination of low-Z elements such as carbon, nitrogen, and oxygen in atmospheric aerosol particles is of interest in studying environmental pollution. Conventional electron probe microanalysis technique has a limitation for the determination of the low-Z elements, mainly because the Be window in an energy-dispersive X-ray (EDX) detector hinders the detection of characteristic X-rays from light elements. The feasibility of low-Z element determination in individual particles using a windowless EDX detector is investigated. To develop a method capable of identifying chemical species of individual particles, both the matrix and the geometric effects of particles have to be evaluated. X-rays of low-Z elements generated by an electron beam are so soft that important matrix effects, mostly due to X-ray absorption, exist even within particles in the micrometer size range. Also, the observed radiation, especially that of light elements, experiences different extents of absorption, depending on the shape and size of the particles. Monte Carlo calculation is applied to explain the variation of observed X-ray intensities according to the geometric and chemical compositional variation of individual particles, at different primary electron beam energies. A comparison is carried out between simulated and experimental data, collected for standard individual particles with chemical compositions as generally observed in marine and continental aerosols. Despite the many fundamental problematic analytical factors involved in the observation of X-rays from low-Z elements, the Monte Carlo calculation proves to be quite reliable to evaluate those matrix and geometric effects. Practical aspects of the Monte Carlo calculation for the determination of light elements in individual particles are also considered.

A versatile Monte Carlo program for quantitative particle analysis in electron probe X-ray microanalysis is presented. The program includes routines for simulating electron-solid interactions in microparticles lying on a flat surface and calculating the generated X-ray signal. Simulation of the whole X-ray spectrum as well as phi(z) curves is possible. The most important facility of the program is the reverse Monte Carlo quantification of the chemical composition of microparticles, including low-Z elements, such as C, N, O, and F. This quantification method is based on the combination of a single scattering Monte Carlo simulation and a robust successive approximation. An iteration procedure is employed; in each iteration step, the Monte Carlo simulation program calculates characteristic X-ray intensities, and a new set of concentration values for chemical elements in the particle is determined. When the simulated X-ray intensities converge to the measured ones, the input values of elemental concentrations used for the simulation are determined as chemical compositions of the particle. This quantification procedure was evaluated by investigating various types of standard particles, and good accuracy of the methodology was demonstrated. A methodology for heterogeneity assessment of single particles is also described.