John T. Armstrong

Objective: To determine test methods and standards that can be used to evaluate and optimize the various electron scattering models and other physical parameters that are used in Monte Carlo calculations of x-ray emission from layered and irregular sample surfaces.

Problem: Electron microbeam analysis is very well suited for qualitative characterization of layered materials, particles and rough surfaces; however accurate quantitative analysis of specimens with irregular boundaries has remained illusive. Many industrial applications require accurate analyses of such samples and a variety of methods have been proposed [1-2]. One of the most promising approaches is the use of Monte Carlo simulations to model the electron and x-ray path lengths in complex samples [1,3]. However, the various models and physical parameters commonly used in these calculations give significantly different analytical results.

Approach: We have modified the Monte Carlo algorithms of Joy and Armstrong [3] to develop a versatile simulation program that calculates x-ray emission from thin films, particles and rough surfaces of given boundary conditions. The program allows defocusing or rastering of the electron beam to any given size, and bombardment of any portion of the particle or rough surface. The program calculates the electron trajectories at a rate of approximately one million electrons per minute and is modularized to allow for easy substitution for the various physical expressions that make up the model. We have applied this program to determine analytical, geometrical and compositional conditions that are particularly sensitive to the choice of physical parameters used in the Monte Carlo models.

Results: The results are in generally good agreement with previously developed geometric corrections and with experimental analyses [1]. They are particularly useful in showing the magnitude of electron sidescattering and its effect on x-ray production. This scattering is sensitive to the parameters used by the program and results in large variation in the relative x-ray intensities with closeness to a sample side, with beam to sample angle, and with beam voltage. For example, Monte Carlo calculations show variation in emitted x-ray intensities of particles of AlNi by greater than a factor of 100 with model-dependent variations of over a factor of two.

Future Plans: The Monte Carlo calculations are being used to identify the best measurement experiments (such as stepping the beam toward sample edges or sequentially changing beam to sample angle) and sample compositions to test the physical parameters. Appropriate analyses will then be made on such specimens (in cooperation with the research efforts of John Small and Dale Newbury) to determine the best equations to use for quantitative analysis.

Publications:
[1] Armstrong J T (1991) Electron Probe Quantitation, eds K F J Heinrich and D E Newbury (New York: Plenum) pp 261-315.

[2] Small J A, Armstrong J T, Bright D S and Thorne B B (1998) Microscopy Microanal. 4 (suppl. 2) pp 184-5.

[3] Armstrong J T (2000) Microbeam Analysis 2000, eds D B Williams and R Shimizu (Inst. Phys. Conf. Ser. No. 165) pp 429-430.

Web Contact micro@nist.gov