|
|
3.3.4 Consistency of Secondary Ion Mass Spectrometry and Neutron Depth Profiling Measurements
Kevin J. Coakley Statistical Engineering Division, ITL
George Lamaze and Heather Chen-Mayer Analytical Chemistry Division, CSTL
David S. Simons Surface and Microanalysis Science Division, CSTL Neutron Depth Profiling (NDP) is a nondestructive method for analysis of the concentration profile of an element in material. Inferences about the concentration depth profile are based on the observed energy spectrum of charged particles emitted due to specific nuclear reactions. The detector response function (DRF) is a probability transition matrix which relates the depth of emission to the expected energy spectrum of the detected particles. The DRF depends on the geometries of the emitter and detector, and assumed models for the stopping power of the material, energy straggling, multiple scattering and detector energy resolution. In previous work, we developed a computer code to predict the DRF. Over the last year, we improved the model to account for non-Gaussian energy resolution functions. In a calibration experiment, the energy resolution function of the NDP detector was measured. For low energies, we model the energy resolution function as an exponential. For higher energies, we using a B-spline expansion. A study of the consistency of NDP and SIMS data continued. To get more conclusive results, a new silicon sample was prepared. In the new sample, the boron profile is sharper than in previous samples. The depth profile of Boron in a Silicon sample was measured by Secondary Ion Mass Spectrometry (SIMS). Based on the measured SIMS profile and the modeled DRF, we predict the NDP energy spectrum.
Figure 17: Based on the energy spectrum of alpha particles from a boron surface deposit (top), we esitimate the energy resolution function of the detector for a 1472.6 keV alpha particle (bottom). The vertical line is at 1472.6 keV. For low energies, we model the energy resolution function as an exponential. For higher energies, we used a B-spline approximation. In our modeling, we assume that the energy resolution function is shift invariant.
Date created: 7/20/2001 |