| AVS 54th International Symposium | |
| Applied Surface Science | Monday Sessions |
| Session AS-MoA |
| Session: | Quantitative Surface Analysis II. Electron Spectroscopies: (Honoring the contributions of Martin Seah, NPL, and Cedric Powell, NIST) |
| Presenter: | M.R. Keenan, Sandia National Laboratories |
| Authors: | M.R. Keenan, Sandia National Laboratories V.S. Smentkowski, General Electric Global Research J.A. Ohlhausen, Sandia National Laboratories |
| Correspondent: | Click to Email |
The time-to-digital converters typically employed for ToF-SIMS have a dead-time that is significant with respect to the distribution of times-of-flight characteristic of a given nominal mass. When data are acquired with high mass resolution, it is generally the case that only a single secondary ion can be detected at any given mass per primary ion pulse, regardless of how many ions actually reach the detector. Consequently, the data become non-linear in the sense that the signal intensity is no longer proportional to the local composition. This phenomenon often manifests itself as "saturation," for instance, finding isotope ratios that deviate significantly from generally accepted values. The data non-linearity also poses significant problems for multivariate statistical analysis techniques (MVSA), which fundamentally assume an underlying linear additive model. In these cases, extraneous components are often discovered that simply describe the non-linearity and can ultimately confound chemical interpretation. Dead-time correction procedures have been described, previously, for time-of-flight mass spectrometry. These typically rely upon observing a statistically significant number of ion pulses at each volume element, a situation that is not practical for ToF-SIMS spectral images composed of large numbers of pixels, 3D depth profiles and stage rastering measurements. In this paper, various dead-time effects on ToF-SIMS spectral images will be described. We will also propose a novel data pretreatment protocol that is suitable for use with MVSA. The new method involves a transformation to linearize the data, which is similar to previously published corrections, together with a scaling step to properly account for the non-uniform noise characteristic of dead-time-limited data. The ability of the new protocol to enable the extraction of accurate, meaningful components while excluding noise during MVSA will be illustrated with simple, intuitive examples.
Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000.