Paper AS-MoA7
Post-Acquisition Mass Resolution Improvement in Time-Of-Flight Secondary Ion Mass Spectrometry (TOF-SIMS)

Monday, October 31, 2011, 4:00 pm, Room 102

Time-of-flight secondary ion mass spectrometers employing pulsed primary ion beams provide excellent mass resolution, on the order of 10,000 (full-width-at-half-maximum) over most of the spectral range. Unfortunately, even with all instrumental parameters optimized, ultimate mass resolution can only be achieved by sampling a relatively small area on a smooth surface oriented perpendicular to the extraction optics, under a uniform electric field. It is often difficult to meet these four criteria simultaneously.

These criteria fall into two categories, geometric and electrical. Mass resolution degradation due to geometric factors is the result of a distribution of flight times for ions of the same mass, caused by secondary ions originating from different vertical and horizontal positions within the analysis area, and, for rastered primary ion beams, by differences in the flight times of primary ions across the rastered area. Partial correction of these problems can be achieved in real time through hardware and software compensation, but the instrument must be well-tuned. For insulators, electrical effects may be convoluted with geometric factors and influence mass resolution in a number of unpredictable ways.

For data acquired in “raw” mode (full spectrum at every pixel), it is sometimes possible to correct for these real-world difficulties after data acquisition. Two approaches are employed. The first involves subdividing the analysis area into a regular grid of smaller regions and extracting mass spectra from each region. The extracted spectra are individually calibrated by an automated process, and all or an optimized portion of the spectra are summed to produce a new spectrum with higher mass resolution than the original total spectrum. Interestingly, the spectral calibration information can be used as a diagnostic tool for instrument alignment and tuning.

The second approach is effective for improving mass resolution in spectra of rough surfaces, such as fabrics. Unlike the first approach, the analysis area is not subdivided into a regular pattern. Rather, spectra are obtained from regions of similar height, identified by any of four methods ranging from manual selection of regions-of-interest to automated pixel selection using principal component analysis and multivariate curve resolution. The automated methods have the advantage of simultaneously optimizing the mass resolution and the spectral counts without having to take a trial-and-error approach.

With these methods, mass resolution improvements of 20% - 50% are typical for smooth surfaces, and much larger improvements can be achieved for rough surfaces.