Secondary Ion Mass Spectrometry (SIMS) is based on the detection of ionised atoms, molecules, or molecular fragments sputtered as a result of the bombardment of primary ions onto the surface of the sample to be analysed. Primary ion beam energies are typically of a few keV and its species range within a variety of monoatomic (e.g. Ar+, Ga+, Bi+) and cluster ions (C60+,Bin+,Au3+). After an atomic collision cascade (which leads to a collective molecular motion regime for polymers [1]), only the fragments from the first few monolayers of the sample will have sufficient energy to overcome the surface binding energy and leave the sample, with less than 1% of the emitted species being ionised and therefore detectable by the SIMS technique [2]. Modern instruments will contain primary ion probes capable of rastering the samples surfaces and time-of-flight (ToF) detection systems with high speed electronics, which enables parallel detection of a large range of masses with very high sensitivity and specificity [3]. Most surface analysis laboratories use ToF-SIMS spectrometers in dual beam depth profile mode, which will typically generate hyperspectral image datasets distributed throughout a 3D cuboid containing more than 256 x 256 x 500 voxels with each voxel containing from 20,000 to 2,000,000 spectral channels. (this is the same text I wrote for the other paper, I am struggling to find another way to introduce it).
Over the last twenty years, the use of multivariate analysis (MVA) methods has increased significantly within the SIMS community enabling the processing of large amounts of complex data in a reasonable amount of time and at the same time extract the maximum chemical information from the data. Such spread of MVA demanded standardization of the methodology and appropriate software. There is a number of reviews and tutorials that set the standard way of dealing with multivariate data within the SIMS community [4–9]. In terms of software, the most used spectrometers manufacturers (Iontof, Ionoptika, Phisical Electronics) do not provide a complete set of MVA tools in their analysis software, which make researchers go for independently developed alternatives. The three most used software for MVA within the SIMS community are the PLS.MIA toolbox by Eigenvector research [10], the NBtoolbox, developed by Graham [11] and the MCR-ALS toolbox developed by Jaumot, Gargallo, de Juan and Tauler [12]. This paper presents an alternative GUI that runs as a Matlab app or a standalone software with its main merit being on the data visualisation tools and the capacity of dealing with large sparse datasets, typical of imaging and 3D ToF-SIMS data.
The most complex dataset one can generate with ToF-SIMS is a 3D dataset. Using a dual beam set up it is possible to have lateral and depth information arranged in a 3D hyperspectral cuboid. In order to perform MVA on those datasets, they must first be unfolded in a manner that enables the final results to be folded back without loss of spatial information. Figure 1 shows a flowchart describing how the unfolding process is done for a 3D dataset.
Figure 1: Flowchart describing how the unfolding process is done for a 3D dataset.
In case of images datasets there are no further levels and only level 1 is unfolded and processed whereas in case of depth profiling data, the lateral information is collapsed into one data point per level. For spectra analysis, every measurement is regarded as a sample and technical repeats are acquired often to increase statistics.
MVA Tools | simsMVA 