Multivariate image analysis strategies for ToF‐SIMS images with topography

Despite the benefits of multivariate analysis methods, many challenges remain with their robust applications to real-life samples relevant to industry. Here, we use hair fibres pre-treated with a multi-component formulation to investigate different multivariate analysis strategies for complex time-of-flight secondary ion mass spectrometry (ToF-SIMS) images obtained in practical analysis. This is challenging because of extreme topography, a large number of unknown chemical components and detector saturation. We compare results from principal component analysis (PCA) and multivariate curve resolution (MCR) with no scaling, Poisson scaling and binomial scaling. Because of severe topography, scaling methods are modified to operate in the spectral domain only. We propose the use of a maximum ion intensity spectrum to highlight localised chemical features and diagnose detector saturation. Dead time correction with suitable data scaling is demonstrated to be essential for the detection of small, localised chemical variations. While PCA results are difficult to interpret, MCR results resemble secondary ion mass spectrometry (SIMS) spectra and distributions directly. MCR is also superior to manual analysis for the detection of an important interaction between multiple ingredients. However, unlike PCA, the scores and loadings obtained on different MCR factors are correlated. The consequence of this for the optimal resolution of independent chemical features is discussed in detail. Binomial scaling is identified as the most appropriate data scaling method for this image due to detector saturation. This study provides a robust analysis strategy for complex ToF-SIMS images, essential for increasingly complex multi-organic surfaces and biomaterials.