Scanning probe microscopy (SPM) techniques have become the mainstay of nanoscience and nanotechnology by providing easy to use, non-invasive structural imaging and manipulation on the nanometer and atomic scales. Beyond topographic imaging, SPMs have found an extremely broad range of applications for probing electrical, magnetic and mechanical properties – often at the level of several tens of nanometers, opening the pathway for understanding materials functionality and interactions on these length scales.

For more than a decade since the introduction of commercial microscopes in 1992, the SPM has developed as a primarily qualitative method. In imaging modes, only a single or a small number of parameters describing the local properties are obtained and the information contained in complementary images is usually interpreted solely within the limits of a cursory examination. In spectroscopic modes, the spatial correlations between adjacent points are generally ignored.

In a drastic contrast, the last five years have seen tremendous progress in force-based SPMs. The emergence of digital and field-programmable gate array electronics have greatly increased the data acquisition and processing speed, allowing multiple information channels to be acquired without compromising image acquisition speed. Similarly, recent advances in the theoretical understanding of SPM and increasing market competition have led to the emergence of a slew of multimodal and spectroscopic SPM methods, such as dual excitation frequency SPM (Asylum), HarmoniX (Veeco) and configurable multiple frequency lock-ins by Agilent and Nanonis. The unresolved challenge of these techniques is the necessity to analyse multiple parallel data channels to prevent information overload in the human observer.

In this work, we introduce an approach for the analysis of multi-dimensional, spectroscopic-imaging data based on PCA combined with analysis of spatial correlations. PCA selects and ranks relevant response components based on variance within the data. It is shown that for examples with small relative variations between spectra, the first few PCA components closely coincide with results obtained using model fitting, and achieves this at rates approximately four orders of magnitude faster. For cases with strong response variations, PCA allows an effective approach to rapidly process, de-noise and compress data. The prospects of PCA combined with correlation function analysis of component maps as a universal tool for data analysis and representation in microscopy are discussed.

The researchers presented their results in Nanotechnology.