Identifying Nanoscale Structure–Function Relationships Using Multimodal Atomic Force Microscopy, Dimensionality Reduction, and Regression Techniques

Correlating nanoscale chemical specificity with operational physics is a long-standing goal of functional scanning probe microscopy (SPM). We employ a data analytic approach combining multiple microscopy modes using compositional information in infrared vibrational excitation maps acquired via photoinduced force microscopy (PiFM) with electrical information from conductive atomic force microscopy. We study a model polymer blend comprising insulating poly­(methyl methacrylate) (PMMA) and semiconducting poly­(3-hexylthiophene) (P3HT). We show that PiFM spectra are different from FTIR spectra but can still be used to identify local composition. We use principal component analysis to extract statistically significant principal components and principal component regression to predict local current and identify local polymer composition. In doing so, we observe evidence of semiconducting P3HT within PMMA aggregates. These methods are generalizable to correlated SPM data and provide a meaningful technique for extracting complex compositional information that is impossible to measure from any one technique.