Application of pan-sharpening algorithm for correlative multimodal imaging using AFM-IR

Abstract

The coupling of atomic force microscopy with infrared spectroscopy (AFM-IR) offers the unique capability to characterize the local chemical and physical makeup of a broad variety of materials with nanoscale resolution. However, in order to fully utilize the measurement capability of AFM-IR, a three-dimensional dataset (2D map with a spectroscopic dimension) needs to be acquired, which is prohibitively time-consuming at the same spatial resolution of a regular AFM scan. In this paper, we provide a new approach to process spectral AFM-IR data based on a multicomponent pan-sharpening algorithm. This approach requires only a low spatial resolution spectral and a limited number of high spatial resolution single wavenumber chemical maps to generate a high spatial resolution hyperspectral image, greatly reducing data acquisition time. As a result, we are able to generate highresolution maps of component distribution, produce chemical maps at any wavenumber available in the spectral range, and perform correlative analysis of the physical and chemical properties of the samples. We highlight our approach via imaging of plant cell walls as a model system and showcase the interplay between mechanical stiffness of the sample and its chemical composition. We believe our pan-sharpening approach can be more generally applied to different material classes to enable deeper understanding of that structure-property relationship at the nanoscale.

Correlated mechanochemical maps of Arabidopsis thaliana primary cell walls using atomic force microscope infrared spectroscopy

“…To more accurately delineate local concentrations of specific cell wall functional groups and components, NMF spectral deconvolution was applied (Borodinov et al, 2019 ; Kulkarni et al, 2018 ; Labbe et al, 2005 ; Lin et al, 2018 ; Montcuquet et al, 2010 ; Zhang et al, 2018 ). NMF of the AFM-IR dataset (i.e., input matrix) requires a subjective, user-specified number of factors.…”

“…We provide a sample preparation technique suited for AFM-IR on A. thaliana stem sections, which eliminates chemical contamination. Since many plant cell wall components have overlapping spectral features and because IR spectral imaging creates large and complex datasets, we applied non-negative matrix factorisation (NMF) to reduce the dimensionality of the AFM-IR data into a more manageable number of components (Borodinov et al, 2019 ; Montcuquet et al, 2010 ). These non-negative matrix factors were used to identify the IR spectral features representing different cell wall components.…”

Role of Decomposition Product Ions in Hysteretic Behavior of Metal Halide Perovskite

“…Therefore, we utilized a machine learning methodology to interpret the full tr-ToF-SIMS data set. In particular, we used non-negative matrix factorization (NMF), − which allows one to deconvolute the full data set into linear combinations of a limited number of statistically significant endmembers (more information about method and it is mathematical form can be found in Methods). This allowed us to simplify the complex mass spectra (Figure S6) into a few subgroups by considering their time-dependent behavior, where the ions in each subgroup have similar behavior.…”

A Robust Neural Network for Extracting Dynamics from Electrostatic Force Microscopy Data

“…Furthermore, the resulting data are not always directly plottable or interpretable with a simple physical interpretation. Broadly, researchers are interested in applying these techniques to a variety of systems, from hybrid organic–inorganic metal-halide perovskites to conjugated polymers to study phenomena such as charging and local quantum efficiency or ion dynamics. ,,,,,,− Machine learning (ML) and data science techniques provide us with excellent tools to help process these data and study these systems. ,,− Methods of signal boosting, − dimensionality reduction and representation, ,,, and decomposition based in ML and data science are being developed to extract information efficiently and accurately from SPM images and signals. ,,, …”

“…3,14,26−29 Methods of signal boosting, 30−32 dimensionality reduction and representation, 2,11,19,31 and decomposition based in ML and data science are being developed to extract information efficiently and accurately from SPM images and signals. 14,26,27,33 In practice, the use of ML for SPM data processing provides many potential advantages, namely, robustness against noise, speed in application, and ease of handling multidimensional data. 3,11,26,[29][30][31]34 Here, we demonstrate the use of a neural network (NN) to process data-rich SPM images from a feedback-free implementation of trEFM (Figure 1a).…”