Deep learning to establish structure property relationships of impact copolymers from AFM phase images

Abstract

AFM phase images were collected on impact copolymer samples that differ in bulk mechanical properties and microstructure. A deep learning model based on a convolutional neural net (CNN) successfully classified some combinations of ICP's based on microstructure. A separate regression-based CNN correlated the AFM phase images with various bulk mechanical properties, showing good results with yield strength and ultimate elongation percentage and weak results with flexural modulus and notched izod. The results observed from the deep learning model reveal a relationship between the microstructures as captured by the AFM phase images with the different bulk material properties.

Emerging machine learning strategies for diminishing measurement uncertainty in SPM nanometrology

“…Identifies the degradation of tip geometry [129,130], distinguishes desirable and undesirable tip states [131], accesses the state of tip geometry using incomplete scans [132], automates imaging operation [133], automates AFM imaging on single molecules (i.e., DNA molecules) [134], manages the probe quality, selects an appropriate scanning region, and accesses the acquired image [135], derives the highresolution images from the low-resolution ones [136], recognizes NPs of heterogeneous catalysis formed on HOPG [137][138][139], identifies the nanostructured patterns of LSEC fenestrae [140,141], leads the binary segmentation of noisy images of Au NPs formed on Si [142], predicts molecular structures [143], detects various nanostructure patterns of micro-and NPs in a mixture formed on Si [144], classifies the charged and uncharged defects on H-Si (100) [145], classifies the types of molecules [146], identifies impact copolymer [147], classifies different classes of particle and background SVM Finds an optimal hyperplane that clearly distinguishes data points in high dimensional space…”

“…Sun et al [158] developed a k-means clustering model with a combination of a singular value decomposition de-noising process to identify the local microstructures on low carbon steel based on the AFM images of topographical, surface potential, and capacitance gradient mapping. Yablon et al [147,153]…”

On machine learning analysis of atomic force microscopy images for image classification, sample surface recognition

“…22 These and other factors are paramount when combining AFM with ML. ML analysis has also been used to control image acquisition in AFM, 17,23 to improve the accuracy of nanomechanical measurements, 24,25 to help do the reconstruction of sample properties and structures that are difficult to find using classical mathematical methods, [25][26][27][28][29] and to control imaging. 23,[30][31][32] These applications are typically specific to particular models and samples.…”