Convolutional neural network applied for nanoparticle classification using coherent scatterometry data

Abstract

The analysis of 2D scattering maps generated in scatterometry experiments for detection and classification of nanoparticles on surfaces is a cumbersome and slow process. Recently, deep learning techniques have been adopted to avoid manual feature extraction and classification in many research and application areas, including optics. In the present work, we collected experimental datasets of nanoparticles deposited on wafers for four different classes of polystyrene particles (with diameters of 40, 50, 60, and 80 nm) plus a background (no particles) class. We trained a convolutional neural network, including its architecture optimization, and achieved 95% accurate results. We compared the performance of this network to an existing method based on line-by-line search and thresholding, demonstrating up to a twofold enhanced performance in particle classification. The network is extended by a supervisor layer that can reject up to 80% of the fooling images at the cost of rejecting only 10% of original data. The developed Python and PyTorch codes, as well as dataset, are available online.

Intelligent control of nanoparticle synthesis through machine learning

“…Moreover, to solve the problem of being able to accurately measure the size of nanoparticles, transmission electron microscopy is used to acquire nanoparticle images online, 144 and then the method of processing the images by computer to identify the centers and contours of nanoparticles to determine their size is also expected. Convolutional neural networks [145][146][147] can process, identify, and segment images very well. If convolutional neural networks can be used to measure the size of nanoparticles, this will also provide advances for the effective and accurate control of machine learning-assisted preparation of nanoparticles.…”

Intelligent nanoscope for rapid nanomaterial identification and classification

“…[8][9][10][11][12][13][14] CNN exploits the spatial locality of an image by using convolutional filters, and CNN image classification methods have demonstrated high accuracy while saving computational cost for size-based image classification, ranging from large objects (e.g., firearms) to small objects (e.g., nanomaterials). [15][16][17][18][19][20][21][22] However, image identification and classification of nanomaterials is limited by the image resolution, the field of view of the sample, and the processing time.…”

Identification of Model Particle Mixtures Using Machine-Learning-Assisted Laser Diffraction

“…In addition, this proposed technique allows for the determination of the predominance (or balance) between particle geometries. These features might be relevant for monitoring particle contamination in industrial manufacturing processes [45].…”

“…Secondly, our technique permits the identification of particle shapes in two-component heterogeneous mixtures, resolving not only the shape of the particles that make up the mixture, but also the predominance (or balance) between particle geometries. This feature might be relevant for monitoring particle contamination in industrial manufacturing processes [30].…”