One way of expediting materials development is to decrease the need for new experiments by making greater use of published literature. Here, we use data mining and automated image analysis to gather new insights on nanoporous gold (NPG) without conducting additional experiments or simulations. NPG is a three-dimensional porous network that has found applications in catalysis, sensing, and actuation. We assemble and analyze published images from among thousands of publications on NPG. These images allow us to infer a quantitative description of NPG coarsening as a function of time and temperature, including the coarsening exponent and activation energy. They also demonstrate that relative density and ligament size in NPG are not correlated, indicating that these microstructure features are independently tunable. Our investigation leads us to propose improved reporting guidelines that will enhance the utility of future publications in the field of dealloyed materials.
Miniaturized machines have open up a new dimension of chemistry, studied usually as an average over numerous molecules or for a single molecule bound on a robust substrate. Mechanical motions at a single molecule level, however, are under quantum control, strongly coupled with fluctuations of its environment — a system rarely addressed because an efficient way of observing the nanomechanical motions in real time is lacking. Here, we report sub-millisecond sub-Å precision in situ video imaging of a single fullerene molecule shuttling, rotating, and interacting with a vibrating carbon nanotube at 0.625 milliseconds(ms)/frame or 1600 fps, using an electron microscope, a fast camera, and a denoising algorithm. We have achieved in situ observation of the mechanical motions of a molecule coupled with vibration of a carbon nanotube with standard error as small as 0.9 millisecond in time and 0.01 nm in space. We have revealed rich molecular dynamics, where motions are non-linear, stochastic and often non-repeatable, and a work and energy relationship at a molecular level previously undetected by time-averaged measurements or microscopy. The molecular video recording at a 1600-fps rate exceeds by 100 times the previous records of continuous recording of molecular motions.
This study examined the improvement of microscopy segmentation intersection over union accuracy by transfer learning from a large dataset of microscopy images called MicroNet. Many neural network encoder architectures were trained on over 100,000 labeled microscopy images from 54 material classes. These pre-trained encoders were then embedded into multiple segmentation architectures including UNet and DeepLabV3+ to evaluate segmentation performance on created benchmark microscopy datasets. Compared to ImageNet pre-training, models pre-trained on MicroNet generalized better to out-of-distribution micrographs taken under different imaging and sample conditions and were more accurate with less training data. When training with only a single Ni-superalloy image, pre-training on MicroNet produced a 72.2% reduction in relative intersection over union error. These results suggest that transfer learning from large in-domain datasets generate models with learned feature representations that are more useful for downstream tasks and will likely improve any microscopy image analysis technique that can leverage pre-trained encoders.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.