Search

We have demonstrated atomic-scale defect characterization in scanning tunneling microscopy images of single crystal tungsten diselenide using an ensemble of U-Net-like convolutional neural networks. Coordinates, counts, densities, and spatial extents were determined from almost 16 000 defect detections, leading to the rapid identification of defect types and their densities. Our results show that analysis aided by machine learning can be used to rapidly determine the quality of transition metal dichalcogenides and provide much needed quantitative input, which may improve the synthesis process. [ABSTRACT FROM AUTHOR]

Chemical vapour deposition (CVD) synthesis of graphene on copper has been broadly adopted since the first demonstration of this process1. However, widespread use of CVD-grown graphene for basic science and applications has been hindered by challenges with reproducibility2 and quality3. Here we identify trace oxygen as a key factor determining the growth trajectory and quality for graphene grown by low-pressure CVD. Oxygen-free chemical vapour deposition (OF-CVD) synthesis is fast and highly reproducible, with kinetics that can be described by a compact model, whereas adding trace oxygen leads to suppressed nucleation and slower/incomplete growth. Oxygen affects graphene quality as assessed by surface contamination, emergence of the Raman D peak and decrease in electrical conductivity. Epitaxial graphene grown in oxygen-free conditions is contamination-free and shows no detectable D peak. After dry transfer and boron nitride encapsulation, it shows room-temperature electrical-transport behaviour close to that of exfoliated graphene. A graphite-gated device shows well-developed integer and fractional quantum Hall effects. By highlighting the importance of eliminating trace oxygen, this work provides guidance for future CVD system design and operation. The increased reproducibility and quality afforded by OF-CVD synthesis will broadly influence basic research and applications of graphene.Assessment of surface contamination shows that trace oxygen is a key factor influencing the trajectory and quality of graphene grown by low-pressure chemical vapour deposition, with oxygen-free synthesis showing increased reproducibility and quality. [ABSTRACT FROM AUTHOR]

We have demonstrated atomic-scale defect characterization in scanning tunneling microscopy images of single crystal tungsten diselenide using an ensemble of U-Net-like convolutional neural networks. Coordinates, counts, densities, and spatial extents were determined from almost 16 000 defect detections, leading to the rapid identification of defect types and their densities. Our results show that analysis aided by machine learning can be used to rapidly determine the quality of transition metal dichalcogenides and provide much needed quantitative input, which may improve the synthesis process. [ABSTRACT FROM AUTHOR]

Biomechanical models of the gecko adhesive system typically focus on setal mechanics from a single gecko species, Gekko gecko. In this study, we compared the predictions from three mathematical models with experimental observations considering an additional gecko species Phelsuma grandis, to quantify interspecific variation in setal micromechanics. We also considered the accuracy of our three focal models: the frictional adhesion model, work of detachment model, and the effective modulus model. Lastly, we report a novel approach to quantify the angle of toe detachment using the Weibull distribution. Our results suggested the coupling of frictional and adhesive forces in isolated setal arrays, first observed in G. gecko is also present in P. grandis although P. grandis displayed a higher toe detachment angle, suggesting they produce more adhesion relative to friction than G. gecko. We also found the angle of toe detachment accurately predicts a species' maximum performance limit when fit to a Weibull distribution. When considering the energy stored during setal attachment, we observed less work to remove P. grandis arrays when compared with G. gecko, suggesting P. grandis arrays may store less energy during attachment, a conclusion supported by our model estimates of stored elastic energy. Our predictions of the effective elastic modulus model suggested P. grandis arrays to have a lower modulus, E, but our experimental assays did not show differences in moduli between the species. The considered mathematical models successfully estimated most of our experimentally measured performance values, validating our three focal models as template models of gecko adhesion (see Full and Koditschek in J Exp Biol 202(23):3325-3332, ), and suggesting common setal mechanics for our focal species and possibly for all fibular adhesives. Future anchored models, built upon the above templates, may more accurately predict performance by incorporating additional parameters, such as variation in setal length and diameter. Variation in adhesive performance may affect gecko locomotion and as a result, future ecological observations will help to determine how species with different performance capabilities use their habitat. [ABSTRACT FROM AUTHOR]