Defect detection in atomic-resolution images via unsupervised learning with translational invariance

Crystallographic defects can now be routinely imaged at atomic resolution with aberration-corrected scanning transmission electron microscopy(STEM)at high speed,with the potential for vast volumes of data to be acquired in relatively short times or through autonomous experiments that can continue over very long periods.Automatic detection and classification of defects in the STEM images are needed in order to handle the data in an efficient way.However,like many other tasks related to object detection and identification in artificial intelligence,it is challenging to detect and identify defects from STEM images.Furthermore,it is difficult to deal with crystal structures that have many atoms and low symmetries.Previous methods used for defect detection and classification were based on supervised learning,which requires human-labeled data.In this work,we develop an approach for defect detection with unsupervised machine learning based on a one-class support vector machine(OCSVM).We introduce two schemes of image segmentation and data preprocessing,both of which involve taking the Patterson function of each segment as inputs.We demonstrate that this method can be applied to various defects,such as point and line defects in 2D materials and twin boundaries in 3D nanocrystals.

Self-driven highly responsive p-n junction In Se heterostructure near-infrared light detector

Photodetectors converting light signals into detectable photocurrents are ubiquitously in use today.To improve the compactness and performance of next-generation devices and systems,low dimensional materials provide rich physics to engineering the light-matter interaction.Photodetectors based on two-dimensional(2D)material van der Waals heterostructures have shown high responsivity and compact integration capability,mainly in the visible range due to their intrinsic bandgap.The spectral region of near-infrared(NIR)is technologically important,featuring many data communication and sensing applications.While some initial NIR 2D material-based detectors have emerged,demonstrations of doping-junction-based 2D material photodetectors with the capability to harness the charge-separation photovoltaic effect are yet outstanding.Here,we demonstrate a 2D p-n van der Waals heterojunction photodetector constructed by vertically stacking p-type and n-type indium selenide(In Se)flakes.This heterojunction charge-separation-based photodetector shows a threefold enhancement in responsivity in the NIR spectral region(980 nm)as compared to photoconductor detectors based on p-or n-only doped In Se.We show that this junction device exhibits self-powered photodetection operation,exhibits few p A-low dark currents,and is about 3-4 orders of magnitude more efficient than the state-of-the-art foundry-based devices.Such capability opens doors for low noise and low photon flux photodetectors that do not rely on external gain.We further demonstrate millisecond response rates in this sensitive zero-bias voltage regime.Such sensitive photodetection capability in the technologically relevant NIR wavelength region at low form factors holds promise for several applications including wearable biosensors,three-dimensional(3D)sensing,and remote gas sensing.