Controlling crystalline material defects is crucial, as they affect properties of the material that may be detrimental or beneficial for the final performance of a device. Defect analysis on the sub-nanometer scale is enabled by high-resolution (scanning) transmission electron microscopy [HR(S)TEM], where the identification of defects is currently carried out based on human expertise. However, the process is tedious, highly time consuming and, in some cases, yields ambiguous results. Here we propose a semi-supervised machine learning method that assists in the detection of lattice defects from atomic resolution microscope images. It involves a convolutional neural network that classifies image patches as defective or non-defective, a graph-based heuristic that chooses one non-defective patch as a model, and finally an automatically generated convolutional filter bank, which highlights symmetry breaking such as stacking faults, twin defects and grain boundaries. Additionally, we suggest a variance filter to segment amorphous regions and beam defects. The algorithm is tested on III-V/Si crystalline materials and successfully evaluated against different metrics, showing promising results even for extremely small training data sets. By combining the data-driven classification generality, robustness and speed of deep learning with the effectiveness of image filters in segmenting faulty symmetry arrangements, we provide a valuable open-source tool to the microscopist community that can streamline future HR(S)TEM analyses of crystalline materials.
Transmission electron microscopy has been a promising candidate for mapping atomic orbitals for a long time. Here, we explore its capabilities by a first principles approach. For the example of defected graphene, exhibiting either an isolated vacancy or a substitutional nitrogen atom, we show that three different kinds of images are to be expected, depending on the orbital character. To judge the feasibility of visualizing orbitals in a real microscope, the effect of the optics aberrations is simulated. We demonstrate that, by making use of energy-filtering, it should indeed be possible to map…
We demonstrate that the sub-atomically focused beam of a scanning transmission electron microscope (STEM) can be used to controllably manipulate individual dopant atoms in a 2D graphene lattice. We demonstrate the manipulation of adsorbed source materials and the graphene lattice with the electron beam such that individual vacancy defects can be controllably passivated by Si substitutional atoms. We further demonstrate that these Si defects may be directed through the lattice via e-beam control or modified (as yet, uncontrollably) to form new defects which can incorporate new atoms into the gr…
Niobium-based Superconducting Radio Frequency (SRF) cavity performance is sensitive to localized defects that give rise to quenches at high accelerating gradients. In order to identify these material defects on bulk Nb surfaces at their operating frequency and temperature, it is important to develop a new kind of wide bandwidth microwave microscopy with localized and strong RF magnetic fields. By taking advantage of write head technology widely used in the magnetic recording industry, one can obtain ~200 mT RF magnetic fields, which is on the order of the thermodynamic critical field of Nb, on…
A non-destructive optical technique described in this paper is an effective new tool for the investigation of defects in semiconductors. The basic instrument for this technique---a mid-IR-laser microscope---being sensitive to accumulations of free carriers enables the study of both accumulations of electrically-active defects or impurities in bulk semiconductors and doped domains in semiconductor structures. The optical beam induced scattering mode of this microscope is designed for the investigation of recombination-active defects but unlike EBIC it requires neither Schottky barrier or p--n j…
This article presents a numerical model dedicated to the simulation of field ion microscopy (FIM). FIM was the first technique to image individual atoms on the surface of a material. By a careful control of the field evaporation of the atoms from the surface, the bulk of the material exposed, and, through a digitally processing a sequence of micrographs, a three-dimensional reconstruction can be achieved. 3DFIM is particularly suited to the direct observation of crystalline defects such as vacancies, interstitials, vacancy clusters, dislocations, and any combinations of theses defects that und…
Quantitative measurements of extended defects in crystalline materials are important in understanding material behaviour. X-ray line profile analysis provides a complement to direct counting in the electron microscope, but is an indirect method and requires validation. Previous studies have focused on comparing x-ray analysis to electron microscopy results. Instead, we use simulated defective material with known defect content and apply line profile analysis to calculated diffraction profiles to directly show that line profile analysis can reliably quantify dislocations and stacking faults.
Samples for single-emitter spectroscopy are usually prepared by spin-coating a dilute solution of emitters on a microscope cover slip of silicate based glass (such as quartz). Here, we show that both borosilicate glass and quartz contain intrinsic defect colour centres that fluoresce when excited at 532 nm. In a microscope image the defect emission is indistinguishable from spin-coated emitters. The emission spectrum is characterised by multiple peaks, most likely due to coupling to a silica vibration with an energy of 160-180 meV. The defects are single-photon emitters, do not blink, and have…
Iron diantimonide is a material with the highest known thermoelectric power. By combining scanning transmission electron microscope study with electronic transport neutron, X-ray scattering and first principle calculation we identify atomic defects that control colossal thermopower magnitude and nanoprecipitate clusters with Sb vacancy ordering which induce additional phonon scattering and substantially reduce thermal conductivity. Defects are found to cause rather weak but important monoclinic distortion of the unit cell Pnnm to Pm. The absence of Sb along [010] for high defect concentration …
Artificial intelligence (AI) and machine learning have promised to revolutionize the way we live and work, and one of particularly promising areas for AI is image analysis. Nevertheless, many current AI applications focus on post-processing of data, while in both materials sciences and medicines, it is often critical to respond to the data acquired on the fly. Here we demonstrate an artificial intelligent atomic force microscope (AI-AFM) that is capable of not only pattern recognition and feature identification in ferroelectric materials and electrochemical systems, but can also respond to cla…
Dark-Field X-ray Microscopy (DFXM) is a diffraction-based synchrotron imaging techique capable of imaging defects in the bulk of extended crystalline samples. We present numerical simulations of image-formation in such a microscope using numerical integration of the dynamical Takagi-Taupin Equations (TTE) and wave front propagation. We validate our approach by comparing simulated images to experimental data from a near-perfect single crystal of diamond containing a single stacking fault defect in the illuminated volume.
Precise and scalable defect engineering of 2D nanomaterials is acutely sought-after in contemporary materials science. Here we present defect engineering in monolayer graphene and molybdenum disulfide (MoS$_2$) by irradiation with noble gas ions at 30 keV. Two ion species of different masses were used in a gas field ion source microscope: helium (He$^+$) and neon (Ne$^+$). A detailed study of the introduced defect sizes and resulting inter-defect distance with escalating ion dose was performed using Raman spectroscopy. Expanding on existing models, we found that the average defect size is cons…
University of Wisconsin-Madison（威斯康星大学麦迪逊分校）；University of Michigan - Ann Arbor（密歇根大学安娜堡分校）
In this work, we perform semantic segmentation of multiple defect types in electron microscopy images of irradiated FeCrAl alloys using a deep learning Mask Regional Convolutional Neural Network (Mask R-CNN) model. We conduct an in-depth analysis of key model performance statistics, with a focus on quantities such as predicted distributions of defect shapes, defect sizes, and defect areal densities relevant to informing modeling and understanding of irradiated Fe-based materials properties. To better understand the performance and present limitations of the model, we provide examples of useful…
Defects ~10 nm in size, with number densities ~10^{10} cm^{-2}, form spontaneously beneath ion-milled, etched, or HF-dipped silicon surfaces examined in our Ti-ion getter-pumped transmission electron microscope (TEM) after exposure to air. They appear as weakly-strained non-crystalline intrusions into silicon bulk, that show up best in the TEM under conditions of strong edge or bend contrast. If ambient air exposure is <10 minutes, defect nucleation and growth can be monitored {\em in situ}. Possible mechanisms of formation are discussed.
The scanning mid-IR-laser microscopy was previously demonstrated as an effective tool for characterization of different semiconductor crystals. Now the technique has been successfully applied for the investigation of CZ SixGe1-x -- a promising material for photovoltaics - and multicrystalline silicon for solar cells. In addition, this technique was shown to be appropriate for imaging of polishing-induced defects as well as such huge defects as "pin holes". Besides, previously unexplained "anomalous" (cubic power) dependence of signal of the scanning mid-IR-laser microscope in the optical-beam-…
The population of dislocation defects in a crystalline material strongly influences its properties, so the ability to analyse this population in experimental samples is of great utility. As a complement to direct counting in the transmission electron microscope, quantitative analysis of x-ray diffraction line profiles is an important tool. This is an indirect approach to quantification and so requires careful validation of the physical models that underly the inferential process. Here we undertake to directly evaluate the ability of line profile analysis to quantify aspects of the dislocation …
Deep-learning algorithms enable precise image recognition based on high-dimensional hierarchical image features. Here, we report the development and implementation of a deep-learning-based image segmentation algorithm in an autonomous robotic system to search for two-dimensional (2D) materials. We trained the neural network based on Mask-RCNN on annotated optical microscope images of 2D materials (graphene, hBN, MoS2, and WTe2). The inference algorithm is run on a 1024 x 1024 px2 optical microscope images for 200 ms, enabling the real-time detection of 2D materials. The detection process is ro…
Void defect is a possible origin of ferromagnetic like feature of pure carbon material. Applying density functional theory to void defect induced graphene nano ribbon (GNR), a detailed relationship between multiple spin state and structure change was studied. An equitorial triangle of an initial initial void having six electrons is distorted to isosceles triangle by rebonding carbon atoms. Among possible spin states, the most stable state was Sz=2/2. The case of Sz=4/2 is remarkable that initial flat ribbon turned to three dimentional curled one having highly polarized spin configuration at ri…
A method of the mid-IR-laser microscopy has been proposed for the investigation of the large-scale electrically and recombination active defects in semiconductors and non-destructive inspection of semiconductor materials and structures in the industries of microelectronics and photovoltaics. The basis for this development was laid with a wide cycle of the investigations on the low-angle mid-IR-light scattering in semiconductors. The essence of the technical idea was to apply the dark-field method for spatial filtering of the scattered light in the scanning mid-IR-laser microscope. This approac…
The defect-rich morphology of YBa$_{2}$Cu$_{3}$O$_{7-δ}$ (YBCO) thin films leads to a glass-like arrangement of Abrikosov vortices which causes the resistance to disappear in vanishing current densities. This vortex glass consists of entangled vortex lines and is identified by a characteristic scaling of the voltage-current isotherms. Randomly distributed columnar defects stratify the vortex lines and lead to a Bose glass. Here, we report on the observation of an ordered Bose glass in a YBCO thin film with a hexagonal array of columnar defects with 30 nm spacings. The periodic pinning landscap…
A method of scanning mid-IR-laser microscopy has recently been proposed for the investigation of large-scale electrically and recombination-active defects in semiconductors and non-destructive inspection of semiconductor materials and structures in the industries of microelectronics and photovoltaics. The basis for this development was laid with a wide cycle of investigations on low-angle mid-IR-light scattering in semiconductors. The essence of the technical idea was to apply the dark-field method for spatial filtering of the scattered light in the scanning mid-IR-laser microscope together wi…
Despite the recognition of two-dimensional (2D) systems as emerging and scalable host materials of single photon emitters or spin qubits, uncontrolled and undetermined chemical nature of these quantum defects has been a roadblock to further development. Leveraging the design of extrinsic defects can circumvent these persistent issues and provide an ultimate solution. Here we established a complete theoretical framework to accurately and systematically design quantum defects in wide-bandgap 2D systems. With this approach, essential static and dynamical properties are equally considered for spin…
Population and distribution of defects is one of the primary parameters controlling materials functionality, are often non-ergodic and strongly dependent on synthesis history, and are rarely amenable to direct theoretical prediction. Here, dynamic electron beam-induced transformations in Si deposited on a graphene monolayer are used to create libraries of the possible Si and carbon vacancy defects. Automated image analysis and recognition based on deep learning networks is developed to identify and enumerate the defects, creating a library of (meta) stable defect configurations. The electronic…
