Scanning Tunneling Microscopy Images Research Articles

Autonomous Scanning Tunneling Microscopy Imaging via Deep Learning.

Scanning tunneling microscopy (STM) is a powerful technique that provides the ability to manipulate and characterize individual atoms and molecules with atomic-level precision. However, the processes of scanning samples, operating the probe, and analyzing data are typically labor-intensive and subjective. Deep learning (DL) techniques have shown immense potential in automating complex tasks and solving high-dimensional problems. In this study, we developed an autonomous STM framework powered by DL to enable autonomous operations of the STM without human interventions. Our framework employs a convolutional neural network (CNN) for real-time evaluation of STM image quality, a U-net model for identifying bare surfaces, and a deep Q-learning network (DQN) agent for autonomous probe conditioning. Additionally, we integrated an object recognition model for the automated recognition of different adsorbates. This autonomous framework enables the acquisition of space-averaging information using STM techniques without compromising the high-resolution molecular imaging. We achieved measuring an area of approximately 1.9 μm2 within 48 h of continuous measurement and automatedly generated the statistics on the molecular species present within the mesoscopic area. We demonstrate the high robustness of the framework by conducting measurements at the liquid nitrogen temperature (∼78 K). We envision that the integration of DL techniques and high-resolution microscopy will not only extend the functionality and capability of scanning probe microscopes but also accelerate the understanding and discovery of new materials.

Electrochemical Surface Science: Self-assembly of Porphyrin molecules at single crystal metal electrodes

A detailed understanding of properties and processes at surfaces and interfaces requires at least two types of most basic information, chemical composition and –distribution as well as structure. While surface science in ultrahigh vacuum is blessed with a plethora of high sensitivity and highest spatial and temporal resolution due to the free accessibility of the surfaces by any kind of probe beams, investigations of solid surfaces under ambient conditions, i.e. in contact with gases or liquids, were for a long time restricted to the use of integral photon-based reflection-, absorption-, emission-, and diffraction methods. This “methodological gap” between UHV surface science and environmental interface research became immediately, at least partially, closed after the realization of the scanning tunneling microscope (STM) and following variants of proximity probes (SPM). The full applicability of this class of methods also under ambient conditions opened the door to structure information of solid-liquid interfaces of comparable resolution as in UHV at room temperature, a “quantum leap” for the understanding of e.g. interfacial electrochemistry. This, in turn, highlighted the need of reliable determination of the chemical composition and distribution at solid-liquid interfaces and pushed the development of in situ X-ray photoelectron spectroscopy (XPS).The availability of both techniques, in situ SPM and in situ XPS closes the former methodological gap between the research in UHV and under ambient conditions. In particular, interfacial electrochemistry, being primarily interested in chemical processes at electrode/electrolyte interfaces benefits decisively of this development.In this article, as an example, we present systematic in situ STM measurements and results on the interactions and self-assembly of porphyrins at anion modified metal/electrolyte interfaces, an important class of molecules for the functionalization of surfaces for various applications. Atomically and sub-molecularly resolved potentiostatic and potentiodynamic in situ STM images of such molecular layers are nowadays standard and wait for an in-depth theoretical analysis.

Lanthanide Atoms Induce Strong Graphene Sheet Distortion When Adsorbed on Stone-Wales Defects.

Local curvature in graphene can enhance its reactivity and catalytic activity and can be induced by the adsorption of certain chemical species. By employing periodic density functional theory (DFT) calculations, we demonstrate that significant local curvature can be systematically observed when lanthanide atoms (the full series from La to Lu) are adsorbed on the Stone-Wales (SW) defect in graphene, contrary to that in defect-free graphene. Despite the typical high coordination numbers of lanthanide species, their hapticity is always η2 (and not η5, η6, or η7), where Ln atoms are adsorbed on the (7,7) junction, forming relatively short Ln···C separations. Contrary to the pristine graphene, the SW region undergoes considerable distortion and results in much stronger Ln bonding. The positive charge acquired by Ln atoms upon adsorption on SW is approximately 1.5 times larger than that on defect-free graphene. The high visibility of electron-rich lanthanide species in scanning tunneling microscopy images provides a means to locate SW defects in graphene samples experimentally.

Structure and Defect Identification at Self-Assembled Islands of CO2 Using Scanning Probe Microscopy.

Understanding how carbon dioxide (CO2) behaves and interacts with surfaces is paramount for the development of sensors and materials to attempt CO2 mitigation and catalysis. Here, we combine simultaneous atomic force microscopy (AFM) and scanning tunneling microscopy (STM) using CO-functionalized probes with density functional theory (DFT)-based simulations to gain fundamental insight into the behavior of physisorbed CO2 molecules on a gold(111) surface that also contains one-dimensional metal-organic chains formed by 1,4-phenylene diisocyanide (PDI) bridged by gold (Au) adatoms. We resolve the structure of self-assembled CO2 islands, both confined between the PDI-Au chains as well as free-standing on the surface and reveal a chiral arrangement of CO2 molecules in a windmill-like structure that encloses a standing-up CO2 molecule and other foreign species existing at the surface. We identify these species by the comparison of height-dependent AFM and STM imaging with DFT-calculated images and clarify the origin of the kagome tiling exhibited by this surface system. Our results show the complementarity of AFM and STM using functionalized probes and their potential, when combined with DFT, to explore greenhouse gas molecules at surface-supported model systems.

Low defect density in MoS2 monolayers grown on Au(111) by metal-organic chemical vapor deposition

Monolayers of transition metal dichalcogenides (TMDs) possess high potential for applications in novel electronic and optoelectronic devices and therefore the development of methods for their scalable growth is of high importance. Among different suggested approaches, metal-organic chemical vapor deposition (MOCVD) is the most promising one for technological applications because of its lower growth temperature compared to the most other methods, e.g., conventional chemical vapor or atomic layer deposition (CVD, ALD). Here we demonstrate for the first time the epitaxial growth of MoS2 monolayers on Au(111) by MOCVD at 450 °C. We confirm the high quality of the grown TMD monolayers down to the atomic scale using several complementary methods. These include Raman spectroscopy, non-contact atomic force microscopy (nc-AFM), X-ray photoelectron spectroscopy and scanning tunneling microscopy (STM). The topographic corrugation of the MoS2 monolayer on Au(111), revealed in a moiré structure, was measured as ≈20 pm by nc-AFM. The estimated defect density calculated from STM images of the as-grown MoS2 monolayers is in the order of 1012 vacancies/cm2. The defects are mainly caused by single sulfur vacancies. Our approach is a step forward towards the technologically relevant growth of high-quality, large-area TMD monolayers.

Classification of adsorbates in scanning tunneling microscopy images of Fe3O4(111) surfaces exposed to water and carbon monoxide

Understanding the structure of catalyst surfaces with adsorbed molecules is key to improving catalyst design. Scanning tunneling microscopy (STM) allows the observation of adsorption states and sites and provides insights into diffusion and desorption processes; however, the presence of multiple types of molecules on the surface presents challenges such as the identification of species and verification of reaction progress, particularly at room temperature or higher. In this study, we develop a protocol for the height classification analysis of STM images using the Watershed algorithm. This method is applied to a system involving the co-adsorption of H2O and CO on the Fe3O4(111) surface, which represents the beginning of the water-gas shift reaction. Water molecules and dissociated OH species were identified in STM images of the Fe3O4(111) surface following the adsorption of water. Furthermore, gradual changes in the types of surface species were observed upon exposure of the surface to CO, indicating reaction progression. Our observations suggest that CO may react with molecular water rather than with dissociated OH on Fe sites. Despite its simplicity, the height classification analysis effectively identifies changes in the adsorbates on the catalyst surface. This method can be extended to other catalyst surfaces with adsorbed gasses.

Data-Driven Structure Recognition of Scanning Tunneling Microscopy Images in a Case of Iron Carbide.

Scanning tunneling microscopy (STM) serves as a critical tool for high-resolution surface imaging, yet deciphering the atomic structures from STM images on multielement surfaces, such as oxides and carbides, remains a challenging task that heavily relies on the expertise and intuition of researchers. In this study, we introduce a data-driven method for rapid structural recognition from STM images. This method involves extracting structural features, filtering through a structural database, and matching with simulated STM images and surface energy analyses, thereby providing researchers with several of the most probable structures. We demonstrate the capabilities of this technique using our previously reported iron carbide grown on an Fe(110) crystal. By proposing a candidate structure set and establishing a comprehensive database linking STM images to corresponding structures and surface energies, we selected 6 out of more than 10 000 possible surfaces. On the basis of these 6 recommendations, researchers can conveniently determine the real surface structures. Our work provides an efficient tool for the structure recognition of STM images to construct surface structures, potentially serving as a universal auxiliary tool for STM structural analysis.

The organization of pyridazine adsorbed on Au(1 1 1) electrode in sulfuric and perchloric acids – As probed by in situ STM

The adsorption of pyridazine (PD) on a gold electrode served as a model to study the orientation and arrangement of an organic adsorbate bearing a heterocyclic molecular structure at an electrified interface. Because the two N-ends of PD could interact with a Au electrode, the acid-base equilibrium of PD at an electrified interface can be different from that in a solution phase. Unprotonated and protonated PD interacted with the Au electrode differently, leading to dissimilar spatial structures on the Au(111) electrode. Moreover, having a large dipole moment of 4.22 D, PD adopted different adsorption configurations, as the Au potential was modulated. The indispensable anion present in the supporting electrolyte can compete with PD for surface sites on the Au electrode. Since the potential of a charged conductor is established by the physical contact with an electrolyte, the study of the adsorption of PD on a Au(111) electrode calls for the use of an in situ tool, such as a scanning tunneling microscope (STM), which provides sub-nanometer information of the interface in real-time and real-space. Different phases of adsorbed PD molecules on an ordered Au(111) electrode were revealed as a function of potential, pH and anion. The most notable event for PD adsorbed on Au(111) is attributed to the sharp transition from a disordered state to highly organized structures in 0.1 M H2SO4 and HClO4. The obtained STM images enabled a direct measurement of the spacing between PD admolecules, from which the molecular reorientations on the Au electrode were inferred. PD admolecules flipped from the horizontal to the upright configuration, tethered to Au substrate via the two N-ends, at positive potentials. Different ordered PD structures seen in H2SO4 and HClO4 imply anions were coadsorbed with PD molecule on the Au electrode. The acid-base equilibrium of PD at the Au electrode was examined with STM in pH 1 and 3 media.

Physics and chemistry from parsimonious representations: image analysis via invariant variational autoencoders

Electron, optical, and scanning probe microscopy methods are generating ever increasing volume of image data containing information on atomic and mesoscale structures and functionalities. This necessitates the development of the machine learning methods for discovery of physical and chemical phenomena from the data, such as manifestations of symmetry breaking phenomena in electron and scanning tunneling microscopy images, or variability of the nanoparticles. Variational autoencoders (VAEs) are emerging as a powerful paradigm for the unsupervised data analysis, allowing to disentangle the factors of variability and discover optimal parsimonious representation. Here, we summarize recent developments in VAEs, covering the basic principles and intuition behind the VAEs. The invariant VAEs are introduced as an approach to accommodate scale and translation invariances present in imaging data and separate known factors of variations from the ones to be discovered. We further describe the opportunities enabled by the control over VAE architecture, including conditional, semi-supervised, and joint VAEs. Several case studies of VAE applications for toy models and experimental datasets in Scanning Transmission Electron Microscopy are discussed, emphasizing the deep connection between VAE and basic physical principles. Python codes and datasets discussed in this article are available at https://github.com/saimani5/VAE-tutorials and can be used by researchers as an application guide when applying these to their own datasets.

Challenges to extracting spatial information about double P dopants in Si from STM images

The design and implementation of dopant-based silicon nanoscale devices rely heavily on knowing precisely the locations of phosphorous dopants in their host crystal. One potential solution combines scanning tunneling microscopy (STM) imaging with atomistic tight-binding simulations to reverse-engineer dopant coordinates. This work shows that such an approach may not be straightforwardly extended to double-dopant systems. We find that the ground (quasi-molecular) state of a pair of coupled phosphorous dopants often cannot be fully explained by the linear combination of single-dopant ground states. Although the contributions from excited single-dopant states are relatively small, they can lead to ambiguity in determining individual dopant positions from a multi-dopant STM image. To overcome that, we exploit knowledge about dopant-pair wave functions and propose a simple yet effective scheme for finding double-dopant positions based on STM images.

Determining the density and spatial descriptors of atomic scale defects of 2H–WSe2 with ensemble deep learning

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.

Cryogen-free modular scanning tunneling microscope operating at 4-K in high magnetic field on a compact ultra-high vacuum platform.

One of the daunting challenges in modern low temperature scanning tunneling microscopy (STM) is the difficulty of combining atomic resolution with cryogen-free cooling. Further functionality needs, such as ultra-high vacuum (UHV), high magnetic field (HF), and compatibility with μm-sized samples, pose additional challenges to an already ambitious build. We present the design, construction, and performance of a cryogen-free, UHV, low temperature, and high magnetic field system for modular STM operation. An internal vibration isolator reduces vibrations in this system, allowing for atomic resolution STM imaging while maintaining a low base temperature of ∼4K and magnetic fields up to 9T. Samples and tips can be conditioned in situ utilizing a heating stage, an ion sputtering gun, an e-beam evaporator, a tip treater, and sample exfoliation. In situ sample and tip exchange and alignment are performed in a connected UHV room temperature stage with optical access. Multisite operation without breaking vacuum is enabled by a unique quick-connect STM head design. A novel low-profile vertical transfer mechanism permits transferring the STM between room temperature and the low temperature cryostat.

Provably Trainable Rotationally Equivariant Quantum Machine Learning

Exploiting the power of quantum computation to realize superior machine learning algorithms has been a major research focus of recent years, but the prospects of quantum machine learning (QML) remain dampened by considerable technical challenges. A particularly significant issue is that generic QML models suffer from so-called barren plateaus in their training landscapes—large regions where cost function gradients vanish exponentially in the number of qubits employed, rendering large models effectively untrainable. A leading strategy for combating this effect is to build problem-specific models that take into account the symmetries of their data in order to focus on a smaller, relevant subset of Hilbert space. In this work, we introduce a family of rotationally equivariant QML models built upon the quantum Fourier transform, and leverage recent insights from the Lie-algebraic study of QML models to prove that (a subset of) our models do not exhibit barren plateaus. In addition to our analytical results we numerically test our rotationally equivariant models on a dataset of simulated scanning tunneling microscope images of phosphorus impurities in silicon, where rotational symmetry naturally arises, and find that they dramatically outperform their generic counterparts in practice. Published by the American Physical Society 2024

Machine Learning-Assisted Precision Manufacturing of Atom Qubits in Silicon.

Donor-based qubits in silicon, manufactured using scanning tunneling microscope (STM) lithography, provide a promising route to realizing full-scale quantum computing architectures. This is due to the precision of donor placement, long coherence times, and scalability of the silicon material platform. The properties of multiatom quantum dot qubits, however, depend on the exact number and location of the donor atoms within the quantum dots. In this work, we develop machine learning techniques that allow accurate and real-time prediction of the donor number at the qubit site during STM patterning. Machine learning image recognition is used to determine the probability distribution of donor numbers at the qubit site directly from STM images during device manufacturing. Models in excess of 90% accuracy are found to be consistently achieved by mitigating overfitting through reduced model complexity, image preprocessing, data augmentation, and examination of the intermediate layers of the convolutional neural networks. The results presented in this paper constitute an important milestone in automating the manufacture of atom-based qubits for computation and sensing applications.

Self-assembled oligomeric structures of an asymmetric molecular linker; 4-isocyanophenyl disulfide on Au(111)

Para-substituted benzenes, such as 1,4-benzene dithiol and 1,4-phenyl diisocyanide, have been observed to oligomerize on the Au(111) surface by incorporating gold adatoms extracted from the substrate. This work investigates if oligomerization occurs for an analogous but asymmetric linker, 4-isocyanophenyl disulfide (ICPD) on Au(111). This molecule is comprised of both disulfide and isocyanide terminal groups attached to the phenyl ring. The resulting surface structures formed on Au(111) following exposure to ICPD are studied using scanning tunneling microscopy (STM). 1,4-isocyanophenyl thiolate (ICPT), formed through scission of ICPD’s disulfide bond, was also found to oligomerize on the surface, and potential oligomer structures and binding geometries are proposed with the aid of density functional theory (DFT) calculations, along with simulated STM images of the resulting structures. It is observed in this work that ICPT forms oligomeric structures that cover large sections of the substrate and appear to create etch pits resulting from gold atom extraction. Numerous potential binding geometries are investigated based on the distances between substrate gold atom adsorption sites compared to the monomer length. Selected structural candidates were optimized using DFT and were used to generate simulated STM images using the Tersoff–Hamann method to compare with experiment. It has been shown previously that the isocyanide- and thiol-connected oligomers conduct electrons, suggesting the possibility that the asymmetric oligomers found here might form the basis for fabricating molecular diodes.

Prediction of single-layer antimony oxyselenide (Sb2O2Se2): metal-to-semiconductor transition via hydrogenation

In this study, the structural, electronic, vibrational, and mechanical properties of single-layer Antimony Oxyselenide (Sb2O2Se2) and its hydrogenated structure (Sb2O2Se2H2) are investigated by performing density functional theory-based first principles calculations. Geometry optimizations reveal that single-layer Sb2O2Se2 crystallizes in tetragonal structure which is shown to possess dynamical stability by means of phonon band dispersions. In addition, the mechanical stability of the predicted single layer is satisfied via the linear-elastic parameters. Electronically, it is revealed that single-layer Sb2O2Se2 exhibits metallic behavior whose highest occupied states are found to arise from the surface Se atoms, may be an indication for tuning the electronic features via surface functionalization. For the surface modification of Sb2O2Se2, top of each Se atom is saturated with a H atom and fully hydrogenated single-layer Sb2O2Se2H2 is shown to be an in-plane anisotropic structure. Phonon band dispersion calculations indicate the dynamical stability of Sb2O2Se2H2. Mechanically stable Sb2O2Se2H2 is found to possess anisotropic linear-elastic behavior, which is much softer than its pristine structure. Moreover, electronically a metallic-to-semiconducting transition is shown to occur as the unoccupied Se-orbitals are saturated via H atoms. Our work offers insights into prediction of a novel single-layer material, namely Sb2O2Se2, and reports the chemically-driven semiconducting behavior via hydrogenation, which may lead to the use of hydrogenated structure in solar cell, photoelectrode, or photocatalyst applications owing to its suitable band gap.

Vacancy Ordering in Ultrathin Copper Oxide Films on Cu(111).

Oxide thin films grown on metal surfaces have wide applications in catalysis and beyond owing to their unique surface structures compared to their bulk counterparts. Despite extensive studies, the atomic structures of copper surface oxides on Cu(111), commonly referred to as "44" and "29", have remained elusive. In this work, we demonstrated an approach for the structural determination of oxide surfaces using element-specific scanning tunneling microscopy (STM) imaging enhanced by functionalized tips. This approach enabled us to resolve the atomic structures of "44" and "29" surface oxides, which were further corroborated by noncontact atomic force microscopy (nc-AFM) measurements and Monte Carlo (MC) simulations. The stoichiometry of the "44" and "29" frameworks was identified as Cu23O16 and Cu16O11, respectively. Contrary to the conventional hypothesis, we observed ordered Cu vacancies within the "44" structure manifesting as peanut-shaped cavities in the hexagonal lattice. Similarly, a combination of Cu and O vacancies within the "29" structure leads to bean-shaped cavities within the pentagonal lattice. Our study has thus resolved the decades-long controversy on the atomic structures of "44" and "29" surface oxides, advancing our understanding of copper oxidation processes and introducing a robust framework for the analysis of complex oxide surfaces.

Formation of Supernarrow Borophene Nanoribbons

AbstractBorophenes have sparked considerable interest owing to their fascinating physical characteristics and diverse polymorphism. However, borophene nanoribbons (BNRs) with widths less than 2 nm have not been achieved. Herein, we report the experimental realization of supernarrow BNRs. Combining scanning tunneling microscopy imaging with density functional theory modeling and ab initio molecular dynamics simulations, we demonstrate that, under the applied growth conditions, boron atoms can penetrate the outermost layer of Au(111) and form BNRs composed of a pair of zigzag (2,2) boron rows. The BNRs have a width self‐contained to ∼1 nm and dipoles at the edges to keep them separated. They are embedded in the outermost Au layer and shielded on top by the evacuated Au atoms, free of the need for post‐passivation. Scanning tunneling spectroscopy reveals distinct edge states, primarily attributed to the localized spin at the BNRs’ zigzag edges. This work adds a new member to the boron material family and introduces a new physical feature to borophenes.

Formation of Supernarrow Borophene Nanoribbons.

Borophenes have sparked considerable interest owing to their fascinating physical characteristics and diverse polymorphism. However, borophene nanoribbons (BNRs) with widths less than 2 nm have not been achieved. Herein, we report the experimental realization of supernarrow BNRs. Combining scanning tunneling microscopy imaging with density functional theory modeling and ab initio molecular dynamics simulations, we demonstrate that, under the applied growth conditions, boron atoms can penetrate the outermost layer of Au(111) and form BNRs composed of a pair of zigzag (2,2) boron rows. The BNRs have a width self-contained to ∼1 nm and dipoles at the edges to keep them separated. They are embedded in the outermost Au layer and shielded on top by the evacuated Au atoms, free of the need for post-passivation. Scanning tunneling spectroscopy reveals distinct edge states, primarily attributed to the localized spin at the BNRs' zigzag edges. This work adds a new member to the boron material family and introduces a new physical feature to borophenes.

Studies of indium nanostructures growth models on A3B6 layered templates

This article explores the surface architecture formation on layered A3B6 semiconductors, such as In4Se3, InSe, and InTe, focusing on kinetics of surface roughness parameters, such as RMS, skewness and kurtosis, and power spectral density (PSD) under indium deposition studied through large-scale 1 × 1 µm2 and 150 × 150 nm2 scanning tunneling microscopy (STM) images analysis. Our findings reveal common kinetics of surface parameters across different crystals, suggesting fundamental similarities in their interface kinetics. Particularly, PSD analysis elucidates the role of low-frequency spatial structures, underscoring their significance in surface roughness kinetics to some extent independent of the specific layered crystal surface structure.