Condensed Matter

The adsorption of AlCl 3 on Si(100) and the effect of annealing the AlCl 3 -dosed substrate was studied to reveal key surface processes for the development of atomic-precision acceptor-doping techniques. This investigation was performed via scanning tunneling microscopy (STM), X-ray photoelectron spectroscopy (XPS), and density functional theory (DFT) calculations. At room temperature, AlCl 3 readily adsorbed to the Si substrate dimers and dissociated to form a variety of species. Annealing of the AlCl 3 -dosed substrate at temperatures below 450 ??C produced unique chlorinated aluminum chains (CACs) elongated along the Si(100) dimer row direction. An atomic model for the chains is proposed with supporting DFT calculations. Al was incorporated into the Si substrate upon annealing at 450 ??C and above, and Cl desorption was observed for temperatures beyond 450 ??C. Al-incorporated samples were encapsulated in Si and characterized by secondary ion mass spectrometry (SIMS) depth profiling to quantify the Al atom concentration, which was found to be in excess of 10 20 cm ?? across a ??2.7 nm thick δ -doped region. The Al concentration achieved here and the processing parameters utilized promote AlCl 3 as a viable gaseous precursor for novel acceptor-doped Si materials and devices for quantum computing.

Linearized augmented planewaves combined with local-orbitals (LAPW+lo) are arguably the most precise basis set to represent Kohn-Sham states. When employed within real-time time-dependent density functional theory (RT-TDDFT), they promise ultimate precision achievable for exploring the evolution of electronic excitations. In this work, we present an implementation of RT-TDDFT in the full-potential LAPW+lo code exciting. We benchmark our results against those obtained by linear-response TDDFT with exciting and by RT-TDDFT calculations with the Octopus code, finding a satisfactory level of agreement. To illustrate possible applications of our implementation, we have chosen three examples: the dynamic behavior of excitations in MoS 2 induced by a laser pulse, the third harmonic generation in silicon, and a pump-probe experiment in diamond.

Crystal structure prediction is one of the major unsolved problems in materials science. Traditionally, this problem is formulated as a global optimization problem for which global search algorithms are combined with first principle free energy calculations to predict the ground-state crystal structure given only a material composition or a chemical system. These ab initio algorithms usually cannot exploit a large amount of implicit physicochemical rules or geometric constraints (deep knowledge) of atom configurations embodied in a large number of known crystal structures. Inspired by the deep learning enabled breakthrough in protein structure prediction, herein we propose AlphaCrystal, a crystal structure prediction algorithm that combines a deep residual neural network model that learns deep knowledge to guide predicting the atomic contact map of a target crystal material followed by reconstructing its 3D crystal structure using genetic algorithms. Based on the experiments of a selected set of benchmark crystal materials, we show that our AlphaCrystal algorithm can predict structures close to the ground truth structures. It can also speed up the crystal structure prediction process by predicting and exploiting the predicted contact map so that it has the potential to handle relatively large systems. We believe that our deep learning based ab initio crystal structure prediction method that learns from existing material structures can be used to scale up current crystal structure prediction practice. To our knowledge, AlphaCrystal is the first neural network based algorithm for crystal structure contact map prediction and the first method for directly reconstructing crystal structures from materials composition, which can be further optimized by DFT calculations.

Atomic layer deposition (ALD) is an essential tool in semiconductor device fabrication that allows the growth of ultrathin and conformal films to precisely form heterostructures and tune interface properties. The self-limiting nature of the chemical reactions during ALD provides excellent control over the layer thickness. However, in contrast to idealized growth models, it is experimentally challenging to create continuous monolayers by ALD because surface inhomogeneities and precursor steric interactions result in island growth during film nucleation. Thus, the ability to create pin-hole free monolayers by ALD would offer new opportunities for controlling interfacial charge and mass transport in semiconductor devices, as well as for tailoring surface chemistry. Here, we report full encapsulation of c-plane gallium nitride (GaN) with an ultimately thin (~3 ?) aluminum oxide (AlOx) monolayer, which is enabled by the partial conversion of the GaN surface oxide into AlOx using a combination of trimethylaluminum deposition and hydrogen plasma exposure. Introduction of monolayer AlOx significantly modifies the physical and chemical properties of the surface, decreasing the work function and introducing new chemical reactivity to the GaN surface. This tunable interfacial chemistry is highlighted by the reactivity of the modified surface with phosphonic acids under standard conditions, which results in self-assembled monolayers with densities approaching the theoretical limit. More broadly, the presented monolayer AlOx deposition scheme can be extended to other dielectrics and III-V-based semiconductors, with significant relevance for applications in optoelectronics, chemical sensing, and (photo)electrocatalysis.

Atomic scale friction, an indispensable element of nanotechnology, requires a direct access to, under actual growing shear stress, its successive live phases: from static pinning, to depinning and transient evolution, eventually ushering in steady state kinetic friction. Standard tip-based atomic force microscopy generally addresses the steady state, but the prior intermediate steps are much less explored. Here we present an experimental and simulation approach, where an oscillatory shear force of increasing amplitude leads to a one-shot investigation of all these successive aspects. Demonstration with controlled gold nanocontacts sliding on graphite uncovers phenomena that bridge the gap between initial depinning and large speed sliding, of potential relevance for atomic scale time and magnitude dependent rheology.

We analytically calculate the electronic friction tensor for a molecule near a metal surface in the case that the electronic Hamiltonian is complex-valued, e.g. the case that there is spin-orbit coupling and/or an external magnetic field. In such a case, {\em even at equilibrium}, we show that the friction tensor is not symmetric. Instead, the tensor is the real-valued sum of one positive definite tensor (corresponding to dissipation) plus one antisymmetric tensor (corresponding to a Berry pseudomagnetic force). Moreover, we find that this Berry force can be much larger than the dissipational force, suggesting the possibility of strongly spin-polarized chemicurrents or strongly spin-dependent rate constants for systems with spin-orbit coupling.

The performance of machine learning interatomic potentials relies on the quality of the training dataset. In this work, we present an approach for generating diverse and representative training data points which initiates with \it{ab initio} calculations for bulk structures. The data generation and potential construction further proceed side-by-side in a cyclic process of training the neural network and crystal structure prediction based on the developed interatomic potentials. All steps of the data generation and potential development are performed with minimal human intervention. We show the reliability of our approach by assessing the performance of neural network potentials developed for two inorganic systems.

Complementary to the development of highly three-dimensional (3D) integrated circuits in the continuation of Moore's law, there has been a growing interest in new 3D deformation strategies to improve device performance. To continue this search for new 3D deformation techniques, it is essential to explore beforehand - using computational predictive methods - which strain tensor leads to the desired properties. In this work, we study germanium (Ge) under an isotropic 3D strain on the basis of first-principle methods. The transport and optical properties are studied by a fully ab initio Boltzmann transport equation and many-body Bethe-Salpeter equation (BSE) approach, respectively. Our findings show that a direct band gap in Ge could be realized with only 0.34% triaxial tensile strain (negative pressure) and without the challenges associated with Sn doping. At the same time a significant increase in refractive index and carrier mobility - particularly for electrons - is observed. These results demonstrate that there is a huge potential in exploring the 3D deformation space for semiconductors - and potentially many other materials - in order to optimize their properties.

Stochastic microstructure reconstruction involves digital generation of microstructures that match key statistics and characteristics of a (set of) target microstructure(s). This process enables computational analyses on ensembles of microstructures without having to perform exhaustive and costly experimental characterizations. Statistical functions-based and deep learning-based methods are among the stochastic microstructure reconstruction approaches applicable to a wide range of material systems. In this paper, we integrate statistical descriptors as well as feature maps from a pre-trained deep neural network into an overall loss function for an optimization based reconstruction procedure. This helps us to achieve significant computational efficiency in reconstructing microstructures that retain the critically important physical properties of the target microstructure. A numerical example for the microstructure reconstruction of bi-phase random porous ceramic material demonstrates the efficiency of the proposed methodology. We further perform a detailed finite element analysis (FEA) of the reconstructed microstructures to calculate effective elastic modulus, effective thermal conductivity and effective hydraulic conductivity, in order to analyse the algorithm's capacity to capture the variability of these material properties with respect to those of the target microstructure. This method provides an economic, efficient and easy-to-use approach for reconstructing random multiphase materials in 2D which has the potential to be extended to 3D structures.

Layered optical devices with a lateral gradient can be fabricated through electrochemical synthesis of porous silicon (PS) using a position dependent etching current density $\bm j(\bm r_\|)$. Predicting the local value of $\bm j(\bm r_\|)$ and the corresponding porosity $p(\bm r_\|)$ and etching rate $v(\bm r_\|)$ is desirable for their systematic design. We develop a simple analytical model for the calculation of $\bm j(\bm r_\|)$ within a prism shaped cell. Graded single layer PS samples were synthesized and their local calibration curves p vs $\bm j$ and v vs $\bm j$ were obtained from our model and their reflectance spectra. The agreement found between the calibration curves from different samples shows that from one sample we could obtain full calibration curves which may be used to predict, design, and fabricate more complex non-homogeneous multilayered devices with lateral gradients for manifold applications.