In this work, we studied the thermal characteristics of flat heaters made of aluminum with a strip heating element in the form of carbon fiber. In order to provide the necessary insulation of the heating element from the metal base, a layer of porous anodic aluminum oxide with a thickness of 20 μm was formed on the aluminum surface. The ends of the carbon fiber filament were metallized with a layer of copper for subsequent soldering during the assembly of the electric heater. The carbon fiber filament of electric heater had an electrical resistance of 60 Ohms. Studies of the propagation of heat fluxes in the volume of a board made of aluminum with nanoporous aluminum oxide were carried out using thermal imaging measurements. The paper presents the dependence of temperature changes on the surface of the lid of a heating element made of aluminum and on the opposite side — heat transfer side with heating time. The results showed that the heat generated by a linear heating element of carbon fiber, quickly distributed throughout the entire volume of the aluminum plate of the heating element. This indicates a high thermal conductivity of the aluminum base of the heater, the parameters of which allow to achieve the required thermal characteristics of the heater.

Models that describe bipolar resistive switching in planar microstructures based on oxide compounds (Bi2Sr2CaCu2O8+x, Nd2-xCexCuO4-y) and bismuth selenide are considered. Metal-isolator-metal planar-type meristor heterostructures were investigated, in which the micro-size is formed by an electrode whose diameter is much smaller than the total size of the structure (it can be both Chervinsky-type microjunctions and film electric electrodes). Another important feature of these heterostructures is the presence of a surface layer several tens of nanometers thick with specific conductivity significantly reduced relative to volume. The change in the resistive properties of such heterostructures is caused by the formation or destruction of the conductive channel through the above-mentioned layer. Numerical simulation has shown that the bipolar resistive switching is significantly influenced by the electrical field distribution topology. A “critical field” model is proposed to describe experimentally observed memristor effects in investigated heterostructures. In this model it is assumed that the change in specific conductivity occurs in those parts of the surface layer where the electric field strength exceeds some critical value. The model of the “critical field” is based on the numerical calculation of the distribution of electrical potential on the distribution of specific conductivity in the structure. In addition, the model allowing to analyze the influence of electrodiffusion of oxygen ions on resistive switching in heterostructures based on Bi2Sr2CaCu2O8+x is considered. At numerical realization of the models a combination of the integro-differential approximation of the differential equations, the multi-grid approach for localization of heterogeneities of physical characteristics, the iterative decomposition method and composite adaptive meshes was used. It allowed tracking the processes under investigation with necessary accuracy. The comparison of simulation results with experimental data is presented.

A general set of ideas related to the memristors modeling is presented. The memristor is considered to be a partially ordered physical and chemical system that is within the “edge of chaos“ from the point of view of nonlinear dynamics. The logical and historical relationship of memristor physics, nonlinear dynamics, and neuromorphic systems is illustrated in the form of a scheme. We distinguish the nonlinearity into external ones, when we describe the behavior of an electrical circuit containing a memristor, and internal ones, which are caused by processes in filament region. As a simulation model, the attention is drawn to the connectionist approach, known in the theory of neural networks, but applicable to describe the evolution of the filament as the dynamics of a network of traps connected electrically and quantum-mechanically. The state of each trap is discrete, and it is called an “oscillator“. The applied meaning of the theory of coupled maps lattice is indicated. The high-density current through the filament can lead to the need to take into account both discrete processes (generation of traps) and continuous processes (inclusion of some constructions of solid body theory into the model).
However, a compact model is further developed in which the state of such a network is aggregated to three phase variables: the length of the filament, its total charge, and the local temperature. Despite the apparent physical meaning, all variables have a formal character, which is usually inherent in the parameters of compact models. The model consists of one algebraic equation, two differential equations, and one integral connection equation, and is derived from the simplest Strukov’s model. Therefore, it uses the “window function” approach. It is indicated that, according to the Poincare—Bendixon theorem, this is sufficient to explain the instability of four key parameters (switching voltages and resistances ON/OFF) at a cycling of memristor. The Fourier spectra of the time series of these parameters are analyzed on a low sample of experimental data. The data are associated with the TiN/HfOx/Pt structure (0 < x < 2). A preliminary conclusion that requires further verification is the predominance of low frequencies and the stochasticity of occurrence ones.

The article discusses the use of hybrid HPC clusters for the execution of software designed to calculate the electronic structure and atomic scale materials modeling. Modern software systems, which are designed to solve the problems of materials science, use the capabilities of various hardware computing accelerators to increase productivity. The use of such computing technologies requires the adaptation of application program code to hybrid computing architectures, which include classic central processing units (CPUs) and specialized graphics accelerators (GPUs).
The use of large computing hybrid systems requires the development of methods for ensuring the workloading of such computing systems that will allow efficient use of computing resources and avoid equipment downtime.
First of all, these methods should allow parallel execution of user applications using computational accelerators. However, in practice, software environments designed to solve application problems cannot be deployed in the same computing environment due to software incompatibility. In order to overcome this limitation and ensure the parallel execution of diverse types of materials science tasks, the creation of individual task execution environments based on virtualization technologies and cloud technologies.
The continuation of virtualization technologies and the provision of cloud services is the construction of digital platforms. The article proposes the use of a digital platform for hosting scientific materials science services that provide calculations using various application software systems. Digital platforms make it possible to provide a unified user interface to scientific materials science services. The platform provides opportunities for finding the necessary scientific services, transferring source data and results between users, the platform and hybrid high-performance clusters.

Information about the structure and properties of materials is especially important when working with micro-and nanoscale objects due to the high complexity of their obtaining. This makes it relevant to use computer modeling to predict the required characteristics of materials. Electronic, magnetic, mechanical, and other properties of crystalline substances are determined by their structure-the periodicity of the lattice and the symmetry of the unit cell. This article discusses metal oxides with the general chemical formulas MeO (metals: Ca, Cd, Mg), MeO2 (metals: Hf, Ce, Zr), Me2O3 (metals: Er, Nd, Sc, Mn, Tl) and Me3O4 (using Fe as an example) and a cubic symmetry type crystal lattice — structural types NaCl (rock salt), Fluorite, Bixbyite, Spinel accordingly. The paper describes the model of ion-atomic radii, which is widely used in the modeling of crystalline metal oxides. The application of the annealing simulation algorithm for calculating the metric parameters of the compounds under consideration is shown. The software implementation of the algorithm presented in this paper allows us to determine the coordinates of the atoms that are included in the elementary cell of the crystal lattice, calculate the lattice constant and the density of the packing of atoms in the crystal cell using the specified chemical formula and the space group symmetry. These structural characteristics can be used as input parameters for determining electronic, magnetic, and other properties. The article compares the values of lattice constants obtained as a result of modeling with experimental data.

The property of natural parallelization of matrix-vector operations inherent in memristor crossbars creates opportunities for their effective use in neural network computing. Analog calculations are orders of magnitude faster in comparison to calculations on the central processor and on graphics accelerators. Besides, mathematical operations energy costs are significantly lower. The essential feature of analog computing is its low accuracy. In this regard, studying the dependence of neural network quality on the accuracy of setting its weights is relevant. The paper considers two convolutional neural networks trained on the MNIST (handwritten digits) and CIFAR_10 (airplanes, boats, cars, etc.) data sets. The first convolutional neural network consists of two convolutional layers, one subsample layer and two fully connected layers. The second one consists of four convolutional layers, two subsample layers and two fully connected layers. Calculations in convolutional and fully connected layers are performed through matrix-vector operations that are implemented on memristor crossbars. Sub-sampling layers imply the operation of finding the maximum value from several values. This operation can be implemented at the analog level. The process of training a neural network runs separately from data analysis. As a rule, gradient optimization methods are used at the training stage. It is advisable to perform calculations using these methods on CPU. When setting the weights, 3—4 precision bits are required to obtain an acceptable recognition quality in the case the network is trained on MNIST. 6-10 precision bits are required if the network is trained on CIFAR_10.

The article gives an overview of the main currently used models for the formation of photoresist masks and the problems in which they are applied. The main stages of «full physical» modeling of mask formation are briefly considered in the case of both traditional DNQ photoresists and CA photoresists. The concept of compact models (VT5 and CM1), which predict the contour of the resist mask for a full-sized device topology is considered. Examples of some calculations using both full physical modeling and compact models are given. Using a full physical modeling of the resist mask formation the lithographic stack was optimized for a promising technological process. The optimum thickness ratios for the binary BARC used in the water immersion lithographic process are found. The problem of determining the optimal number of calibration structures that maximally cover the space of aerial image parameters was solved. To solve this problem, cluster analysis was used. Clustering was carried out using the k-means method. The optimal sample size was from 300 to 350 structures, the mean square error in this case is 1.4 nm, which slightly exceeds the noise of the process for 100 nm structures. Using SEM images for calibrating the VT5 model allows reducing the standard error of 40 structures to 1.18 nm.

Magnetotransport in submicron devices formed on the basis of GaAs/AlGaAs structures is simulated by the method of nonequilibrium Green functions. In the one-particle approximation, the influence of a perpendicular magnetic field on electron transmission through a quasi-one-dimensional quantum dot and the Aharonov—Bohm interferometer is considered. Two-terminal conductance and magnetic moment of the devices are calculated. Two-dimensional patterns of equilibrium (persistent) currents are obtained. The correlations between energy dependences of magnetic moment and conductance are considered. For the quasi-one-dimensional quantum dot, regular conductance oscillations similar to the ABOs were found at low magnetic fields (0.05—0.4 T). In the case of a ring interferometer, the contribution to the total equilibrium current and magnetic moment at a given energy can change sharply both in magnitude and in sign when the magnetic field changes within the same Aharonov—Bohm oscillation. The conductance through the interferometer is determined not by the number of propagating modes, but rather by the influence of triangular quantum dots at the entrances to the ring, causing back scattering. Period of calculated ABOs corresponds to that measured for these devices.

In the modern world, knowledge and high technologies determine the effectiveness of the economy, can radically improve the quality of life of people, modernize infrastructure and public administration, and ensure law and order and security. The creation of a research infrastructure based on a high-performance hybrid cluster enabled detailed calculations of complex phenomena and processes without full-scale experiments. It has become possible to most efficiently apply modern methods of multiscale computer modeling when developing prototypes of new materials with desired properties for their further synthesis. Such approaches can significantly reduce the cost and speed up the development of modern technologies for producing new semiconductor materials for nanoelectronics, composite materials for the aerospace industry and others. Thus, the use of multiscale modeling methods in combination with the use of high-performance software tools made it possible to create a computer model of a nanoscale heterostructure, develop tools for predictive computer modeling of the physical structure of nanoelectronic devices, the neuromorphic architecture of multilevel memory devices, defect formation in composite materials, and others.

The article discusses methods of consolidating scientific services of a digital platform for integrating a set of scientific services for various fields of science for conducting interdisciplinary research. Solutions for creating consolidated services can be widely used for multilevel, multiscale modeling in the field of materials science, which provides complex modeling at several levels of the hierarchy. Currently, this problem is being solved by creating multicomponent hierarchical software systems on corporate computing systems. With the advent of high-performance cloud computing platforms, it will be possible to order services for solving particular modeling problems as a scientific service. In this case, the tasks of complex hierarchical modeling will be solved by a consolidated service - a service providing sequential-parallel execution of complex modeling components in the form of specialized scientific services. The description of the processes for the provision of scientific services is based on the research methodology and is a research plan (the work process mapping), which describes a set of operations related to time and includes a list of necessary resources for their implementation. In modern conditions of the development of a microservice approach to the creation of computing systems and the evolution of the Service Oriented Architecture and of the Enterprise Service Bus integration, special attention is paid to the problems of efficient integration of platform services. The paper proposes to supplement the existing description of a scientific service with the possibility of ordering a third-party service based on agile integration. This approach will allow at the present stage of development of service architectures to overcome the shortcomings of centralized systems such as Enterprise Service Bus and take advantage of the elasticity of cloud computing and a microservice approach to creating information and computing systems.

Editorial article

Editorial article