rus
Issue #5/2025
I. O. Dryagin, A. N. Manin, A. P. Gorshkova, N. L. Istomina
Choice of Materials for Creating Micro-OptoElectromechanical Switches for Next-Generation Communication Systems
Choice of Materials for Creating Micro-OptoElectromechanical Switches for Next-Generation Communication Systems
DOI: 10.22184/1993-7296.FRos.2025.19.5.400.406
The paper discusses the selection of materials for micro-opto-electromechanical switches (MOEMS) used in new generation communication systems. The focus is on the optical and electrical characteristics of materials, such as dielectric permittivity, refractive index, and bandgap. A study of the effect of these parameters on the efficiency of optical systems is carried out. Machine learning methods were used to predict the properties of materials, and promising materials with high dielectric permittivity were identified. The results obtained can be useful in the development of new methods for designing optical and radio-frequency communication components.
The paper discusses the selection of materials for micro-opto-electromechanical switches (MOEMS) used in new generation communication systems. The focus is on the optical and electrical characteristics of materials, such as dielectric permittivity, refractive index, and bandgap. A study of the effect of these parameters on the efficiency of optical systems is carried out. Machine learning methods were used to predict the properties of materials, and promising materials with high dielectric permittivity were identified. The results obtained can be useful in the development of new methods for designing optical and radio-frequency communication components.
Теги: диэлектрическая проницаемость искусственные материалы машинное обучение микро-опто- электромеханические переключатели (моэмс) оптические компоненты систем связи показатель преломления ширина запрещенной зоны
Choice of Materials for Creating Micro-Opto-Electromechanical Switches for Next-Generation Communication Systems
I. O. Dryagin, A. N. Manin, A. P. Gorshkova, N.L. Istomina
Moscow Aviation Institute (National Research University), Moscow,Russia
The paper discusses the selection of materials for micro-opto-electromechanical switches (MOEMS) used in new generation communication systems. The focus is on the optical and electrical characteristics of materials, such as dielectric permittivity, refractive index, and bandgap. A study of the effect of these parameters on the efficiency of optical systems is carried out. Machine learning methods were used to predict the properties of materials, and promising materials with high dielectric permittivity were identified. The results obtained can be useful in the development of new methods for designing optical and radio-frequency communication components.
Keywords: micro-opto-electromechanical switches (MOEMS), dielectric permittivity, refractive index, bandgap, machine learning, optical components of communication systems, artificial materials
Article received: 10.04.2025
Article accepted: 05.05.2025
Introduction
The development of modern communication systems imposes increased requirements on the materials used, in particular, on their optical and electrical properties [1]. Micro-opto-electromechanical switches (MOEMS) use materials that have high dielectric permittivity, low absorption loss and high stability in operation over a wide frequency range. The key parameters in the selection of materials are the refractive index, the bandgap and the ability to polarize [2].
Researches [3, 4] showed that polycrystalline materials with high dielectric permittivity have promising characteristics for use in MOEMS. One of the important advantages of crystalline materials is that by varying the width of the band gap, it is possible to create artificial materials with a specified effective dielectric permittivity. The permittivity of semiconductor materials depends on the sum of certain types of atomic vibrations in the lattice or their electron shells. A graphical analysis of the distribution of the effective dielectric permittivity (fig. 1) demonstrates that most materials have values in the range of 0 to 20, with a smooth decay for higher dielectric permittivity values. This indicates the availability of a wide selection of materials, among which it is necessary to highlight the most suitable for specific engineering tasks.
The dependence of the dielectric permittivity on the bandgap (fig. 2) shows that for materials with a narrow bandgap (less than 3 eV), significant fluctuations in permeability are possible, while for wide-bandgap materials (>4 eV) the permittivity is consistently low. This must be taken into account when designing optical systems, as this characteristic affects the absorption coefficient and reflection loss [5].
An important parameter is the refractive index, which affects the permittivity (fig. 3). As a consequence, materials with high permeability also have a higher refractive index, which can be used to create efficient waveguide structures and optical switches [6].
To select and evaluate the possibilities of using materials as MOEMS, it is necessary to try out many combinations of electro-optical characteristics. The use of machine learning (ML) methods allows this process to be automated.
Study of Methods for Applying Machine Learning to Identify Materials for the Design of Devices Operating in the Optical Frequency Range
The use of machine learning (ML) techniques automates the process of finding and evaluating materials suitable for use in MOEMS [7, 8]. In the course of the study, a gradient boosting model was built, which predicts the values of dielectric permittivity based on data on the bandgap, refractive index, and electronic structure of materials.
The evaluation of the model quality showed the following results: the root mean square error (MSE) was 41.12, the mean absolute error (MAE) was 4.12, and the coefficient of determination (R2) reached 0.62. This indicates that the model is sufficiently predictive and can be used in practical problems of selecting materials (fig. 4).
Analysis of the significance of the features revealed that the key parameters affecting the dielectric permittivity are the refractive index and the bandgap. The correlation map (fig. 5) confirmed a high degree of correlation between these parameters, which makes it possible to effectively predict the optical properties of materials.
The predicted values of the effective dielectric permittivity showed that most artificial materials with high permeability are observed in the region of materials with a low bandgap. This is especially important for the design of optical devices, where high signal transmission efficiency and minimization of losses are required.
In addition, a correlation map was constructed (fig. 6), illustrating the relationship between the main parameters of the materials. It shows a significant influence of the band gap width and dielectric permittivity on the refractive index.
An analysis of the distribution of permittivity values predicted for the optical range (fig. 7) showed that the maximum distribution density is observed in the region of materials with a narrow bandgap and high dielectric permittivity. This demonstrates the potential advantages of such materials in the design of devices.
Well-known databases [9] were used to search for promising materials with the required electrophysical parameters. The developed model predicted complex composite semiconductor materials with high dielectric permittivity and narrow bandgap, making them promising for applications in communication systems. The best materials predicted by the model are presented in the Table.
Summary
The identified materials have a combination of low bandgap width and high dielectric permittivity, which makes them promising for use in high-frequency components of optical and radio frequency devices. These properties can significantly reduce signal transmission losses, improve reflectivity, and increase the efficiency of waveguide structures and switches.
The use of machine learning methods has shown its effectiveness in predicting the properties of materials, which opens up new opportunities for automated search for optimal materials with specified characteristics. In the future, we plan to expand the range of materials we study and improve our forecasting model, which will allow us to find new and effective solutions for next-generation communication systems.
Authors
Dryagin Ivan O., Senior Lecturer, Department of Innovation Management (317), FSBEI HE Moscow Aviation Institute (National Research University), Moscow, Russia.
ORCID: 0000-0002-4545-2305
Manin Andrey N., Technician, Scientific Research Department 317, FSBEI HE Moscow Aviation Institute (National Research University), Moscow, Russia
Gorshkova Alina P., Technician, Scientific Research Department 317, FSBEI HE Moscow Aviation Institute (National Research University), Moscow, Russia.
Istomina Natalia L., Dr. Sci. (Phys.&Math.), prof., Department 317, FSBEI HE Moscow Aviation Institute (National Research University), Moscow, Russia.
ORCID: 0000-0001-6008-1226
Contribution of the authors
Dryagin I.O.: formation of the idea, development of the concept;
Manin A.N.: preparation of data for processing, formation of illustrative content and interpretation of data; Gorshkova A.P.: research of the predictive ability of the model; Istomina N.L.: experimental and theoretical verification of the reliability of the obtained results.
Conflict of interest
The authors carried out the work in collaboration and in accordance with the specified contribution. The authors agreed with the edits made to the overall manuscript text
I. O. Dryagin, A. N. Manin, A. P. Gorshkova, N.L. Istomina
Moscow Aviation Institute (National Research University), Moscow,Russia
The paper discusses the selection of materials for micro-opto-electromechanical switches (MOEMS) used in new generation communication systems. The focus is on the optical and electrical characteristics of materials, such as dielectric permittivity, refractive index, and bandgap. A study of the effect of these parameters on the efficiency of optical systems is carried out. Machine learning methods were used to predict the properties of materials, and promising materials with high dielectric permittivity were identified. The results obtained can be useful in the development of new methods for designing optical and radio-frequency communication components.
Keywords: micro-opto-electromechanical switches (MOEMS), dielectric permittivity, refractive index, bandgap, machine learning, optical components of communication systems, artificial materials
Article received: 10.04.2025
Article accepted: 05.05.2025
Introduction
The development of modern communication systems imposes increased requirements on the materials used, in particular, on their optical and electrical properties [1]. Micro-opto-electromechanical switches (MOEMS) use materials that have high dielectric permittivity, low absorption loss and high stability in operation over a wide frequency range. The key parameters in the selection of materials are the refractive index, the bandgap and the ability to polarize [2].
Researches [3, 4] showed that polycrystalline materials with high dielectric permittivity have promising characteristics for use in MOEMS. One of the important advantages of crystalline materials is that by varying the width of the band gap, it is possible to create artificial materials with a specified effective dielectric permittivity. The permittivity of semiconductor materials depends on the sum of certain types of atomic vibrations in the lattice or their electron shells. A graphical analysis of the distribution of the effective dielectric permittivity (fig. 1) demonstrates that most materials have values in the range of 0 to 20, with a smooth decay for higher dielectric permittivity values. This indicates the availability of a wide selection of materials, among which it is necessary to highlight the most suitable for specific engineering tasks.
The dependence of the dielectric permittivity on the bandgap (fig. 2) shows that for materials with a narrow bandgap (less than 3 eV), significant fluctuations in permeability are possible, while for wide-bandgap materials (>4 eV) the permittivity is consistently low. This must be taken into account when designing optical systems, as this characteristic affects the absorption coefficient and reflection loss [5].
An important parameter is the refractive index, which affects the permittivity (fig. 3). As a consequence, materials with high permeability also have a higher refractive index, which can be used to create efficient waveguide structures and optical switches [6].
To select and evaluate the possibilities of using materials as MOEMS, it is necessary to try out many combinations of electro-optical characteristics. The use of machine learning (ML) methods allows this process to be automated.
Study of Methods for Applying Machine Learning to Identify Materials for the Design of Devices Operating in the Optical Frequency Range
The use of machine learning (ML) techniques automates the process of finding and evaluating materials suitable for use in MOEMS [7, 8]. In the course of the study, a gradient boosting model was built, which predicts the values of dielectric permittivity based on data on the bandgap, refractive index, and electronic structure of materials.
The evaluation of the model quality showed the following results: the root mean square error (MSE) was 41.12, the mean absolute error (MAE) was 4.12, and the coefficient of determination (R2) reached 0.62. This indicates that the model is sufficiently predictive and can be used in practical problems of selecting materials (fig. 4).
Analysis of the significance of the features revealed that the key parameters affecting the dielectric permittivity are the refractive index and the bandgap. The correlation map (fig. 5) confirmed a high degree of correlation between these parameters, which makes it possible to effectively predict the optical properties of materials.
The predicted values of the effective dielectric permittivity showed that most artificial materials with high permeability are observed in the region of materials with a low bandgap. This is especially important for the design of optical devices, where high signal transmission efficiency and minimization of losses are required.
In addition, a correlation map was constructed (fig. 6), illustrating the relationship between the main parameters of the materials. It shows a significant influence of the band gap width and dielectric permittivity on the refractive index.
An analysis of the distribution of permittivity values predicted for the optical range (fig. 7) showed that the maximum distribution density is observed in the region of materials with a narrow bandgap and high dielectric permittivity. This demonstrates the potential advantages of such materials in the design of devices.
Well-known databases [9] were used to search for promising materials with the required electrophysical parameters. The developed model predicted complex composite semiconductor materials with high dielectric permittivity and narrow bandgap, making them promising for applications in communication systems. The best materials predicted by the model are presented in the Table.
Summary
The identified materials have a combination of low bandgap width and high dielectric permittivity, which makes them promising for use in high-frequency components of optical and radio frequency devices. These properties can significantly reduce signal transmission losses, improve reflectivity, and increase the efficiency of waveguide structures and switches.
The use of machine learning methods has shown its effectiveness in predicting the properties of materials, which opens up new opportunities for automated search for optimal materials with specified characteristics. In the future, we plan to expand the range of materials we study and improve our forecasting model, which will allow us to find new and effective solutions for next-generation communication systems.
Authors
Dryagin Ivan O., Senior Lecturer, Department of Innovation Management (317), FSBEI HE Moscow Aviation Institute (National Research University), Moscow, Russia.
ORCID: 0000-0002-4545-2305
Manin Andrey N., Technician, Scientific Research Department 317, FSBEI HE Moscow Aviation Institute (National Research University), Moscow, Russia
Gorshkova Alina P., Technician, Scientific Research Department 317, FSBEI HE Moscow Aviation Institute (National Research University), Moscow, Russia.
Istomina Natalia L., Dr. Sci. (Phys.&Math.), prof., Department 317, FSBEI HE Moscow Aviation Institute (National Research University), Moscow, Russia.
ORCID: 0000-0001-6008-1226
Contribution of the authors
Dryagin I.O.: formation of the idea, development of the concept;
Manin A.N.: preparation of data for processing, formation of illustrative content and interpretation of data; Gorshkova A.P.: research of the predictive ability of the model; Istomina N.L.: experimental and theoretical verification of the reliability of the obtained results.
Conflict of interest
The authors carried out the work in collaboration and in accordance with the specified contribution. The authors agreed with the edits made to the overall manuscript text
Readers feedback
