Accepted
,
Abstract +
In recent years, capacitively coupled plasmas driven by ultra-low frequency source have garnered increasing attention, because they are beneficial for generating ions with high energy and small scattering angle, which aligns well with the current trend in high aspect ratio etching. Since the sheath becomes thicker when a ultra-low frequency source is applied, the secondary electron emission becomes significant. Indeed, these energetic secondary electrons could enhance the ionization process and even influence the discharge mode. In this work, a two-dimensional fluid model is employed to study the influence of secondary electrons on the dual frequency capacitively coupled plasmas under different ultra-low frequency voltages, secondary electron emission coefficients and inter-electrode gaps. The high frequency is fixed at 13.6 MHz, and the ultra-low frequency is fixed at 400 kHz. First, by using the ion energy dependent secondary electron emission coefficient, it is shown that the electron density first decreases and then increases with ultra-low frequency voltage. This is because on one hand, the higher ultra-low frequency voltage leads to thicker sheath, and therefore, the effective discharge volume is compressed. On the other hand, secondary electrons emitted from electrodes could obtain more energy, and thus enhance the ionization process. By comparing with the results obtained with fixed secondary electron emission coefficients, it is found that in the low voltage range, the evolution of the electron density is similar to that with fixed coefficient of 0.1. While, in the high voltage range, the growth of the electron density is even more pronounced than that with fixed coefficient of 0.2, indicating that the enhancement of the secondary electron effect by ultra-low frequency voltage is non-linear. Finally, the impact of discharge gap on the plasma properties has also been discussed. It is shown that with the increase of inter-electrode gap from 2 cm to 4 cm, the maximum ionization rate becomes lower, but the electron density rises significantly, and the plasma radial uniformity is improved. When inter-electrode gap is large, secondary electrons could collide with neutral species fully, and thus their influence on the electron density at high ultra-low frequency voltage is more pronounced. The results obtained in this paper are helpful to understand the influence of ultra-low frequency source on the secondary electron effect, and provide some guidance for the optimization of plasma processing.
, , Received Date: 2024-12-12
Abstract +
, , Received Date: 2024-11-24
Abstract +
Birefringence, as a fundamental parameter of optical crystals, plays a vital role in numerous optical applications such as phase modulation, light splitting, and polarization, thereby making them key materials in laser science and technology. The significant birefringence of vanadate polyhedra provides a new approach for developing birefringent materials. In this study, first-principles calculations are used to investigate the band structures, density of states (DOS), electron localization functions (ELFs), and birefringence behaviors of four alkali metal vanadate crystals AV3O8 (A = Li, Na, K, Rb). The computational results show that all AV3O8 crystals have indirect band gaps, whose values are 1.695, 1.898, 1.965, and 1.984 eV for LiV3O8, NaV3O8, KV3O8, and RbV3O8, respectively. The DOS analysis reveals that near the Fermi level, the conduction band minima (CBM) in these vanadates are predominantly occupied by the outermost orbitals of V atoms, while the valence band maxima (VBM) are primarily contributed by O-2p orbitals. The O-2p orbitals also exhibit strong localization near the Fermi level. Combined with highest occupied molecular orbital-lowest unoccupied molecular orbital (HOMO-LUMO) analysis and population analysis, the bonding interactions in all four crystals mainly arise from the hybridization between V-3p and O-2p orbitals, indicating strong covalent bonding in V—O bonds. Through the analysis of structure-property relationships, the large birefringence is primarily attributed to the pronounced structural anisotropy, high anisotropy index of responsive electron distribution, unique arrangement of anionic groups, and d-p orbital hybridization between V-3d and O-2p orbitals. The calculated birefringence values at a wavelength of 1064 nm for LiV3O8, NaV3O8, KV3O8, and RbV3O8 are 0.28, 0.30, 0.28, and 0.27, respectively.
Abstract +
Orbitronic devices have attracted considerable interest due to their unique advantage of independence from strong spin-orbit coupling. Light metal chromium (Cr), with high orbital Hall conductivity, exhibits significant potential for application in orbit-spintronic devices. In this study, we present experimental verification of the inverse orbital Hall effect (IOHE) in Cr thin films and systematically investigate the underlying physical mechanisms of orbital-to-charge current conversion. The Cr/Ni and Pt/Ni heterostructures were fabricated on Al2O3 substrates via magnetron sputtering. Terahertz time-domain spectroscopy was employed to measure the terahertz emission signal. The Cr/Ni heterostructures exhibits the same positive terahertz polarity as the ISHE-dominant Pt/Ni heterostructures, despite the Cr layer owing negative spin Hall angle, which confirms the IOHE of Cr/Ni heterostructure. In the Cr/Ni heterostructures, femtosecond laser excitation generates spin current in the ferromagnetic Ni layer, which is converted into orbital current via its spin-orbit coupling. This orbital current propagates into the Cr layer where it is transformed into charge current through the IOHE. Furthermore, increasing the Cr thickness (2-40 nm) weakens the terahertz emission of Cr/Ni heterostructures due to enhanced optical absorption of Cr layers reducing spin current generation in Ni layers. However, optimizing Ni thickness (3-10 nm) significantly enhances the terahertz emission by improving the spin-orbital conversion efficiency. This work provides experimental evidence for IOHE in Cr films and demonstrates the crucial role of ferromagnetic layer engineering in spin-to-orbit conversion efficiency, offering innovative perspectives for the design and performance optimization of orbitronic devices.
,
Abstract +
Large-area capacitively coupled discharges are widely used in plasma enhanced chemical vapor deposition (PECVD) processes for solar cell and display manufacturing. With the increase of the chamber size and driving frequency for the purpose of higher production efficiency, the non-uniformity of deposited film induced by standing wave effects becomes severe and deserves more attention and in-depth studies. Based on a fluid model coupled with transmission line model, the potential amplitude distribution on the powered electrode as well as the plasma characteristics in a capacitive plasma sustained in a silane/hydrogen discharge driven at 27.12 MHz, with 2 m square electrodes, are investigated. This work identifies three critical control parameters: pressure, silane content, and input power, with particular emphasis on radial wave attenuation caused by electron-neutral elastic collisions. The simulation results are validated by industrial experimental results, confirming the relationship between the distributions of potential amplitude on the powered electrode and the film thickness.
Two distinct mechanisms emerge from the analysis. Under low silane content with high power conditions, the surface wave radial attenuation is not significant and the surface wave wavelength variations dominate the potential amplitude distribution on the powered electrode. Conversely, in the case of high silane content and low power, significant radial attenuation of the surface wave leads to noticeable weakening of the standing wave effect due to higher electron-neutral collision frequency. Neglecting the radial attenuation of the surface wave would result in significant deviations in the potential amplitude distribution on the powered electrode, as shown in the following figure.
Strategies such as adjusting power input positions or using multiple power input are studied to improve uniformity, but the improvements are still limited. Although it requires strict parameter control and machining precision, the shaped electrode demonstrates remarkable uniformity improvement of the potential distribution. In future work, it is necessary to further analyze the impact of the standing wave effects on the radial distributions of electron, ions, and neutral radicals under complex conditions, such as different chamber structures, gas flows, and temperature distributions, as well as the impact on the quality of deposited films. This will enable a more comprehensive and accurate study of standing wave effects, providing support and guidance for solving real industrial problems.
Two distinct mechanisms emerge from the analysis. Under low silane content with high power conditions, the surface wave radial attenuation is not significant and the surface wave wavelength variations dominate the potential amplitude distribution on the powered electrode. Conversely, in the case of high silane content and low power, significant radial attenuation of the surface wave leads to noticeable weakening of the standing wave effect due to higher electron-neutral collision frequency. Neglecting the radial attenuation of the surface wave would result in significant deviations in the potential amplitude distribution on the powered electrode, as shown in the following figure.
Strategies such as adjusting power input positions or using multiple power input are studied to improve uniformity, but the improvements are still limited. Although it requires strict parameter control and machining precision, the shaped electrode demonstrates remarkable uniformity improvement of the potential distribution. In future work, it is necessary to further analyze the impact of the standing wave effects on the radial distributions of electron, ions, and neutral radicals under complex conditions, such as different chamber structures, gas flows, and temperature distributions, as well as the impact on the quality of deposited films. This will enable a more comprehensive and accurate study of standing wave effects, providing support and guidance for solving real industrial problems.
,
Abstract +
Circular cross-section plasma is the most basic form of tokamak plasma and the fundamental configuration for magnetic confinement fusion experiments. This article is based on the HL-2A limiter discharge experiments to study the magnetohydrodynamic (MHD) equilibrium and MHD instability of tokamak plasmas in circular cross-section. The results showed that when q0=0.95, the internal kink mode of m/n=1/1 is always unstable. The increase in plasma β (=the ratio of thermal pressure to magnetic pressure) can lead to the appearance of external kink modes. The combination of axial safety factor q0 and edge safety factor qa determines the equilibrium configuration of the plasma and also affects the MHD stability of the equilibrium, but its growth rate is also related to the size of β. Under the condition of qa>2 and q0 slightly greater than 1, it is easy to achieve the stabilization of internal and surface kink modes. However the plasma becomes unstable again and the instability intensity increases as q0 continues to increase when q0 exceeds 1. As the poloidal specific pressure (βp) increases, the MHD instability develops, the equilibrium configuration of MHD elongates laterally, and the Shafranov displacement increases, which in turn has the effect for suppressing instability. Calculations have shown that the maximum β value imposed by the ideal MHD mode in a plasma with free boundary in tokamak experiments is proportional to the normalized current IN (=Ip(MA)/a (m)B0(T)), and the maximum specific pressure is calibrated as β(max)~2.01IN. The operational β limit of HL-2A circular cross-section plasma is approximately βNc≈2.0. A too high q0 is not conducive to MHD stability and leads to a decrease in the β limit. When q0=1.3, we obtain a maximum β_N of approximately 1.8. Finally, based on the existing circular cross-section plasma, some key factors affecting the operational β and the relationship between achievable high β and the calculated ideal β limits were discussed.
,
Abstract +
Spin-orbit torque (SOT) based on the spin-orbit coupling (SOC) effect has gained increasing attention in magnetic information storage, logical operation and neuron simulation devices because it can effectively manipulate magnetization switching, chiral magnetic domain wall, and magnetic skyrmion motions. Further improvement of the SOT efficiency and reduction of the driving current density are crucial scientific problems to be solved for high-density and low-power applications of SOT-based spintronic devices. The heavy rare-earth metal dysprosium (Dy) possesses a relatively strong SOC due to the partially filled f orbital electrons (4f10), which is expected to generate spin Hall torques. In this paper, the impact of Dy thicknesses on the SOT efficiency and SOT-driven magnetic reversal was explored in the Dy/Pt/[Co/Pt]3 magnetic multilayers, where the rare-earth Dy and [Co/Pt]3 were used as the spin-source layer and the perpendicularly magnetized ferromagnetic layer, respectively. A series of Dy/Pt/[Co/Pt]3 heterostructures with various Dy layer thicknesses (tDy) of 1, 3, 5 and 7 nm were fabricated by ultrahigh-vacuum magnetron sputtering. The perpendicular magnetic anisotropy, SOT efficiency, spin Hall angle and current-induced magnetization switching were characterized using the magnetic property and electrical transport measurements. The results showed that the switching field and magnetic anisotropic field decreased with an increase in tDy, revealing that the magnetic parameters can be regulated by the bottom Dy layer due to their structural sensitivity. However, both damping-like SOT efficiency and effective spin Hall angle (θSHeff) gradually increased with increasing tDy, indicating that the rare-earth Dy can provide additional spin current to enhance the SOT efficiency apart from the contribution of Pt/[Co/Pt]3. Particularly, the maximum θSHeff of 0.379±0.008 was achieved when tDy was 7 nm. According to the fitting analysis of the drift-diffusion model, the intrinsic spin Hall angle and spin diffusion length of the rare-earth Dy were extracted to be 0.260±0.039 and 2.234±0.383 nm, respectively, suggesting that Dy can be used as an ideal spin-source material. In addition, the critical switching current density (Jc) gradually decreased with the increase in tDy, and Jc reached a minimum value of approximately 5.3×106 A/cm2 at tDy=7 nm, which is primarily attributed to the increase of the damping-like SOT and slight decrease of the switching field. These results experimentally demonstrate a strong spin Hall effect of the rare-earth Dy, and provide a feasible route for designing SOT-based spintronic devices with low-power dissipation.
,
Abstract +
Packed bed dielectric barrier discharge (PB-DBD) is extremely popular in plasma catalysis applications, which can significantly improve the selectivity and energy efficiency of the catalytic processes. In order to achieve some complex chemical reactions, it is necessary to mix different materials in practical applications. In this work, based on the two-dimensional particle-in-cell/Monte Carlo collision (PIC/MCC) method, the discharge evolution in PB-DBD packed with two mixed dielectrics is numerically simulated to reveal the discharge characteristics. Due to the polarization of dielectric columns, the enhancement of electric field induces streamers at the bottom of the dielectric columns with high electrical permittivity (εr). The streamers propagate downward in the voids between the dielectric columns with low εr, which finally transitions into volume discharges. Then, a new streamer forms near the upper dielectric plate and propagates downward along the void of the dielectric columns with high εr. Moreover, electron density in between the columns with high εr is lower than that in between the dielectric columns with low εr. In addition, the numbers of e, N2+, O2+ and O2- present different profiles versus time. All of e, N2+ and O2+ increase in number before 0.8 ns. After 0.8 ns, the number of electrons decreases with time, while the numbers of N2+ and O2+ keep almost constant. During the whole process, the number of O2- keeps increasing versus time. The reason for the different temporal profiles can be analyzed as follows. The sum of electrons deposited on the dielectric and those lost in attachment reaction is greater than the number of electrons generated by ionization reaction, resulting in the declining electron trend. Comparatively, the deposition of N2+ and O2+ on the dielectric almost balances with their generation, leading to the constant numbers of N2+ and O2+. In addition, the variation of averaged electron density (ne) and averaged electron temperature (Te) in the voids between the dielectric columns are also analyzed under different experimental parameters. Simulation results indicate that both of them decrease with the increase in pressure or the decrease in voltage amplitude. Moreover, they increase with enlarging dielectric column radius. In addition, ne increases and then decreases with the increase of N2 content in the working gas, while Te monotonically increases. The variations of ne and Te in the voids can be explained as follows. With increasing pressure, the increase of collision frequency and the decrease of average free path lead to less energy obtained per unit time by electrons from the electric field, resulting in the decreasing Te. Moreover, the first Townsend ionization coefficient decreases with a reduction in Te, resulting in less electrons produced per unit time. Hence, both ne and Te decrease with increasing pressure. Additionally, Te is mainly determined by electric field strength. Therefore, the rising voltage amplitude results in the increase of and Te. Based on the same reason with pressure, nealso increases with increasing voltage amplitude. Consequently, both ne and Te increase with increasing voltage amplitude. In addition, the surface area of dielectric columns increases with enlarging dielectric column radius. Therefore, more polarized charges are induced on the inner surface of the dielectric column, inducing a stronger electric field outside. Accordingly, the enlarging dielectric column radius results in the increase of ne and Te. Moreover, the variation of ne with N2 content is analyzed from the ionization rate, and that of Te is obtained from analyzing the ionization thresholds of N2 and O2.
, , Received Date: 2024-08-09
Abstract +
To reveal the load mechanism of wall damage induced by bubble collapse, the near-wall cavitation bubble collapse evolution is numerically simulated using an improved multi-relaxation-time lattice Boltzmann method (MRT-LBM), and the dynamic behavior of near-wall cavitation bubble is systematically analyzed. First, the improved multi-relaxation pseudopotential model with a modified force scheme is introduced and validated through the Laplace law and thermodynamic consistency. Subsequently, the near-wall bubble collapse evolution is simulated using the improved model, and the process of the bubble collapse evolution is obtained. The accuracy of the numerical simulation results is confirmed by comparing with previous experimental results. Based on the obtained flow field information, including velocity and pressure distributions, the dynamic behaviors during the bubble collapse are thoroughly analyzed. The results show that the micro-jets released during the near-wall bubble collapse primarily originate from the first collapse, while the shock waves are generated during both the first and the second collapse. Notably, the intensity of the shock waves produced during the second collapse is significantly higher than that during the first collapse. Furthermore, the distribution characteristics of pressure and velocity on the wall during the near-wall bubble collapse are analyzed, revealing the load mechanism of wall damage caused by bubble collapse. The results show that the wall is subjected to the combined effects of shock waves and micro-jets: shock waves cause large-area surface damage due to their extensive propagation range, whereas micro-jets lead to concentrated point damage with their localized high-velocity impact. In summary, this study elucidates the evolution of near-wall bubble collapse and the load mechanism of wall damage induced by bubble collapse, which provides theoretical support for further utilizing the cavitation effects and mitigation of cavitation-induced damage.
, , Received Date: 2024-11-30
Abstract +
In this paper, a method of designing broadband reconfigurable polarization-converting metasurface operating in L-band is proposed. This method can also be used to directly modulate the information by using two modulation modes: binary amplitude shift keying (BASK) and binary phase shift keying (BPSK). Switching the ' ON/OFF state of PIN diode can be used to modify the amplitude and phase responses of the cross-polarized reflection of the element in a frequency band of 1.17–1.66 GHz, thereby creating a 1-bit digital coding meta-atom. By changing the real-time coding patterns of amplitude and phase, the reconfigurable metasurface can control beams and information modulation. Simulation results show that by changing the coding patterns of the metasurface, twin-beams and four-beams with different reflection angles can be obtained which fully validates the control ability of dynamic far-field beam. As an experimental verification, a reconfigurable metasurface consisting of 10×10 meta-atoms is fabricated, and its beam control and information modulation functions are tested. The far-field patterns of the metasurface with different coding phase distributions are measured. Furthermore, modulation signals of varying high/low voltage levels and rates are loaded onto the metasurface, in order to control its modulation mode and rate. The modulated signals reflected from metasurface are captured by a high-speed radio-frequency (RF) oscilloscope at varying rates and reflection angles, and then demodulated so as to recover the original information. On this basis, a metasurface wireless communication system based on BASK and BPSK is constructed to transmit digital image information in a real-world environment. In the experiment, the image is first represented by a sequence of '0' and '1' bits, corresponding to the operational state sequence of the metasurface used for transmitting information. The field programmable gate array (FPGA) is then used to generate signals with high and low voltage levels in real time according to the sequence of working states of the metasurface, and to modulate the carrier signal irradiated onto the metasurface. Therefore, the signal is converted into a modulated signal and received by the antenna. Finally, the signal is demodulated by the universal software radio peripheral (USRP) and transmitted to the terminal equipment, yielding the constellation diagrams and enabling the recovering of the images. The image information recovered under both modulation schemes verifies that the system can achieve real-time modulation and transmission of digital information. The proposed metasurface and the design method may be used in many fields, such as satellite communications and digital broadcasting.
, , Received Date: 2024-10-26
Abstract +
The manta ray is a large marine species, which has the ability of gliding efficiently and flapping rapidly. It can autonomously switch between various motion modes, such as gliding, flapping, and group swimming, based on ocean currents and seabed conditions. To address the computational resource and time constraints of traditional numerical simulation methods in modeling the manta ray’s three-dimensional (3D) large-deformation flow field, this study proposes a novel generative artificial intelligence approach based on a denoising probabilistic diffusion model (surf-DDPM). This method predicts the surface flow field of the manta ray by inputting a set of motion parameter variables. Initially, we establish a numerical simulation method for the manta ray’s flapping mode by using the immersed boundary method and the spherical function gas kinetic scheme (IB-SGKS), generating an unsteady flow dataset comprising 180 sets under frequency conditions of 0.3–0.9 Hz and amplitude conditions of 0.1–0.6 body lengths. Data augmentation is then performed. Subsequently, a Markov chain for the noise diffusion process and a neural network model for the denoising generation process are constructed. A pretrained neural network embeds the motion parameters and diffusion time step labels into the flow field data, which are then fed into a U-Net for model training. Notably, a transformer network is incorporated into the U-Net architecture to enable the handling of long-sequence data. Finally, we examine the influence of neural network hyperparameters on model performance and visualize the predicted pressure and velocity fields for multi-flapping postures that were not included in the training set, followed by a quantitative analysis of prediction accuracy, uncertainty, and efficiency. The results demonstrate that the proposed model achieves fast and accurate predictions of the manta ray’s surface flow field, characterized by extensive high-dimensional upsampling. The minimum PSNR value and SSIM value of the predictions are 35.931 dB and 0.9524, respectively, with all data falling within the 95% prediction interval. Compared with CFD simulations, the single-condition simulations by using AI model show that the prediction efficiency is enhanced by 99.97%.
,
Abstract +
Droplet microfluidics technology presents significant potential for applications in chemical analysis, biological detection, and material preparation. Passive droplet generation methods can rapidly achieve droplet formation by relying on the geometric characteristics of microchannels and shear flow. As a typical structure, the influence of fluid parameters and symmetry differences in cross microchannels on the droplet generation process has not been fully studied. Therefore, this paper uses the lattice Boltzmann method to conduct numerical simulation studies on droplet generation in symmetric and asymmetric cross microchannels, systematically analyzing the action mechanisms of capillary number, viscosity ratio, and microchannel symmetry. First, this study verifies the computational reliability of the numerical model through two classic cases, i.e., the droplet deformation under planar shear flow and stationary droplets on ideal solid surfaces. Then, this work focuses on the three flow stages in symmetric cross microchannels, i.e., the interface immersion stage, the shear-induced breakup stage, and the droplet migration and coalescence stage, analyzing the collaborative mechanism of capillary number and viscosity ratio. In the symmetric cross microchannel structure, the capillary number is the main factor determining the droplet size in the cross microchannel. With the increase of the capillary number, the surface tension gradually weakens, causing the liquid bridge at the droplet neck to break more easily and generate droplets. In contrast, the effect of the viscosity ratio on the droplet size is relatively small, but it can suppress the generation of sub-droplets and improve the uniformity of droplets by adjusting the viscous resistance of the continuous phase. On this basis, the study further quantifies the impact of microchannel symmetry on the droplet generation process in cross microchannels. In the asymmetric cross microchannel structure, the microchannel deviation breaks the flow symmetry and weakens the cooperative shearing effect of the oil-phase fluid on the immersion structure of the water-phase fluid. When the microchannel deviates within the centerline range of the water-phase microchannel, the droplet size increases significantly with the increase of the microchannel deviation. This is mainly because the oil-phase fluid on the side far from the deviation first squeezes the immersion structure of the water-phase fluid, and then the oil-phase fluid near the deviation side performs secondary squeezing on the immersion structure, resulting in the continuous elongation of the neck liquid bridge of the immersion structure and the offset of the shear position along the microchannel deviation direction. When the microchannel deviation exceeds the centerline position of the water-phase microchannel, the interface fracture of the water-phase immersion structure mainly relies on the double squeeze effect of the oil-phase fluid and the surface tension of water-phase fluid, and the droplet size tends to be stable. The relevant research results provide a theoretical basis for microchannel design and fluid parameter regulation in droplet microfluidics and further promote the application and development of droplet microfluidic technology.
,
Abstract +
Controlling of tungsten (W) impurity core accumulation is of great significance for the steady-state operation of tokamaks. This work mainly investigates the effect of neoclassical transport on the core accumulation of W impurities using STRAHL code. The study focuses on the HL-3 device, which will use tungsten divertor and conduct research under argon gas injection discharge conditions. In the simulation, the edge and core background plasma parameters are obtained by SOLPS-ITER and OMFIT simulations, respectively. The distribution of tungsten impurities in the boundary region is simulated using the IMPEDGE code. The edge anomalous transport coefficient in STRAHL is adjusted accordingly, and the simulation results are compared with those from IMPEDGE to ensure consistency in impurity distribution between the core and edge. In the core region, a numerical scan is performed to adjust the simulation results so that the energy radiation matches the setting values, thereby determining the specific turbulence convection velocity. By setting the coefficients for both the core and boundary regions, a complete distribution of W impurities from boundary to the core is obtained. To account the neoclassical transport effects, the neoclassical transport coefficients are calculated using the subroutine NEOART and applied to the impurity transport simulation, and the simulation region is set from ρ= 0.0 to 0.9. On this basis, the transport of W impurities with and without neoclassical convection is simulated. The simulation results show that without neoclassical convection, anomalous transport dominates the impurity transport, which is directed inward and enhances impurity accumulation in the core, and the core impurity density reaches 1.1×1016 m-3. After introducing neoclassical convection whose direction is outward, it can offset the inward anomalous convection and significantly reduces the W impurity density in the core, significantly reducing the core tungsten impurity density to 4.0×1015 m-3. In additional, the neoclassical convection in the region of ρ = 0.72 - 0.90 plays a more important role in reducing the core impurity density. Further analysis of the components of neoclassical convection shows that the PS (Pfirsche-Schlüter) component dominates the neoclassical convection term, which is mainly driven by the ion temperature gradient term. Therefore, experimentally, plasma heating can be used to enhance the temperature gradient and suppress impurity core accumulation.
,
Abstract +
The research aims to predict the thermodynamic stability of rare-earth compounds using machine learning (ML) models, providing crucial data support for advanced materials design and facilitating the discovery of new rare-earth compounds.
In terms of methods, this study is based on a dataset consisting of 280,569 compounds. The formation energies of these compounds were obtained through density functional theory (DFT) calculations. A system of 145 feature descriptors was constructed, covering stoichiometric properties, statistical properties of elements, electronic structure properties, and properties of ionic compounds, to comprehensively describe the characteristics of rare-earth compounds. Two ML models, random forest (RF) and neural network (NN), were selected to perform classification and regression tasks respectively. The 5-fold cross-validation was used to improve the reliability of the models. The min-max scaling technique was applied for data preprocessing, and an ensemble learning architecture was constructed to address the limitations of single model.
In the classification task, the RF and NN algorithms performed remarkably well. With 5-fold cross-validation, the accuracy reached approximately 0.97, and the F1 score was around 0.98, enabling the precise classification of compounds into stable or unstable categories. In the regression task, the mean absolute errors (MAE) of the formation energy predictions by the RF and NN models were 0.055 eV/atom and 0.071 eV/atom, respectively. This indicates that the model predictions are highly accurate and can, to a certain extent, replace complete DFT calculations. In the prediction analysis of systems outside the test set, six representative components were selected from the Materials Project database, covering binary, ternary, and quaternary systems. The prediction errors of all compositions were controlled within 0.5 eV/atom, and the error percentages were lower than 25%, demonstrating the strong extrapolation prediction ability of the models. When predicting the binary phase diagrams of rare-earth compounds La-Al and Ce-H using the trained models, the convex hull phase diagrams constructed through the ensemble learning architecture, which combines the prediction results of the RF and NN models, were highly consistent with those constructed from the Open Quantum Materials Database. The models successfully captured several metastable phases that were not present in multiple databases. Moreover, the convex hull distances of the predicted phases were mostly less than 0.1 eV/atom, with the maximum not exceeding 0.2 eV/atom.
In conclusion, this study successfully used ML models to predict the thermodynamic stability of rare-earth compounds. The constructed models demonstrated strong capabilities in classification and regression tasks. The ensemble learning architecture effectively improved the model performance, providing a promising tool for materials discovery in the field of rare-earth science and contributing to the research and development of new rare-earth compounds and the design of advanced materials.
,
Abstract +
Rare-earth elements possess unique atomic structures characterized by multiple unpaired 4f orbital electrons in inner shells, high atomic magnetic moments, and strong spin-orbit coupling. These attributes endow them with rich electronic energy levels, enabling the formation of compounds with diverse valence states and coordination environments. Consequently, rare-earth materials often exhibit exceptional magnetic properties and complex magnetic domain structures, making them critical for high-tech industrial development. The intricate magnetic configurations, diverse types of magnetic coupling, and direct/indirect magnetic exchange interactions in these materials not only facilitate the development of novel functional devices but also pose significant challenges to fundamental research. With the rapid advancement of data mining techniques, the emergence of big data and artificial intelligence has provided researchers with a new approach to efficiently analyze vast experimental and computational datasets, thereby accelerating the exploration and development of rare-earth magnetic materials. This paper focuses on rare-earth permanent magnetic materials, rare-earth magnetocaloric materials, and rare-earth magnetostrictive materials, detailing the application progress of data mining techniques in property prediction, composition and process optimization, and microstructural analysis. It also delves into current challenges and future trends, aiming to provide a theoretical foundation for deepening the integration of data mining technologies with rare-earth magnetic material research.
- 1
- 2
- 3
- 4
- 5
- ...
- 14
- 15
