Preparation method of high-permeability soft magnetic ferrite core material

Abstract

The invention relates to the technical field of soft magnetic ferrite materials, in particular to a preparation method of a high-permeability soft magnetic ferrite core material, which comprises the following steps: firstly, preparing FeSiCrBCu soft magnetic alloy powder with uniform components and controllable particle size through a single-roller quenching and hydrogen decrepitation process, and laying a structural foundation for excellent magnetic performance; and then a mechanical stirring-ultrasonic synergistic coating technology is utilized, so that the ferrite precursor uniformly coats the alloy powder, an effective core-shell structure is formed, and solid-phase reaction and crystal grain uniform growth during sintering are promoted. Besides, a multi-task deep learning model (MTL-DL) based on transfer learning is adopted to carry out multi-objective collaborative optimization on the sintering process, accurately match and dynamically adjust key parameters, and accurately regulate and control the material densification process, the grain size and the phase distribution on the microcosmic level, so that the problem that the microstructure and component uniformity are difficult to accurately control in the prior art is solved; therefore, the problems of poor material performance and high power loss are solved.

Description

Technical Field

[0001] This invention relates to the field of soft magnetic ferrite materials technology, and in particular to a method for preparing a high-permeability soft magnetic ferrite core material. Background Technology

[0002] With the rapid development of electronic information technology, various electronic devices are constantly moving towards miniaturization, high frequency, and high efficiency, which places more stringent performance requirements on soft magnetic ferrite core materials. As a key component in electronic devices that enables the conversion between electrical and magnetic energy, the performance of soft magnetic ferrite core materials directly affects the overall performance and efficiency of the electronic devices. High-permeability soft magnetic ferrite core materials, due to their ability to achieve large magnetic induction intensity at lower magnetic field strengths, effectively reduce core energy loss and improve the energy conversion efficiency of electronic devices, have shown great application potential in many fields such as switching power supplies, transformers, and inductors, becoming a current research hotspot in the field of magnetic materials.

[0003] However, existing methods for preparing high-permeability soft ferrite core materials make it difficult to precisely control the microstructure and compositional uniformity of the materials during the preparation process. This results in poor performance in terms of permeability and saturation magnetic induction, as well as relatively high power loss, which fails to meet the demands of modern electronic devices for high-performance soft ferrite core materials. Summary of the Invention

[0004] The purpose of this invention is to provide a method for preparing a high-permeability soft magnetic ferrite core material, aiming to solve the technical problem in the prior art where it is difficult to accurately control the microstructure and composition uniformity of the material during the preparation process, resulting in poor performance in terms of permeability and saturation magnetic induction intensity, as well as relatively high power loss, which cannot meet the technical requirements of modern electronic devices for high-performance soft magnetic ferrite core materials.

[0005] To achieve the above objectives, the present invention provides a method for preparing a high-permeability soft magnetic ferrite core material, comprising the following steps:

[0006] Amorphous alloy strips were produced using Fe, Si, Cr, B, and Cu as raw materials via a single-roll rapid cooling method, with the roll speed controlled at 25-35 m / s and the cooling rate ≥10. 5 K / s, after hydrogen crushing, FeSiCrBCu soft magnetic alloy powder with a particle size of 10-50μm was obtained;

[0007] Weigh the raw materials according to the molar ratio Fe2O3:Mn3O4:ZnO = 52-54:26-28:18-20, add deionized water and dispersant, and prepare a suspension with a solid content of 25-35%.

[0008] The obtained soft magnetic alloy powder was added to the prepared suspension and treated by mechanical stirring-ultrasound to make the precursor uniformly coated on the powder surface. After drying, the core-shell structured composite powder was obtained.

[0009] The core-shell structured composite powder was pressed into shape, sintered in an atmosphere sintering furnace, and cooled with the furnace to obtain a high-performance soft magnetic ferrite composite material.

[0010] When sintering in an atmosphere sintering furnace, the sintering scheme is formulated as follows:

[0011] A multi-task deep learning model (MTL-DL) based on transfer learning is used to optimize sintering process parameters. This model simultaneously predicts permeability, saturation magnetic induction intensity and power loss, and outputs Pareto front solution set.

[0012] When formulating the sintering scheme, the multi-task deep learning model (MTL-DL) used takes sintering temperature, holding time, heating rate, oxygen partial pressure and cooling rate as inputs. It extracts common features by sharing the bottom feature layer and optimizes the three objectives simultaneously by using the task-specific output layer. The model uses Bayesian optimization for hyperparameter tuning and uses transfer learning strategy to quickly adapt to the new equipment environment using small sample data.

[0013] The multi-task deep learning model (MTL-DL) is a hybrid architecture based on a Transformer encoder and a multilayer perceptron (MLP). The Transformer encoder is used to enhance features and extract correlations from the input five-dimensional process parameter sequence. Its output is concatenated with the standardized material composition feature vector and input into a shared bottom-layer MLP for nonlinear feature fusion. Finally, the magnetic permeability, saturation magnetic induction intensity, and power loss values ​​are predicted by regressing through three independent task-specific MLP output layers.

[0014] The model is trained based on a historical sintering process database, which contains 500 pairs of process parameters and corresponding performance indicators. A loss function with uncertainty weighting is used for multi-task training to balance the learning rate of different tasks. The transfer learning strategy is as follows: on the basis of the pre-trained model, the Transformer encoder weights are frozen, and only a portion of the sample data (50-100 sets) from the new equipment is used to fine-tune the shared MLP and the task-specific MLP to achieve rapid adaptation.

[0015] In the step of “using Fe, Si, Cr, B, and Cu as raw materials, amorphous alloy ribbons are prepared by rapid cooling, and FeSiCrBCu soft magnetic alloy powder with a particle size of 10-50 μm is obtained after hydrogen crushing treatment”, the mass percentage of each element is: Fe 76-78%, Si 8-10%, Cr 4-6%, B 4-5%, Cu 1-2%, and the median particle size (D50) of Cr element is ≤2 μm, in order to enhance interfacial diffusion and reduce coercivity.

[0016] In the preparation of the suspension, a composite additive is added to the suspension, which includes: 0.02-0.04% MoO3, 0.01-0.03% TiO2, 0.01-0.02% SnO2, 0.005-0.015% Bi2O3 and 0.003-0.008% Co2O3.

[0017] The composite additive also includes 0.001-0.003% Gd2O3 to improve high-temperature magnetic stability.

[0018] In the step of “adding the obtained soft magnetic alloy powder to the prepared suspension and treating it by mechanical stirring-ultrasound combination”, the mechanical stirring speed is 300-500 rpm, the ultrasonic power is 600-800W, the treatment time is 30-60 minutes, and the pulsed ultrasonic mode is used to prevent local overheating. The drying conditions are vacuum drying at 80-100℃ for 4-6 hours.

[0019] During sintering, the sintering process is carried out in a mixed atmosphere of nitrogen and oxygen. The sintered material needs to be vacuum annealed at 550-650℃ for 2-3 hours. After annealing, it is cooled to below 150℃ at a rate of no more than 0.5℃ / min before being taken out of the furnace.

[0020] This invention discloses a method for preparing high-permeability soft magnetic ferrite core materials. In practical application, this invention uses a single-roll quenching and hydrogen annealing process to prepare FeSiCrBCu soft magnetic alloy powder with uniform composition and controllable particle size, laying a structural foundation for obtaining excellent magnetic properties. Furthermore, through mechanical stirring-ultrasonic synergistic coating technology, the ferrite precursor is uniformly coated on the surface of the alloy powder, forming an effective core-shell structure, thereby promoting uniform solid-state reaction and grain growth during sintering. Moreover, this invention employs a multi-task deep learning model (MTL-DL) based on transfer learning to perform multi-objective synergistic optimization of the sintering process, enabling precise matching and dynamic adjustment of key parameters such as temperature, atmosphere, and time, thereby accurately controlling the densification process, grain size, and phase distribution of the material at the microscopic level. This method solves the technical problem in existing technologies where it is difficult to precisely control the microstructure and compositional uniformity of the material during preparation, resulting in poor permeability, saturation magnetic induction intensity performance, and relatively high power loss, failing to meet the technical requirements of modern electronic devices for high-performance soft magnetic ferrite core materials. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart of the preparation method of the high permeability soft magnetic ferrite core material of the present invention. Detailed Implementation

[0023] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, but should not be construed as limiting the present invention.

[0024] Please see Figure 1 , Figure 1 This is a flowchart of the preparation method of the high permeability soft magnetic ferrite core material of the present invention.

[0025] This invention provides a method for preparing a high-permeability soft magnetic ferrite core material.

[0026] Includes the following steps:

[0027] S1. Amorphous alloy thin strips are produced using Fe, Si, Cr, B, and Cu as raw materials via a single-roll rapid cooling method. The roll speed is controlled at 25-35 m / s, and the cooling rate is ≥10. 5K / s, after hydrogen crushing, FeSiCrBCu soft magnetic alloy powder with a particle size of 10-50μm was obtained;

[0028] In this specific embodiment, the mass percentages of each element are: Fe 76-78%, Si 8-10%, Cr 4-6%, B 4-5%, Cu 1-2%, and the median particle size (D50) of Cr is ≤2μm, in order to enhance interfacial diffusion and reduce coercivity.

[0029] S2. Weigh the raw materials according to the molar ratio Fe2O3:Mn3O4:ZnO=52-54:26-28:18-20, add deionized water and dispersant, and prepare a suspension with a solid content of 25-35%.

[0030] In this specific embodiment, a composite additive is added to the suspension during preparation. The composite additive includes: 0.02-0.04% MoO3, 0.01-0.03% TiO2, 0.01-0.02% SnO2, 0.005-0.015% Bi2O3, and 0.003-0.008% Co2O3.

[0031] The composite additive also includes 0.001-0.003% Gd2O3 to improve high-temperature magnetic stability.

[0032] S3. The obtained soft magnetic alloy powder is added to the prepared suspension and treated by mechanical stirring-ultrasound to make the precursor uniformly coated on the powder surface. After drying, the core-shell structured composite powder is obtained.

[0033] In this specific embodiment, the mechanical stirring speed is 300-500 rpm, the ultrasonic power is 600-800 W, the processing time is 30-60 minutes, and the pulsed ultrasonic mode (on / off time ratio = 5:1) is used to prevent local overheating. The drying conditions are vacuum drying at 80-100℃ for 4-6 hours.

[0034] S4. The core-shell composite powder is pressed into shape, sintered in an atmosphere sintering furnace, and cooled with the furnace to obtain a high-performance soft magnetic ferrite composite material.

[0035] The sintering process is as follows: heat the temperature from room temperature to 400-500℃ at a rate of 1-3℃ / min and hold for 1-2 hours to completely remove the binder and lubricant (bonding); then continue to heat the temperature to the final sintering temperature of 1100-1250℃ at a rate of 2-5℃ / min and hold for 2-4 hours.

[0036] For this specific embodiment, when sintering in an atmosphere sintering furnace, the sintering scheme is formulated as follows:

[0037] A multi-task deep learning model (MTL-DL) based on transfer learning is used to optimize sintering process parameters. This model simultaneously predicts permeability, saturation magnetic induction intensity and power loss, and outputs Pareto front solution set.

[0038] When formulating the sintering scheme, the multi-task deep learning model (MTL-DL) used takes sintering temperature, holding time, heating rate, oxygen partial pressure and cooling rate as inputs. It extracts common features by sharing the bottom feature layer and optimizes the three objectives simultaneously by using the task-specific output layer. The model uses Bayesian optimization for hyperparameter tuning and uses a transfer learning strategy to quickly adapt to the new equipment environment using small sample data.

[0039] The multi-task deep learning model (MTL-DL) is a hybrid architecture based on a Transformer encoder and a multilayer perceptron (MLP). The Transformer encoder is used to enhance features and extract correlations from the input five-dimensional process parameter sequence. Its output is concatenated with the standardized material composition feature vector and input into a shared bottom-layer MLP for nonlinear feature fusion. Finally, the magnetic permeability, saturation magnetic induction intensity, and power loss values ​​are predicted by regressing through three independent task-specific MLP output layers.

[0040] The model is trained based on a historical sintering process database, which contains 500 pairs of process parameters and corresponding performance indicators. A loss function with uncertainty weighting is used for multi-task training to balance the learning rate of different tasks. The transfer learning strategy is as follows: on the basis of the pre-trained model, the Transformer encoder weights are frozen, and only a portion of the sample data (50-100 sets) from the new equipment is used to fine-tune the shared MLP and the task-specific MLP to achieve rapid adaptation.

[0041] During sintering, the sintering process is carried out in a mixed atmosphere of nitrogen and oxygen. The sintered material needs to be vacuum annealed at 550-650℃ for 2-3 hours. After annealing, it is cooled to below 150℃ at a rate of no more than 0.5℃ / min before being taken out of the furnace.

[0042] Secondly, during sintering, the oxygen partial pressure is dynamically adjusted by a PID controller, based on the optimization suggestions output in real time by a multi-task deep learning model.

[0043] Furthermore, the Fe, Si, Cr, B, and Cu raw materials are all pure metals or master alloy ingots with a purity of ≥99.9%. Surface treatment is required before use to remove oxide scale and oil, followed by drying.

[0044] The Fe2O3, Mn3O4, and ZnO raw materials are all industrial-grade or electronic-grade powders with a purity of ≥99.5%. They must be dried at 100-120℃ for at least 2 hours before use to remove adsorbed moisture.

[0045] Furthermore, the specific process for preparing amorphous alloy strips using the single-roll rapid cooling method is as follows: the prepared alloy raw materials are heated to complete melting and homogeneous composition in a vacuum induction melting furnace. Then, the molten metal is sprayed onto the surface of a high-speed rotating water-cooled copper roller using pressurized argon gas, with the roller speed controlled at 25-35 m / s and the cooling rate ≥10. 5 K / s, rapid cooling forms a continuous amorphous alloy ribbon.

[0046] Finally, after sintering, the sintered magnetic core undergoes necessary machining (such as grinding and chamfering) to achieve the final dimensional requirements.

[0047] A coil is wound on the magnetic core, and key soft magnetic properties such as permeability (μ), saturation magnetic induction (Bs), coercivity (Hc), and core loss (Pcv) are measured using equipment such as an impedance analyzer and a BH analyzer.

[0048] The method for preparing a high-permeability soft magnetic ferrite core material according to this embodiment, in practical application, utilizes a single-roll quenching and hydrogen annealing process to prepare FeSiCrBCu soft magnetic alloy powder with uniform composition and controllable particle size, laying a structural foundation for obtaining excellent magnetic properties. Furthermore, through mechanical stirring-ultrasonic synergistic coating technology, the ferrite precursor is uniformly coated on the surface of the alloy powder, forming an effective core-shell structure, thereby promoting uniform solid-state reaction and grain growth during sintering. Moreover, this invention employs a multi-task deep learning model (MTL-DL) based on transfer learning to perform multi-objective synergistic optimization of the sintering process, enabling precise matching and dynamic adjustment of key parameters such as temperature, atmosphere, and time, thereby accurately controlling the densification process, grain size, and phase distribution of the material at the microscopic level. This approach solves the technical problem in existing technologies where it is difficult to precisely control the microstructure and compositional uniformity of the material during preparation, resulting in poor permeability, saturation magnetic induction intensity performance, and relatively high power loss, failing to meet the technical requirements of modern electronic devices for high-performance soft magnetic ferrite core materials.

[0049] The above description discloses only one preferred embodiment of the present invention, and should not be construed as limiting the scope of the present invention. Those skilled in the art will understand that all or part of the processes of the above embodiments can be implemented, and equivalent changes made in accordance with the claims of the present invention are still within the scope of the invention.

Claims

1. A method for preparing a high-permeability soft magnetic ferrite core material, characterized in that, Includes the following steps: Amorphous alloy strips were produced using Fe, Si, Cr, B, and Cu as raw materials via a single-roll rapid cooling method, with the roll speed controlled at 25-35 m / s and the cooling rate ≥10. 5 K / s, after hydrogen crushing, FeSiCrBCu soft magnetic alloy powder with a particle size of 10-50μm was obtained; Weigh the raw materials according to the molar ratio Fe2O3:Mn3O4:ZnO = 52-54:26-28:18-20, add deionized water and dispersant, and prepare a suspension with a solid content of 25-35%. The obtained soft magnetic alloy powder was added to the prepared suspension and treated by mechanical stirring-ultrasound to make the precursor uniformly coated on the powder surface. After drying, the core-shell structured composite powder was obtained. The core-shell structured composite powder was pressed into shape, sintered in an atmosphere sintering furnace, and cooled with the furnace to obtain a high-performance soft magnetic ferrite composite material.

2. The method for preparing the high permeability soft magnetic ferrite core material as described in claim 1, characterized in that, When sintering in an atmosphere sintering furnace, the sintering scheme is formulated as follows: A multi-task deep learning model (MTL-DL) based on transfer learning is used to optimize sintering process parameters. This model simultaneously predicts permeability, saturation magnetic induction intensity and power loss, and outputs Pareto front solution set.

3. The method for preparing the high permeability soft magnetic ferrite core material as described in claim 2, characterized in that, When formulating the sintering scheme, the multi-task deep learning model (MTL-DL) used takes sintering temperature, holding time, heating rate, oxygen partial pressure and cooling rate as inputs. It extracts common features by sharing the bottom feature layer and optimizes the three objectives simultaneously by using the task-specific output layer. The model uses Bayesian optimization for hyperparameter tuning and uses a transfer learning strategy to quickly adapt to the new equipment environment using small sample data.

4. The method for preparing the high permeability soft magnetic ferrite core material as described in claim 3, characterized in that, The multi-task deep learning model (MTL-DL) is a hybrid architecture based on a Transformer encoder and a multilayer perceptron (MLP). The Transformer encoder is used to enhance features and extract correlations from the input five-dimensional process parameter sequence. Its output is concatenated with the standardized material composition feature vector and input into a shared bottom-layer MLP for nonlinear feature fusion. Finally, the permeability, saturation magnetic flux density, and power loss values ​​are predicted by regressing through three independent task-specific MLP output layers. The model is trained based on a historical sintering process database, which contains 500 pairs of process parameters and corresponding performance indicators. A loss function with uncertainty weighting is used for multi-task training to balance the learning rate of different tasks. The transfer learning strategy is as follows: on the basis of the pre-trained model, the Transformer encoder weights are frozen, and only a portion of the sample data (50-100 sets) from the new equipment is used to fine-tune the shared MLP and the task-specific MLP to achieve rapid adaptation.

5. The method for preparing the high permeability soft magnetic ferrite core material as described in claim 4, characterized in that, In the step "using Fe, Si, Cr, B, and Cu as raw materials, an amorphous alloy ribbon is prepared by rapid quenching, and after hydrogen crushing, FeSiCrBCu soft magnetic alloy powder with a particle size of 10-50 μm is obtained", the mass percentage of each element is: Fe 76-78%, Si 8-10%, Cr 4-6%, B 4-5%, Cu 1-2%, and the median particle size (D50) of Cr element is ≤2 μm, in order to enhance interfacial diffusion and reduce coercivity.

6. The method for preparing the high permeability soft magnetic ferrite core material as described in claim 5, characterized in that, In preparing the suspension, a composite additive is also added to the suspension, which includes: 0.02-0.04% MoO3, 0.01-0.03% TiO2, 0.01-0.02% SnO2, 0.005-0.015% Bi2O3 and 0.003-0.008% Co2O3.

7. The method for preparing the high permeability soft magnetic ferrite core material as described in claim 6, characterized in that, The composite additive also includes 0.001-0.003% Gd2O3 to improve high-temperature magnetic stability.

8. The method for preparing the high permeability soft magnetic ferrite core material as described in claim 7, characterized in that, In the step "adding the obtained soft magnetic alloy powder to the prepared suspension and treating it by mechanical stirring-ultrasound", the mechanical stirring speed is 300-500 rpm, the ultrasonic power is 600-800W, the treatment time is 30-60 minutes, and the pulsed ultrasonic mode is used to prevent local overheating. The drying conditions are vacuum drying at 80-100℃ for 4-6 hours.

9. The method for preparing the high permeability soft magnetic ferrite core material as described in claim 8, characterized in that, During sintering, the sintering process is carried out in a mixed atmosphere of nitrogen and oxygen. After sintering, the material needs to be vacuum annealed at 550-650℃ for 2-3 hours. After annealing, it is cooled to below 150℃ at a rate of no more than 0.5℃ / min before being taken out of the furnace.

Images