Broadband terahertz wave absorber based on combination of composite material and metamaterial as well as design method and application of broadband terahertz wave absorber
By combining graphene/nanocellulose composites with graphene metamaterials and utilizing their strong dielectric loss and electromagnetic resonance loss mechanisms, the absorption band is expanded, solving the problem of balancing the thickness and frequency band of terahertz absorbers, achieving thin thickness and wide bandwidth absorption, reducing processing difficulty, and having good engineering application prospects.
Patent Information
- Application Number
- CN202510749658.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-06
- Publication Date
- 2025-09-26
AI Technical Summary
Existing terahertz absorbers are difficult to achieve both thin thickness and wide-band absorption, and are difficult to process, which restricts their practical application.
By combining graphene/nanocellulose composites with graphene metamaterials, and utilizing the strong dielectric loss of the graphene/nanocellulose composites and the electromagnetic resonance loss mechanism of the metamaterials, the absorption band is expanded through the synergistic effect of the local electromagnetic resonance of the graphene metamaterial layer and the dielectric loss of the nanocellulose dielectric layer. The geometric structure parameters are optimized through machine learning to achieve a balance between thin thickness and wide bandwidth.
It achieves wide-band terahertz wave absorption at a relatively thin thickness, reduces preparation complexity, improves design efficiency, and has good engineering application prospects.
Smart Images
Figure CN120709734A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of terahertz wave absorbing technology, and in particular to a broadband terahertz absorber based on the combination of composite materials and metamaterials, and a design method and application thereof. Background Art
[0002] Terahertz waves, with their strong penetrating properties, wide bandwidth, and high resolution, offer unique advantages in radar detection and are expected to become a crucial tool for situational awareness and target identification on future battlefields. To address terahertz detection, developing absorbers that are "thin, light, wide, and strong" has become a key development goal in terahertz absorption technology. Currently, terahertz absorption methods are primarily divided into porous materials and metamaterial absorbers. The former achieves broadband absorption through multiple reflections and attenuation of terahertz waves, but typically requires millimeter-scale thicknesses. The latter overcomes this thickness bottleneck through the electromagnetic resonance of metamaterials, but their absorption bandwidth is relatively narrow. Graphene, with its light weight and excellent optoelectronic properties, offers significant advantages in metamaterial absorbers, but still faces the challenge of a narrow absorption bandwidth. Developing new terahertz absorbers that balance thinness and broadband absorption has become a key area of breakthrough in this technology.
[0003] To achieve both thinness and broadband absorption in absorbers, the main design approaches currently employed are to arrange multi-scale metamaterials within the metamaterial layer or to stack multiple metamaterials vertically. In-plane arrangements of multi-scale metamaterials can weaken absorption strength due to electromagnetic coupling between metamaterial units, limiting bandwidth expansion. Layered metamaterial absorbers achieve broadband absorption by stacking metamaterials with different resonant frequencies, leveraging the complementary frequency responses of each layer. However, stacking multiple metamaterial layers increases the thickness of the absorber to a certain extent. Existing technologies have designed a terahertz absorber based on a metal-dielectric-interdigitated graphene structure. Dynamically tuned absorption is achieved by adjusting the graphene's Fermi level and relaxation time, but this dynamic tuning mechanism struggles to maintain broadband absorption. Most currently designed broadband terahertz metamaterial absorbers struggle to achieve both thinness and broadband absorption. Furthermore, the resolution of currently designed terahertz graphene metamaterials is generally above 20 μm, making fabrication difficult. This has limited their development to simulations, severely hindering their transition to practical applications. Therefore, it is urgent to design a new terahertz absorber structure that has good fabrication feasibility while taking into account both thin thickness and wide-band absorption. Summary of the Invention
[0004] In response to the shortcomings of the aforementioned background technology, the present invention primarily overcomes the difficulties currently faced in designing terahertz absorbers, which have limited ability to achieve both thinness and broadband performance, as well as the difficulty in fabricating them. The present invention provides a broadband terahertz absorber based on a composite material combined with a metamaterial, as well as its design method and application. This broadband terahertz absorber combines a graphene / nanocellulose composite material with a graphene metamaterial, leveraging the synergistic effects of the strong dielectric loss generated by the large amount of heterogeneous interface polarization in the graphene / nanocellulose composite material and the electromagnetic resonance loss mechanism of the metamaterial to achieve strong absorption of broadband terahertz waves with a relatively thin thickness.
[0005] The first object of the present invention is to provide a broadband terahertz absorber based on a combination of a composite material and a metamaterial, the broadband terahertz absorber comprising a metal layer, and a graphene / nanocellulose composite material layer, a nanocellulose dielectric layer, and a graphene metamaterial layer sequentially stacked on the metal layer; The graphene / nanocellulose composite material layer is prepared by uniformly dispersing graphene powder and nanocellulose powder in a water solvent and then performing casting. The nanocellulose dielectric layer is prepared by casting a nanocellulose dispersion; The graphene metamaterial layer includes a plurality of graphene metamaterial units periodically arranged on the nanocellulose dielectric layer; Each graphene metamaterial unit consists of a square ring with two mutually intersecting and perpendicular "I" structures nested inside it.
[0006] Preferably, each "I"-shaped structure includes a vertical rod and horizontal rods arranged at both ends of the vertical rod, and the two horizontal rods are arranged relatively parallel; Two mutually intersecting and perpendicular "I"-shaped structures include two "I"-shaped structures in which vertical rods are mutually intersecting and perpendicularly arranged; The two parallel horizontal bars in each "I" structure are of equal length; In each graphene metamaterial unit, a gap is left between the ends of the two cross bars in one "I" structure and the ends of the two cross bars in another "I" structure.
[0007] Preferably, the graphene metamaterial layer is prepared by printing graphene ink according to a preset pattern onto a nanocellulose dielectric layer using a droplet jetting method; wherein the graphene ink is prepared by uniformly dispersing graphene powder in a nanocellulose dispersion; and the preset pattern is formed by a plurality of periodically arranged graphene metamaterial units.
[0008] Preferably, the line width of the square ring, and the line width of the vertical rod and the horizontal rod are all 100-120 μm.
[0009] Preferably, the thickness of the metal layer is 10-30 μm; The graphene / nanocellulose composite material layer has a thickness of 10 to 40 μm; The thickness of the nanocellulose dielectric layer is 10~40 μm.
[0010] A second object of the present invention is to provide a design method for a broadband terahertz absorber based on a combination of composite materials and metamaterials, comprising: According to the geometric characteristics of the absorber, multiple sets of data sets with different geometric parameter combinations are obtained; Based on multiple sets of data sets with different geometric parameter combinations, an absorber model is constructed using electromagnetic simulation software, and an absorption bandwidth data set is obtained through simulation analysis; Using data sets with multiple different geometric parameter combinations as network input and the absorber's absorption bandwidth as network output, the machine learning model based on the fully connected neural network is trained until convergence, obtaining a trained machine learning model. The original absorption bandwidth dataset is sorted in descending order, the last 2% of the data groups are removed, new geometric parameters are randomly generated and added to the dataset, and the supplemented dataset is passed through the trained machine learning model to obtain a new bandwidth dataset; the new bandwidth dataset is sorted in descending order, and the last 2% of the data are removed; after repeating the cycle for at least 100 times, the top 10% of the geometric parameters in the final iteration result are selected; Based on the top 10% of geometric parameters, electromagnetic simulation software was used for modeling and analysis, and the maximum product of the absorber's absorption bandwidth and the average absorption rate within the bandwidth was calculated and compared to obtain a geometric parameter combination that has both wide bandwidth and high absorption rate.
[0011] Preferably, during the training of the machine learning model, the input data set and the output data set are randomized and divided into a training set, a validation set, and a test set in a ratio of 70%, 15%, and 15%; Use the training set to train the machine learning model; Use validation sets and test sets to evaluate the generalization ability and accuracy of the trained machine learning model.
[0012] Preferably, when training the machine learning model, the gradient descent method, the conjugate gradient method or the Newton method is used, and the training is stopped when the output error meets the set requirements.
[0013] Preferably, the electromagnetic simulation software is CST Microwave Studio simulation software.
[0014] The third object of the present invention is to provide a broadband terahertz absorber based on the combination of composite materials and metamaterials for use in terahertz wave absorption.
[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention provides a broadband terahertz absorber based on the combination of composite materials and metamaterials, as well as its design method and application. The absorber provided by the present invention combines a graphene / nanocellulose composite material and a graphene metamaterial, and utilizes the synergistic effect of the strong dielectric loss generated by the polarization of a large number of heterogeneous interfaces of the graphene / nanocellulose composite material and the electromagnetic resonance loss mechanism of the metamaterial, so as to achieve strong absorption of broadband terahertz waves with a relatively thin thickness. Specifically, the local resonance of the metamaterial forms a strong electric field at a specific frequency point, significantly enhancing the polarization relaxation strength of the graphene-nanocellulose interface in the composite material, while the broadband dielectric loss base of the composite material effectively bridges the discrete resonance peaks, thereby effectively expanding the absorption bandwidth, solving the problems of narrow absorption band and large thickness of porous materials in traditional metamaterial absorbers. Compared with other complex and difficult-to-achieve terahertz absorber structures, its graphene metamaterial has lower resolution requirements for the preparation technology and exhibits excellent process compatibility, enabling it to be prepared using a low-cost and simple process, greatly reducing the complexity of absorber preparation.
[0016] This invention uses a machine learning-based computational method to optimize the geometric parameters of the absorber. A neural network model is used to establish a mapping between the absorber's geometric parameters and the absorption bandwidth. An iterative optimization cycle is then used to select the optimal absorber structure, reducing the need for complex manual parameter adjustment calculations and significantly improving design efficiency. This invention provides a thin, broadband, and easily fabricated structure for terahertz absorbers, along with an efficient optimization design method, promising promising engineering applications. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 This is a schematic diagram of the layered structure of a broadband terahertz absorber based on a combination of composite materials and metamaterials. From bottom to top, it consists of a metal plate, a graphene / nanocellulose composite layer, a nanocellulose dielectric layer, and a graphene metamaterial layer.
[0018] Figure 2 This is a flow chart of a design method for a broadband terahertz absorber based on the combination of composite materials and metamaterials.
[0019] Figure 3 Schematic diagram of the device for uniform droplet jet printing of graphene metamaterials.
[0020] In the figure, 1-computer, 2-CCD camera, 3-pulse signal generator, 4-piezoelectric nozzle, 5-graphene droplet, 6-infusion tube, 7-liquid reservoir, 8-LED light, 9-3D motion platform, 10-printing platform, 11-3D motion platform controller.
[0021] Figure 4This is the terahertz absorption spectrum obtained by CST simulation of the designed terahertz absorber. The absorber can achieve 90% absorption in the 0.5 ~ 3 THz frequency band. DETAILED DESCRIPTION
[0022] In order to enable those skilled in the art to better understand and implement the technical solution of the present invention, the present invention is further described below with reference to specific embodiments and drawings, but the embodiments are not intended to limit the present invention.
[0023] The purpose of the present invention is to provide a broadband terahertz absorber based on the combination of composite materials and metamaterials, as well as a design method and application thereof, to overcome the problems that currently designed terahertz absorbers are difficult to achieve both thin thickness and broadband and are difficult to process.
[0024] In order to achieve the above-mentioned object, the first aspect of the present invention provides a broadband terahertz absorber based on the combination of composite materials and metamaterials. Figure 1 As shown, the broadband terahertz absorber includes a metal layer, and a graphene / nanocellulose composite material layer, a nanocellulose dielectric layer, and a graphene metamaterial layer sequentially stacked on the metal layer; the graphene metamaterial layer serves as the top layer, and excites local electromagnetic resonance through artificial microstructures, forming a high-intensity electric field at a specific frequency point to enhance selective loss. The nanocellulose dielectric layer is a carrier of the metamaterial, used to separate the metamaterial layer and the composite material layer to prevent the electromagnetic loss effects of the metamaterial and the composite material from being affected by each other. The graphene / nanocellulose composite material layer relies on the strong polarization relaxation effect of the graphene-nanocellulose heterojunction to provide a broadband dielectric loss substrate, bridging discrete resonance peaks to expand the absorption bandwidth. The metal layer serves as the bottom layer, and through total reflection, it forces the residual wave to return and penetrate the middle layer, driving the electromagnetic wave to lose its energy through multiple reflections.
[0025] The graphene / nanocellulose composite material layer is prepared by uniformly dispersing graphene powder and nanocellulose powder in a water solvent and then performing casting. The nanocellulose dielectric layer is prepared by casting a nanocellulose dispersion; The graphene metamaterial layer includes a plurality of graphene metamaterial units periodically arranged on the nanocellulose dielectric layer; Each graphene metamaterial unit consists of a square ring with two mutually intersecting and perpendicular "I" structures nested inside it.
[0026] Each "I"-shaped structure includes a vertical rod and horizontal rods arranged at both ends of the vertical rod, and the two horizontal rods are arranged relatively parallel; Two mutually intersecting and perpendicular "I"-shaped structures include two "I"-shaped structures in which vertical rods are mutually intersecting and perpendicularly arranged; The two parallel horizontal bars in each "I" structure are of equal length; In each graphene metamaterial unit, a gap is left between the ends of the two cross bars in one "I" structure and the ends of the two cross bars in another "I" structure.
[0027] The graphene metamaterial layer is produced by printing graphene ink onto a nanocellulose dielectric layer according to a preset pattern using a droplet jetting method; wherein the graphene ink is prepared by uniformly dispersing graphene powder in a nanocellulose dispersion; and the preset pattern is formed by a plurality of periodically arranged graphene metamaterial units.
[0028] The line width of the square ring, as well as the line width of the vertical rod and the horizontal rod are all 100 to 120 μm.
[0029] The thickness of the metal layer is 10-30 μm; The graphene / nanocellulose composite material layer has a thickness of 10 to 40 μm; The thickness of the nanocellulose dielectric layer is 10~40 μm.
[0030] For example, a broadband terahertz absorber based on a composite material and metamaterial combination comprises four layers: from bottom to top, a metal layer, a graphene / nanocellulose composite layer, a nanocellulose dielectric layer, and a graphene metamaterial layer. The metal layer is 0.02 mm thick and is made of a highly conductive metal such as copper, aluminum, or silver. The graphene / nanocellulose composite layer and the nanocellulose dielectric layer are both 10 to 40 μm thick. The graphene metamaterial layer is composed of a periodic arrangement of graphene micropatterns, with the metamaterial unit consisting of a square ring with nested perpendicular "I" shapes. The line width of the graphene metamaterial pattern is set to 100 to 120 μm.
[0031] A second aspect of the present invention provides a design method for a broadband terahertz absorber based on a combination of composite materials and metamaterials, comprising: According to the geometric characteristics of the absorber, multiple sets of data sets with different geometric parameter combinations are obtained; Based on multiple sets of data sets with different geometric parameter combinations, an absorber model is constructed using electromagnetic simulation software, and an absorption bandwidth data set is obtained through simulation analysis; Using data sets with multiple different geometric parameter combinations as network input and the absorber's absorption bandwidth as network output, the machine learning model based on the fully connected neural network is trained until convergence, obtaining a trained machine learning model. The original absorption bandwidth dataset is sorted in descending order, the last 2% of the data groups are removed, new geometric parameters are randomly generated and added to the dataset, and the supplemented dataset is passed through the trained machine learning model to obtain a new bandwidth dataset; the new bandwidth dataset is sorted in descending order, and the last 2% of the data are removed; after repeating the cycle for at least 100 times, the top 10% of the geometric parameters in the final iteration result are selected; Based on the top 10% of geometric parameters, electromagnetic simulation software was used for modeling and analysis, and the maximum product of the absorber's absorption bandwidth and the average absorption rate within the bandwidth was calculated and compared to obtain a geometric parameter combination that has both wide bandwidth and high absorption rate.
[0032] During the training of the machine learning model, the input and output data sets are randomized and divided into training, validation, and test sets at a ratio of 70%, 15%, and 15% respectively. Use the training set to train the machine learning model; Use validation sets and test sets to evaluate the generalization ability and accuracy of the trained machine learning model.
[0033] When training a machine learning model, use gradient descent, conjugate gradient method, or Newton method, and stop training when the output error meets the set requirements.
[0034] The electromagnetic simulation software is CST Microwave Studio simulation software.
[0035] For example, see Figure 2 As shown, a design method for a broadband terahertz absorber based on a combination of composite materials and metamaterials includes: Step 1: Extract the geometric characteristic values of the absorber (including the period of the metamaterial, the size of the pattern, the thickness of the nanocellulose dielectric layer and the graphene / nanocellulose composite layer, etc.), assign values to different geometric characteristics using the equal-interval sampling method, and generate a data set of 1,000 sets of different geometric parameter combinations.
[0036] Step 2: Build absorber models in the electromagnetic simulation software CST Microwave Studio using different combinations of geometric parameter values. Set periodic and open boundary conditions to simulate infinite periodic structures. Obtain the corresponding absorption spectrum through simulation analysis. Record the frequency bandwidth where the absorption exceeds 90%.
[0037] Step 3: Build a machine learning model based on a fully connected neural network to model the complex nonlinear mapping relationship between the absorber's geometric parameters and absorption bandwidth. The geometric parameter dataset and the absorption bandwidth dataset are normalized and used as the input and output datasets of the machine learning model, respectively. The neural network model consists of three layers: an input layer, a hidden layer, and an output layer. The input layer has six neurons, representing the six geometric features of the absorber; the hidden layer has 12 neurons; and the output layer has one neuron, representing the absorption bandwidth of the absorber.
[0038] Step 4: Randomize the input and output datasets and divide them into input, validation, and test sets with a ratio of 70%, 15%, and 15%. Use the training set to train the neural network using algorithms such as gradient descent, conjugate gradient, and Newton's method. Training is terminated when the output error meets the specified requirements. Use the validation and test sets to evaluate the generalization ability and accuracy of the machine learning model.
[0039] Step 5: Sort the original absorption bandwidth dataset in descending order, removing the bottom 2% of data sets. New geometric parameters are randomly generated and added to the dataset. After training the updated parameter set using the machine learning model, the newly generated bandwidth dataset is again sorted in descending order, removing the bottom 2% of data. This optimization cycle is repeated 100 times, and the top 10% of parameters from the final iteration are selected as the optimization output.
[0040] Step 6: When determining the optimal geometric parameters, potential fabrication errors were taken into account to ensure a minimum line spacing of 30 μm in the metamaterial structure. CST software was used again for modeling and analysis. By calculating and comparing the maximum product of the metamaterial absorber's absorption bandwidth and the average absorptivity within that bandwidth, a geometric parameter combination that achieved both broadband and high absorptivity was selected.
[0041] In the present invention, a method for preparing a broadband terahertz absorber based on a combination of composite materials and metamaterials according to the geometric parameters obtained by the above-mentioned design method includes: Step 1: Graphene powder and nanocellulose powder are mixed in a mass ratio of approximately 0.3 to 0.8 and added to deionized water. A graphene / nanocellulose dispersion is prepared using simultaneous mechanical stirring and waterbath sonication at a stirring speed of 1600 rpm, an ultrasonic power of 80 W, and a treatment time of approximately 1.5 to 2 hours. A support substrate is selected from materials such as plexiglass, PET, or a glass slide. After ultrasonic cleaning and drying with a balloon, the graphene / nanocellulose dispersion is deposited onto the support substrate using a casting method and dried at room temperature for approximately 48 hours to form a graphene / nanocellulose composite film.
[0042] Step 2: Add nanocellulose powder to deionized water to prepare a nanocellulose dispersion with a mass fraction of approximately 1%. The nanocellulose dispersion is mechanically stirred for 1 to 1.5 hours at 1600 rpm and vacuum defoamed for 1 to 3 minutes to obtain a uniform, bubble-free nanocellulose dispersion. After ultrasonically cleaning the support substrate and drying it with a balloon, the nanocellulose dispersion is deposited onto the support substrate using a casting method and dried at room temperature for approximately 48 hours to form a nanocellulose film.
[0043] Step 3: Graphene sheets and nanocellulose powder were weighed in a mass ratio of approximately 6 to 7. The nanocellulose powder was first added to deionized water and mechanically stirred at 1600 rpm for approximately 10 minutes to prepare a nanocellulose dispersion. The graphene powder was then added to the nanocellulose dispersion and probe ultrasound was used to evenly disperse the graphene sheets at a power of 120 to 180 W for 1.5 to 3 hours to obtain a stably dispersed graphene ink suitable for droplet jetting.
[0044] Step 4, see Figure 3 As shown, the device for uniformly printing graphene metamaterials by microdroplet jetting comprises: Computer 1, CCD camera 2, pulse signal generator 3, piezoelectric nozzle 4, graphene droplet 5, infusion tube 6, liquid reservoir 7, LED light 8, 3D motion platform 9, printing platform 10, 3D motion platform controller 11. Computer 1 controls the pulse signal generator 3 to drive the piezoelectric nozzle 4 to spray graphene droplet 5, and controls the 3D motion platform controller 11 to precisely position the 3D motion platform 9 and printing platform 10. Ink in the liquid reservoir 7 is supplied to the nozzle 4 via the infusion tube 6. Simultaneously, the CCD camera 2 monitors the spraying process under the illumination of the LED light 8 and feeds the image back to the computer 1.
[0045] Ultrasonic cleaning of the piezoelectric nozzle 4, infusion tube 6, and liquid reservoir 7 was performed at a power of 60 to 80 W for 20 to 30 minutes. After the ultrasonic cleaning was completed, the piezoelectric nozzle 4, infusion tube 6, and liquid reservoir 7 were connected. The pulse width and frequency of the pulse signal output by the pulse signal generator 3 were adjusted. When the pulse signal parameters were adjusted to a pulse width of 15 to 20 µs, a frequency of 10 to 15 Hz, and a printing rate of 0.6 to 1 mm / s, the graphene droplets 5 were stably ejected and printed in a uniform, straight line.
[0046] Step 5: Write and execute a printing program on a computer according to the designed graphene metamaterial pattern. By synergistically controlling the movement of the three-dimensional motion platform and the graphene micro-droplet ejection of the piezoelectric nozzle, the graphene micro-droplets are deposited point by point on the surface of the nanocellulose film to form the expected pattern. Loop this printing program to repeat it multiple times for multi-layer graphene printing to obtain a graphene metamaterial layer.
[0047] Step 6: Select a highly conductive metal such as copper, aluminum or silver as the bottom plate of the absorber. After ultrasonic cleaning the metal plate and drying it with a balloon blower, bond the metal plate, the graphene / nanocellulose composite film, and the nanocellulose film printed with graphene metamaterials with a trace amount of adhesive to obtain a formed terahertz absorber.
[0048] The third aspect of the present invention provides an application of a broadband terahertz absorber based on the combination of composite materials and metamaterials in terahertz wave absorption.
[0049] It should be noted that the experimental methods used in the present invention are all conventional methods unless otherwise specified; the reagents and materials used can be purchased on the market unless otherwise specified.
[0050] Example 1 See Figure 1 As shown, a composite-structured broadband graphene terahertz absorber based on the micro-droplet ejection technology includes four layers in total, which are a metal layer, a graphene / nanocellulose composite material layer, a nanocellulose dielectric layer, and a graphene metamaterial layer from bottom to top. The graphene metamaterial layer is composed of periodically arranged graphene micro-patterns, and the metamaterial unit consists of a square ring and an "I" shape nested inside it and perpendicular to each other. The metamaterial units are distributed in an M×N pattern, where both M and N are positive integers. The metal layer is a highly conductive metal such as copper, aluminum, or silver, with a thickness of 0.02 mm. The thickness of the graphene / nanocellulose composite material layer is 10 - 40 μm, the thickness of the nanocellulose dielectric layer is 10 - 40 μm, and the line width of the graphene metamaterial pattern is set to 100 - 120 μm.
[0051] Example 2 Based on the structure of Example 1, a design method for a broadband terahertz absorber based on the combination of composite materials and metamaterials is provided, and the specific steps are as follows: Step 1: Extract the key geometric parameter features of the designed absorber, including the period of the metamaterial unit p , the side length of the square ring pattern l , the vertical bar length of the "I" shape pattern d 1 and the horizontal bar length d 2, and assign different values to these parameters in the way of equally spaced sampling. The specific values are as follows: pϵ[1400, 2400], the value interval is 200 μm; lϵ[1000, 1800], the value interval is 200 μm; d 1ϵ[600, 1400], with a value interval of 200 μm; d 2ϵ[200, 1000], with a value interval of 200 μm; a data set containing 1000 sets of different geometric parameter value combinations is generated by random combination.
[0052] Step 2: Using the electromagnetic simulation software CST Microwave Studio, construct absorber models based on different geometric parameter combinations. Periodic boundary conditions are set on the x- and y-axes, and an open boundary condition is set on the z-axis to simulate an infinite periodic structure. The corresponding absorption spectrum is obtained through frequency domain analysis. The frequency bandwidth where the absorption exceeds 90% is recorded.
[0053] Step 3: Apply Min-Max normalization to the combined absorber geometric parameter dataset and the bandwidth dataset obtained from electromagnetic simulation. The normalized geometric parameter dataset serves as the input dataset for the machine learning model, and the normalized bandwidth dataset serves as the output dataset. A machine learning model based on a fully connected neural network is established. The input layer consists of six neurons, covering the six geometric eigenvalues of the absorber, and the hidden layer consists of 12 neurons, using the ReLU activation function. The output layer consists of one neuron, representing the absorber's absorption bandwidth, using a linear activation function.
[0054] Step 4: Randomly shuffle the existing dataset and divide it into an input set, a validation set, and a test set in a ratio of 70%, 15%, and 15%. Train the neural network using gradient descent, conjugate gradient, and Newton's methods. The neural network uses the mean squared error (MSE) loss function, and the initial training round is set to 500. After each training round, calculate the mean squared error (MSE) loss on the validation set until the mean squared error stops decreasing for 20 consecutive rounds. Training is then terminated. The performance of the machine learning model is evaluated using both the validation and test sets.
[0055] Step 5: Sort the original absorption bandwidth dataset in descending order, and remove the absorber geometric parameters corresponding to the bottom 2% of the narrowest bandwidth datasets. Then, randomly generate 20 new sets of geometric parameters to add to the dataset. Use the trained machine learning model to calculate the absorption bandwidth for these updated geometric parameter sets, and perform the same sorting and removal operations on the newly generated bandwidth datasets. This optimization cycle is repeated 100 times, and the top 10% of parameters from the final iteration are selected as the optimized output.
[0056] Step 6. When determining the optimal geometric parameters, consider potential fabrication errors and ensure that the line spacing of the metamaterial pattern is at least 30 μm. Select an absorber structure with a line spacing greater than 30 μm and re-use the CST software modeling analysis to calculate the product of the absorption bandwidth of the remaining absorbers and the average absorptivity within the bandwidth. Select the absorber with the largest product, which is the optimal absorber structure with both broadband absorption and high absorptivity.
[0057] The optimal geometric parameters are finally obtained: p =1800 μm, l =1200 μm, d 1=800 μm, d 2=400 μm.
[0058] Example 3 Based on the geometric parameters designed in Example 2, a method for preparing a broadband terahertz absorber based on a combination of a composite material and a metamaterial is provided, comprising: Step 1: Graphene powder and nanocellulose powder were mixed in a mass ratio of approximately 1:2 in deionized water. The mixture was then treated with simultaneous mechanical stirring (1600 rpm) and water bath ultrasound (80 W) for 1.5 hours to prepare a dispersion. A plexiglass substrate was used as the support. After ultrasonic cleaning and drying, the dispersion was deposited onto the substrate using a tape-casting process and dried at room temperature for 48 hours to form a graphene / nanocellulose composite film.
[0059] Step 2: Add nanocellulose powder to deionized water to prepare a dispersion with a mass fraction of approximately 1%. The dispersion was mechanically stirred at 1600 rpm for 1.5 hours and vacuum-defoamed for 3 minutes to obtain a uniform, bubble-free dispersion. The dispersion was then deposited onto a clean, dry support substrate by tape casting and cured at room temperature for 48 hours to produce a nanocellulose film.
[0060] Step 3: Weigh graphene sheets and nanocellulose powder in a mass ratio of approximately 7:1. First, add the nanocellulose powder to deionized water and mechanically stir at 1600 rpm for 10 minutes to form a dispersion. Then, add the graphene powder and ultrasonicate with a titanium alloy probe (120 W, 3 h) to obtain a stable graphene ink suitable for microdroplet jetting.
[0061] Step 4: Clean the piezoelectric nozzle 4, infusion tube 6, and liquid reservoir 7 with 80 W ultrasonic cleaning for 30 minutes and then connect and assemble them. Adjust the pulse signal generator 3 to a pulse width of 20 μs and a frequency of 12 Hz, and set the printing rate to 0.8 mm / s to achieve stable ejection and linear printing of graphene droplets.
[0062] Step 5: Based on the preset pattern of the target metamaterial, a computer control program is written. Through the coordinated control of the 3D motion platform and the piezoelectric nozzle, graphene droplets are deposited point by point on the surface of the nanocellulose film. The multi-layer printing process is repeated to build a patterned graphene metamaterial layer.
[0063] Step 6: Use copper foil as the base plate of the absorber. After ultrasonically cleaning the metal plate and blowing it dry with a balloon, glue the copper foil, graphene / nanocellulose composite film, and nanocellulose film printed with graphene metamaterials with a small amount of epoxy resin to form a terahertz absorber.
[0064] Among them, the thickness of the copper foil is 20 μm; the thickness of the graphene / nanocellulose composite film is about 20 μm; the thickness of the nanocellulose film is about 40 μm; and the thickness of the graphene metamaterial is about 1 μm.
[0065] In order to illustrate the performance of the broadband terahertz absorber based on the combination of composite materials and metamaterials provided by the present invention, it is described with reference to the accompanying drawings.
[0066] Figure 4 This is the terahertz absorption spectrum obtained by testing the terahertz absorber prepared in Example 3. The absorber can achieve 90% absorption in the 0.5 to 3 THz frequency band.
[0067] The present invention describes preferred embodiments and their effects. However, those skilled in the art, once informed of the basic inventive concept, may make additional changes and modifications to these embodiments. Therefore, it is intended that the appended claims be interpreted to include the preferred embodiments as well as all changes and modifications that fall within the scope of the invention.
[0068] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.
Claims
1. A broadband terahertz absorber based on a combination of composite materials and metamaterials, characterized in that: The broadband terahertz absorber comprises a metal layer, and a graphene / nanocellulose composite material layer, a nanocellulose dielectric layer and a graphene metamaterial layer stacked in sequence on the metal layer; The graphene / nanocellulose composite material layer is prepared by uniformly dispersing graphene powder and nanocellulose powder in a water solvent and then performing casting. The nanocellulose dielectric layer is prepared by casting a nanocellulose dispersion; The graphene metamaterial layer includes a plurality of graphene metamaterial units periodically arranged on the nanocellulose dielectric layer; Each graphene metamaterial unit consists of a square ring with two mutually intersecting and perpendicular "I" structures nested inside it.
2. The broadband terahertz absorber based on the combination of composite materials and metamaterials according to claim 1, characterized in that: Each "I"-shaped structure includes a vertical rod and horizontal rods arranged at both ends of the vertical rod, and the two horizontal rods are arranged relatively parallel; Two mutually intersecting and perpendicular "I" structures include two "I" structures in which vertical rods are mutually intersecting and perpendicularly arranged; The two parallel horizontal bars in each "I" structure are of equal length; In each graphene metamaterial unit, a gap is left between the ends of the two cross bars in one "I" structure and the ends of the two cross bars in another "I" structure.
3. The broadband terahertz absorber based on the combination of composite materials and metamaterials according to claim 2, characterized in that: The graphene metamaterial layer is produced by printing graphene ink onto a nanocellulose dielectric layer according to a preset pattern using a droplet jetting method; wherein the graphene ink is prepared by uniformly dispersing graphene powder in a nanocellulose dispersion; and the preset pattern is formed by a plurality of periodically arranged graphene metamaterial units.
4. The broadband terahertz absorber based on the combination of composite materials and metamaterials according to claim 2, characterized in that: The line width of the square ring, as well as the line width of the vertical rod and the horizontal rod are all 100 to 120 μm.
5. The broadband terahertz absorber based on the combination of composite materials and metamaterials according to claim 1, characterized in that: The thickness of the metal layer is 10-30 μm; The graphene / nanocellulose composite material layer has a thickness of 10 to 40 μm; The thickness of the nanocellulose dielectric layer is 10~40 μm.
6. A method for designing a broadband terahertz absorber based on a combination of composite materials and metamaterials according to any one of claims 1 to 5, characterized in that: include: According to the geometric characteristics of the absorber, multiple sets of data sets with different geometric parameter combinations are obtained; Based on multiple sets of data sets with different geometric parameter combinations, an absorber model is constructed using electromagnetic simulation software, and an absorption bandwidth data set is obtained through simulation analysis; Using data sets with multiple different geometric parameter combinations as network input and the absorber's absorption bandwidth as network output, the machine learning model based on the fully connected neural network is trained until convergence, obtaining a trained machine learning model. The original absorption bandwidth dataset is sorted in descending order, the last 2% of the data groups are removed, new geometric parameters are randomly generated and added to the dataset, and the supplemented dataset is passed through the trained machine learning model to obtain a new bandwidth dataset; the new bandwidth dataset is sorted in descending order, and the last 2% of the data are removed; After the cycle is repeated at least 100 times, the top 10% of the geometric parameters in the final iteration result are selected; Based on the top 10% of geometric parameters, electromagnetic simulation software was used for modeling and analysis, and the maximum product of the absorber's absorption bandwidth and the average absorption rate within the bandwidth was calculated and compared to obtain a geometric parameter combination that has both wide bandwidth and high absorption rate.
7. The design method of a broadband terahertz absorber based on a combination of composite materials and metamaterials according to claim 6, characterized in that: During the training process of the machine learning model, the input and output data sets are randomized and divided into training set, validation set and test set in the ratio of 70%, 15% and 15% respectively; Use the training set to train the machine learning model; Use validation sets and test sets to evaluate the generalization ability and accuracy of the trained machine learning model.
8. The design method of a broadband terahertz absorber based on a combination of composite materials and metamaterials according to claim 7, characterized in that: When training a machine learning model, use gradient descent, conjugate gradient method, or Newton method, and stop training when the output error meets the set requirements.
9. The design method of a broadband terahertz absorber based on a combination of composite materials and metamaterials according to claim 6, characterized in that: The electromagnetic simulation software is CST Microwave Studio simulation software.
10. Application of a broadband terahertz absorber based on the combination of composite materials and metamaterials in terahertz wave absorption.