A material preparation method based on purification and delamination optimization of montmorillonite

Abstract

The application provides a material preparation method based on purification and delamination synergistic optimization of montmorillonite, and comprises the following steps: S1, pretreating natural montmorillonite raw materials; S2, centrifugal purification by using a purification parameter linear optimization model; S3, preparing a montmorillonite delamination precursor; S4, ultrasonic delamination treatment by using a delamination effect dynamic regulation model; S5, purification and delamination synergistic optimization adjustment; S6, delamination montmorillonite stabilization treatment; S7, material forming and drying; S8, material sintering and performance characterization. Through the whole-process intelligent synergy of "raw material characteristics-purification parameters-delamination effect", the application breaks through the limitations of traditional processes and realizes efficient and stable preparation of montmorillonite materials.

Description

Technical Field

[0001] This invention relates to the field of inorganic non-metallic material preparation technology, and in particular to a material preparation method based on the synergistic optimization of montmorillonite purification and exfoliation. Background Technology

[0002] Montmorillonite, a natural layered silicate mineral, has significant application value in the field of materials due to its unique nanosheet structure, excellent mechanical properties, and chemical stability. However, natural montmorillonite raw materials often contain impurities such as quartz, calcite, and feldspar, and the strong interlayer forces directly restrict its performance. Therefore, it is necessary to purify and remove impurities and achieve layer dispersion through delamination.

[0003] In existing technologies, montmorillonite purification often employs centrifugation, but centrifugation parameters (speed and time) rely on empirical settings and are difficult to dynamically adjust based on raw material characteristics, leading to unstable purification purity. The delamination process often uses ultrasonic-assisted intercalation, but ultrasonic parameters lack correlation with purification effectiveness, easily resulting in uneven layer thickness and agglomeration. Furthermore, the purification and delamination processes are independent of each other, lacking a synergistic optimization mechanism, leading to poor consistency in final material properties and low production efficiency, thus limiting the application of montmorillonite in high-end materials fields. Summary of the Invention

[0004] This invention provides a material preparation method based on the synergistic optimization of montmorillonite purification and exfoliation. The core of this method lies in the dynamic control of parameters throughout the entire process through intelligent algorithms, specifically including:

[0005] Raw material pretreatment: The initial particle size and impurity content of montmorillonite were determined to provide basic data for subsequent parameter optimization;

[0006] Intelligent purification: It adopts a linear optimization model of purification parameters to predict the optimal centrifugation parameters based on the characteristics of raw materials, so as to achieve efficient removal of impurities;

[0007] Intercalation reaction: The ratio of intercalating agent and reaction conditions are adjusted according to the purification purity to provide precursors for exfoliation;

[0008] Dynamic peeling: A dynamic control model for peeling effect is adopted, which correlates purification parameters with intercalation effect and optimizes ultrasonic parameters to achieve precise dispersion of layers;

[0009] Collaborative optimization: A dual-model closed-loop feedback is constructed through a collaborative optimization index to dynamically adjust purification and stripping parameters, ensuring that intermediate product indicators meet the standards.

[0010] Stabilization and molding: Based on the delamination and lamellar characteristics, the addition of stabilizers and the molding process are controlled, and high-performance montmorillonite materials are finally obtained through sintering.

[0011] To achieve the above objectives, the present invention adopts the following technical solution:

[0012] A material preparation method based on the synergistic optimization of montmorillonite purification and exfoliation includes the following steps:

[0013] S1: Pre-treat the natural montmorillonite raw material, determine the average initial particle size distribution and initial impurity content, and obtain the pre-treated montmorillonite.

[0014] S2: Based on the initial average particle size distribution and initial impurity content, the optimized centrifugation speed and optimized centrifugation time were calculated using a linear optimization model of purification parameters; the pretreated montmorillonite was centrifuged and purified according to the optimized centrifugation speed and optimized centrifugation time to obtain a purified montmorillonite suspension and determine its purity.

[0015] S3: Prepare an intercalating agent solution according to the purity, and carry out an intercalation reaction between the intercalating agent solution and the purified montmorillonite suspension to obtain a mixture of exfoliation precursors and determine its intercalation rate.

[0016] S4: Based on purity, intercalation rate and optimized centrifugation speed, the optimized ultrasonic power and optimized ultrasonic time were calculated using a dynamic control model for the delamination effect. The delamination precursor mixture was ultrasonically delaminated according to the optimized ultrasonic power and optimized ultrasonic time to obtain the initial delaminated montmorillonite dispersion and to determine its initial laminated thickness and initial laminated particle size.

[0017] S5: Calculate the synergistic optimization index based on purity, intercalation rate, initial lamella thickness, and initial lamella particle size. If the synergistic optimization index is less than the preset threshold, update the weight parameters of the linear optimization model of purification parameters and the dynamic control model of exfoliation effect respectively, and return to perform centrifugation purification, intercalation reaction, or ultrasonic exfoliation again until the synergistic optimization index is not less than the preset threshold, and obtain the optimized exfoliated montmorillonite dispersion. At the same time, measure its optimized lamella thickness and optimized lamella particle size.

[0018] S6: Prepare a stabilizer solution according to the optimized lamella thickness, and carry out a stabilization reaction between the stabilizer solution and the optimized stripped montmorillonite dispersion to obtain a stabilized stripped montmorillonite dispersion.

[0019] S7: Based on the optimized lamellae particle size, set the casting parameters, cast and dry the stabilized exfoliated montmorillonite dispersion to obtain a material blank;

[0020] S8: Set sintering parameters according to the optimized layer thickness, sinter the material blank, and obtain a montmorillonite-based material.

[0021] In this manual, the training process of the linear optimization model for purification parameters in S2 includes: collecting 300 sets of training datasets containing the mean initial particle size distribution of raw materials, initial impurity content, corresponding optimal centrifugation speed, optimal centrifugation time, and purification purity; using the mean initial particle size distribution and initial impurity content as independent variables, and the optimal centrifugation speed and optimal centrifugation time as dependent variables; fitting the linear equation using the least squares method; and minimizing the mean square error between the predicted centrifugation parameters and the actual optimal centrifugation parameters using the gradient descent method until the loss value is less than 0.01, thus completing the model training.

[0022] In this manual, the training process of the dynamic control model for the exfoliation effect in S4 includes: collecting 300 training datasets containing purification purity, intercalation rate, optimized centrifugation speed, corresponding optimal ultrasonic power, optimal ultrasonic time, initial lamella thickness, and initial lamella particle size; using purification purity, intercalation rate, and optimized centrifugation speed as independent variables, and optimal ultrasonic power and optimal ultrasonic time as dependent variables; training the nonlinear regression equation using the Adam optimizer; and adjusting the parameters by minimizing the weighted error until the loss value is less than 0.05, thus completing the model training.

[0023] In this instruction manual, if the purity measured after centrifugation purification in S2 is less than 95%, the optimized centrifugation time is kept unchanged, the optimized centrifugation speed is increased by 10%, and the centrifugation purification operation is repeated. The purity of the purified montmorillonite suspension is measured again. The above adjustment and measurement steps are repeated until the purity is not less than 95%. Then, the qualified purity, the corresponding optimized centrifugation speed and optimized centrifugation time are transferred to S3 and S5.

[0024] In this instruction manual, if the initial lamella thickness measured after ultrasonic delamination in S4 is greater than 5 nm or the initial lamella particle size is greater than 200 nm, the optimized ultrasonic power is increased by 50 W and the optimized ultrasonic time is extended by 15 min before ultrasonic delamination is performed again. The initial lamella thickness and initial lamella particle size of the initially delaminated montmorillonite dispersion are measured again. The above adjustment and measurement steps are repeated until the initial lamella thickness is no greater than 5 nm and the initial lamella particle size is no greater than 200 nm. Then, the qualified initial lamella thickness and initial lamella particle size are transmitted to S5.

[0025] In this specification, the calculation method for the synergistic optimization index in S5 is as follows: the weights assigned to purity, intercalation rate, initial lamella thickness, and initial lamella particle size are 0.3, 0.2, 0.25, and 0.25, respectively; the synergistic optimization index is obtained by weighted summation based on the ratio of purity to target purity of 95%, the ratio of intercalation rate to target intercalation rate of 85%, the ratio of 1-lamella thickness to target thickness of 5 nm, and the ratio of 1-lamella particle size to target particle size of 200 nm.

[0026] In this manual, the process of updating the parameters of the purification parameter linear optimization model and the layer stripping effect dynamic control model in S5 includes: based on the difference between the co-optimization index and 1, combined with the gradient of the current model loss function, adjusting the linear equation weights of the purification parameter linear optimization model and the nonlinear equation weights of the layer stripping effect dynamic control model according to the preset learning rate of 0.01, so as to ensure that the prediction accuracy of the updated model is improved.

[0027] In this specification, the stabilizer solution in S6 is polyvinyl alcohol, and the mass fraction of the stabilizer solution is set according to the optimized sheet thickness: if the optimized sheet thickness is not greater than 3nm, the mass fraction is 0.3%; if the optimized sheet thickness is greater than 3nm but not greater than 5nm, the mass fraction = 0.3% + 0.05% × (optimized sheet thickness - 3nm).

[0028] In this specification, the casting parameters in S7 are set as follows: if the optimized sheet particle size is no greater than 100nm, the casting speed is 2m / min and the casting film thickness is 0.1mm; if the optimized sheet particle size is greater than 100nm but no greater than 200nm, the casting speed is 1m / min and the casting film thickness is 0.2mm.

[0029] In this manual, the sintering parameters in S8 are set as follows: if the optimized layer thickness is no more than 3nm, the sintering temperature is set to 500℃ and the holding time is set to 2h; if the optimized layer thickness is greater than 3nm but no more than 5nm, the sintering temperature is set to 600℃ and the holding time is set to 3h.

[0030] In summary, the present invention has at least the following beneficial effects:

[0031] 1. Improve process stability: Through the synergistic effect of the linear optimization model of purification parameters and the dynamic control model of delamination effect, intelligent control of raw material characteristics and purification and delamination parameters is achieved, avoiding parameter fluctuations of traditional empirical methods and significantly improving the batch stability of products.

[0032] 2. Optimize the microstructure of the material: Through closed-loop feedback of purification and delamination, the purity, lamellar thickness and particle size distribution of montmorillonite are effectively controlled, reducing impurity interference and lamellar agglomeration, and achieving precise control of the microstructure.

[0033] 3. Enhance material performance consistency: The collaborative optimization mechanism ensures that the key indicators such as interlayer spacing and mechanical properties of the final material remain stable within the target range, solving the problem of large performance fluctuations in traditional methods.

[0034] 4. Expanding the scope of applications: The high-performance montmorillonite materials prepared have controllable structure and stable performance, which can meet the stringent requirements of raw materials in high-end coatings, composite materials and other fields, thus broadening the application scenarios of montmorillonite.

[0035] 5. Improve production efficiency: Intelligent algorithms replace traditional trial and error methods, shorten the parameter optimization cycle, reduce repeated experiments, lower production costs, and are suitable for large-scale industrial production. Attached Figure Description

[0036] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. 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.

[0037] Figure 1 This is a schematic diagram of the material preparation method based on the synergistic optimization of montmorillonite purification and delamination involved in this invention.

[0038] Figure 2 This is a schematic diagram of the centrifugal purification process involved in this invention.

[0039] Figure 3 This is a schematic diagram of the ultrasonic peeling process involved in this invention.

[0040] Figure 4 This is a schematic diagram of the collaborative optimization and shaping process involved in this invention. Detailed Implementation

[0041] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of the embodiments of the invention. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0042] The following disclosure provides many different implementations or examples for carrying out different structures of the embodiments of the present invention. To simplify the disclosure of the embodiments of the present invention, specific examples of components and arrangements are described below. Of course, these are merely examples and are not intended to limit the embodiments of the present invention. Furthermore, reference numerals and / or reference letters may be repeated in different examples of the embodiments of the present invention; such repetition is for simplification and clarity and does not in itself indicate a relationship between the various implementations and / or arrangements discussed.

[0043] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0044] like Figure 1 As shown, this embodiment provides a material preparation method based on the synergistic optimization of montmorillonite purification and exfoliation, including the following steps:

[0045] S1: Pre-treat the natural montmorillonite raw material, determine the average initial particle size distribution and initial impurity content, and obtain the pre-treated montmorillonite.

[0046] S2: Based on the initial average particle size distribution and initial impurity content, the optimized centrifugation speed and optimized centrifugation time were calculated using a linear optimization model of purification parameters; the pretreated montmorillonite was centrifuged and purified according to the optimized centrifugation speed and optimized centrifugation time to obtain a purified montmorillonite suspension and determine its purity.

[0047] S3: Prepare an intercalating agent solution according to the purity, and carry out an intercalation reaction between the intercalating agent solution and the purified montmorillonite suspension to obtain a mixture of exfoliation precursors and determine its intercalation rate.

[0048] S4: Based on purity, intercalation rate and optimized centrifugation speed, the optimized ultrasonic power and optimized ultrasonic time were calculated using a dynamic control model for the delamination effect. The delamination precursor mixture was ultrasonically delaminated according to the optimized ultrasonic power and optimized ultrasonic time to obtain the initial delaminated montmorillonite dispersion and to determine its initial laminated thickness and initial laminated particle size.

[0049] S5: Calculate the synergistic optimization index based on purity, intercalation rate, initial lamella thickness, and initial lamella particle size. If the synergistic optimization index is less than the preset threshold, update the weight parameters of the linear optimization model of purification parameters and the dynamic control model of exfoliation effect respectively, and return to perform centrifugation purification, intercalation reaction, or ultrasonic exfoliation again until the synergistic optimization index is not less than the preset threshold, and obtain the optimized exfoliated montmorillonite dispersion. At the same time, measure its optimized lamella thickness and optimized lamella particle size.

[0050] S6: Prepare a stabilizer solution according to the optimized lamella thickness, and carry out a stabilization reaction between the stabilizer solution and the optimized stripped montmorillonite dispersion to obtain a stabilized stripped montmorillonite dispersion.

[0051] S7: Based on the optimized lamellae particle size, set the casting parameters, cast and dry the stabilized exfoliated montmorillonite dispersion to obtain a material blank;

[0052] S8: Set sintering parameters according to the optimized layer thickness, sinter the material blank, and obtain a montmorillonite-based material.

[0053] In some embodiments, the training process of the linear optimization model for purification parameters in S2 includes: collecting 300 sets of training datasets containing the mean initial particle size distribution of raw materials, the initial impurity content, the corresponding optimal centrifugation speed, the optimal centrifugation time, and the purification purity; using the mean initial particle size distribution and the initial impurity content as independent variables, and the optimal centrifugation speed and the optimal centrifugation time as dependent variables; fitting the linear equation using the least squares method; and minimizing the mean square error between the predicted centrifugation parameters and the actual optimal centrifugation parameters using the gradient descent method until the loss value is less than 0.01, thus completing the model training.

[0054] In some embodiments, the training process of the dynamic control model for delamination effect in S4 includes: collecting 300 training datasets containing purification purity, intercalation rate, optimized centrifugation speed, corresponding optimal ultrasonic power, optimal ultrasonic time, initial lamella thickness, and initial lamella particle size; using purification purity, intercalation rate, and optimized centrifugation speed as independent variables, and optimal ultrasonic power and optimal ultrasonic time as dependent variables; training a nonlinear regression equation using the Adam optimizer; and adjusting parameters by minimizing weighted error until the loss value is less than 0.05, thus completing model training.

[0055] In some embodiments, if the purity measured after centrifugation purification in S2 is less than 95%, the optimized centrifugation time is kept unchanged, the optimized centrifugation speed is increased by 10%, and the centrifugation purification operation is repeated. The purity of the purified montmorillonite suspension is measured again. The above adjustment and measurement steps are repeated until the purity is not less than 95%. Then, the qualified purity, the corresponding optimized centrifugation speed and optimized centrifugation time are transmitted to S3 and S5.

[0056] In some embodiments, if the initial lamella thickness measured after ultrasonic delamination in S4 is greater than 5 nm or the initial lamella particle size is greater than 200 nm, the ultrasonic power is increased by 50 W and the ultrasonic time is extended by 15 min before ultrasonic delamination is performed again. The initial lamella thickness and initial lamella particle size of the initially delaminated montmorillonite dispersion are measured again. The above adjustment and measurement steps are repeated until the initial lamella thickness is no greater than 5 nm and the initial lamella particle size is no greater than 200 nm. Then, the qualified initial lamella thickness and initial lamella particle size are transmitted to S5.

[0057] In some embodiments, the synergistic optimization index in S5 is calculated as follows: the weights of purity, intercalation rate, initial lamella thickness, and initial lamella particle size are assigned to 0.3, 0.2, 0.25, and 0.25, respectively; the synergistic optimization index is obtained by weighted summation based on the ratio of purity to target purity of 95%, the ratio of intercalation rate to target intercalation rate of 85%, the ratio of 1-lamella thickness to target thickness of 5 nm, and the ratio of 1-lamella particle size to target particle size of 200 nm.

[0058] In some embodiments, the process of updating the parameters of the purification parameter linear optimization model and the layer stripping effect dynamic control model in S5 includes: based on the difference between the co-optimization index and 1, combined with the gradient of the current model loss function, adjusting the linear equation weights of the purification parameter linear optimization model and the nonlinear equation weights of the layer stripping effect dynamic control model at a preset learning rate of 0.01, so as to ensure that the prediction accuracy of the updated model is improved.

[0059] In some embodiments, the stabilizer solution in S6 is polyvinyl alcohol, and the mass fraction of the stabilizer solution is set according to the optimized sheet thickness: if the optimized sheet thickness is not greater than 3nm, the mass fraction is 0.3%; if the optimized sheet thickness is greater than 3nm and not greater than 5nm, the mass fraction = 0.3% + 0.05% × (optimized sheet thickness - 3nm).

[0060] In some embodiments, the casting parameters in S7 are set as follows: if the optimized sheet particle size is not greater than 100nm, the casting speed is 2m / min and the casting film thickness is 0.1mm; if the optimized sheet particle size is greater than 100nm and not greater than 200nm, the casting speed is 1m / min and the casting film thickness is 0.2mm.

[0061] In some embodiments, the sintering parameters in S8 are set as follows: if the optimized sheet thickness is no more than 3nm, the sintering temperature is set to 500℃ and the holding time is set to 2h; if the optimized sheet thickness is greater than 3nm and no more than 5nm, the sintering temperature is set to 600℃ and the holding time is set to 3h.

[0062] The technical concept of this invention is as follows:

[0063] This solution addresses the problems of traditional montmorillonite preparation processes, such as reliance on experience-based trial and error for "purification-exfoliation" parameters, poor coordination, and unstable product performance. It proposes a full-process optimization method centered on a "data-driven intelligent model," ultimately achieving efficient preparation of high-performance montmorillonite materials. The solution follows a core logic of "raw material characteristics → intelligent parameter prediction → process control → closed-loop optimization → finished product preparation," comprising eight key steps. Data is rigorously transferred and mutually supportive in each step. The specific logic and core design are as follows:

[0064] S1: Montmorillonite raw material pretreatment (data acquisition and pretreatment)

[0065] S1.1 Systematic Measurement of Raw Material Characteristic Parameters

[0066] Natural montmorillonite ore from Xinghe, Inner Mongolia (a typical montmorillonite producing area), was selected, and the basic parameters were determined using the following methods:

[0067] Initial particle size distribution Determination: Take 50g of raw ore, crush it to 1-2mm particles, weigh 10g and place it in a beaker, add 50mL of deionized water, and magnetically stir for 30min (500r / min) to prepare a suspension; use a laser particle size analyzer (MalvernMastersizer 3000) to test, with a testing range of 0.01-3500μm and a dispersion pressure of 0.5bar. Test each sample three times and take the average value to obtain the result. (Mean particle size distribution), for example, measured .

[0068] Initial impurity content Determination: 2g of raw ore was ground to 300 mesh and analyzed using an X-ray fluorescence spectrometer (Bruker S8 Tiger). Test conditions: Rh target, voltage 60kV, current 50mA, scan time 300s. Impurities (quartz) were calculated by analyzing the characteristic peak intensity. calcite Feldspar The proportion of total mass, i.e. For example, measured .

[0069] Data recording and transmission: , The corresponding raw material batch number (such as XH-2023-058) is recorded in the database and synchronously transmitted to the purification parameter linear optimization model of S2.

[0070] S1.2 Raw material crushing and screening process

[0071] Coarse crushing stage: 2kg of raw ore is fed into a jaw crusher (PE-60×100 type), the crushing gap is set to 5mm, and after crushing, particles with a particle size ≤5mm are obtained. After collection, the particles are rinsed 3 times with deionized water (to remove surface dust) and dried in a 60℃ forced-air drying oven for 12h.

[0072] Fine crushing stage: Take 1 kg of dry coarse particles and put them into a planetary ball mill (QM-3SP4 type). Use agate balls (to avoid contamination), with a ball-to-material ratio of 8:1 (60% of φ10mm balls and 40% of φ5mm balls), a rotation speed of 200 r / min, and a grinding time of 30 min. After stopping the machine, take out the material and pass it through a 200-mesh and a 300-mesh standard inspection sieve (the sieve material is stainless steel, and the aperture is 75μm and 48μm respectively). Collect the material that passes through the 200-300 mesh sieve (particle size 48-75μm), which is the pretreated montmorillonite powder.

[0073] Mass measurement and transfer: The mass of the pretreated powder was weighed using an electronic balance (accuracy 0.01g). ,Will The particle morphology photos (taken with a stereomicroscope) are transmitted to S2 for dispersion preparation.

[0074] S2: Centrifugal purification of montmorillonite (Construction and application of linear optimization model for purification parameters)

[0075] S2.1 Model Construction Details

[0076] The Purification Parameter Linear Optimization Model (PPLO model) is a multiple linear regression model designed to accurately predict the optimal parameters for centrifugal purification based on the initial characteristics of the raw materials, thus avoiding the inefficiency of traditional trial-and-error methods. The centrifugal purification process is as follows: Figure 2 As shown.

[0077] Input variables:

[0078] Initial particle size of raw materials (Unit: μm, defined as the average equivalent diameter distribution of raw material particles, measured by a laser particle size analyzer at a wavelength of 532 nm, with a measurement range of 0.1-1000 μm).

[0079] Initial impurity content (The unit %, is defined as the percentage of the total mass of impurities (quartz, calcite, feldspar) to the total mass of the raw material, determined by XRF under conditions of 40kV voltage and 30mA current).

[0080] Output variables:

[0081] Optimize centrifuge speed (Unit: r / min, defined as the centrifuge rotation speed that maximizes the sedimentation efficiency of impurities).

[0082] Optimize centrifugation time (Unit: min, defined as the centrifugation duration required to fully suspend the target montmorillonite particles).

[0083] Model formula: , As the independent variable, , As the dependent variable, the least squares method is used to fit the linear equation;

[0084] (1)

[0085] (2)

[0086] in:

[0087] ~ For model weight parameters ( , , , , , (Obtained through training, its physical meaning is the influence coefficient of the corresponding variable on the output).

[0088] (Reference speed value, unit r / min) (Time baseline value, in minutes), which is the model bias term to ensure that the parameters are within a reasonable range.

[0089] S2.2 Model Training Process

[0090] Training dataset construction:

[0091] 300 sets of natural montmorillonite raw material samples were collected. Each set of samples contained:

[0092] Raw material characteristics: (50-500μm, 50μm interval) (8%–15%, with 1% intervals);

[0093] Orthogonal experiment results: Through L9(3 4 An orthogonal array was used to design an experiment to test nine sets of centrifugation parameters (speed 2000-4000 r / min, interval 500 r / min; time 10-25 min, interval 5 min), and the purification purity corresponding to each set of parameters was recorded. (Unit: %), to select the optimal parameters for each sample group. (i.e., corresponding) The largest parameter).

[0094] The final dataset is .

[0095] Loss function and training process:

[0096] The loss function is defined as the mean square error between the predicted value and the actual optimal parameters:

[0097] (3)

[0098] in , Let n be the model's predicted value for the i-th sample group, and n=300 be the sample size.

[0099] Training uses gradient descent:

[0100] Initial learning rate The number of iterations is 500.

[0101] The validation set loss is calculated once every 100 iterations (using 30 groups of samples that did not participate in training). Training stops when the validation set loss is less than 0.01 for 3 consecutive times.

[0102] The final trained model has an L=0.008 and a parameter prediction error ≤5%.

[0103] S2.3 Model Application and Centrifugation Operation Details

[0104] Example of parameter calculation for S2.3.1:

[0105] If S1 transmits (>200μm) ,but:

[0106] (1) Substitute into formula (1) to calculate :

[0107] ;

[0108] (2) Because Correction factor (Revision basis: Large-diameter raw materials require higher rotation speeds to promote impurity settling), revised ;

[0109] (3) Substitute into formula (2) to calculate :

[0110] .

[0111] S2.3.2 Centrifugation operation steps:

[0112] (1) Dispersion transfer: Transfer the pretreated montmorillonite dispersion (by mass) prepared in S1 to the pretreated montmorillonite dispersion prepared in S1. Transfer the liquid into six 500mL centrifuge tubes, each containing 300mL of liquid (to avoid spillage during centrifugation).

[0113] (2) Equipment settings: Use a GL-21M high-speed centrifuge (Hunan Xiangyi), place the centrifuge tubes symmetrically, and set the speed. ,time centrifugal acceleration (r=15cm is the centrifugal radius), the calculated value is a≈170g;

[0114] (3) Post-centrifugation treatment: After centrifugation, the solution is divided into three layers (upper layer: montmorillonite suspension; middle layer: transition layer; lower layer: impurity precipitate). Use a pipette to slowly aspirate 90% of the volume of the upper suspension (avoid aspirating the transition layer), and combine them to obtain the purified montmorillonite suspension.

[0115] S2.3.3 Purity Determination and Adjustment:

[0116] XRD (Bruker D8 Advance) was used for determination. (Test conditions: Cu target Kα rays, scanning range 2θ = 5°-60°, step size 0.02°), purity was calculated by the intensity ratio of characteristic peaks: (I represents the intensity of the diffraction peak).

[0117] like (If measured) If it is qualified, record it. , , And transmit to S3 and S5;

[0118] like (If measured) Then, start the second centrifugation: maintain Unchanged, will Increase by 10% (i.e.) After centrifugation, the measurement was repeated until... .

[0119] S3: Preparation of Montmorillonite stripping precursor

[0120] S3.1 Selection and Precise Formulation of Intercalating Agents

[0121] Purity based on S2 transmission (belonging to 95%≤) <97% range), the intercalating agent was determined to be hexadecyltrimethylammonium bromide (CTAB, analytical grade, Aladdin reagent), and its mass ratio to montmorillonite was determined. .

[0122] Preparation of intercalating agent solution: Calculation of required CTAB mass: ,in ( Let S2 be the total mass of the purified montmorillonite suspension. =3400g, solid content 5%, then Therefore Take 28.3g of CTAB and put it into a beaker. Add 500mL of deionized water and stir in a 60℃ constant temperature water bath until completely dissolved (300r / min) to obtain a CTAB solution.

[0123] S3.2 Refined Operation of Intercalation Response

[0124] Reaction system setup: The purified montmorillonite suspension (3400g) obtained from S2 was transferred to a 2L three-necked flask, and a mechanical stirrer (5cm blade diameter) and reflux condenser were installed. The constant temperature water bath temperature was set to 70℃ (because...). If the humidity is in the 95%–97% range, choose the intermediate temperature.

[0125] Gradient feeding and reaction control: The CTAB solution was slowly added to the suspension through a constant-pressure dropping funnel at a dropping rate of 2 mL / min (to avoid excessively high local concentrations). After the addition was complete (approximately 4 hours), stirring was continued for another 4 hours (according to...). (The range is determined), the stirring speed is stabilized at 300 r / min, and the pH value of the system is recorded every 30 min (maintained at 7.0±0.2, if it deviates, it is adjusted with 0.1 mol / L HCl or NaOH).

[0126] Endpoint determination: Take 10 mL of reaction solution, centrifuge (4000 r / min, 10 min), collect the supernatant, and measure the absorbance at 220 nm using a UV-Vis spectrophotometer (Shimadzu UV-2600). Compare the absorbance with the initial CTAB solution absorbance to calculate the intercalation rate. For example, measured (≥85% qualified).

[0127] S3.3 Data Processing and Transmission

[0128] Mass of the exfoliation precursor mixture Intercalation rate The reaction temperature curve was recorded and transmitted to S4 as input parameters for the dynamic control model of the delamination effect; at the same time, 50 mL of the mixture was stored in a refrigerator (4℃) for later use in traceability analysis.

[0129] S4: Ultrasonic stripping treatment of montmorillonite (construction and application of dynamic control model for stripping effect)

[0130] S4.1 Model Construction Details

[0131] The dynamic control model for the exfoliation effect is a nonlinear regression model. It dynamically adjusts the ultrasonic exfoliation parameters by controlling key parameters of the purification process, achieving precise control of the sheet thickness and particle size. The ultrasonic exfoliation process is as follows: Figure 3 As shown.

[0132] Input variables:

[0133] Purification (Unit: %, i.e., the purity of montmorillonite measured in S2).

[0134] Intercalation rate (The unit is %, which is defined as the proportion of intercalating agent molecules entering the interlayer of montmorillonite, and is determined by a UV-Vis spectrophotometer at a wavelength of 220 nm.)

[0135] Optimize centrifuge speed (Unit: r / min, i.e., the output parameters of the model in S2).

[0136] Output variables:

[0137] Optimize ultrasonic power (The unit W is defined as the output power of the ultrasonic equipment, which determines the ultrasonic intensity.)

[0138] Optimize ultrasound time (Unit: min, defined as the duration of ultrasonic treatment).

[0139] Model formula:

[0140] (4)

[0141] (5)

[0142] in:

[0143] ~ For model weight parameters ( , , , , , (Obtained through training, reflecting the degree of influence of input variables on ultrasound parameters).

[0144] (Power reference correction value, in W) (Time base correction value, unit: min) Ensure that the output parameters are within the valid range (power 200-800W, time 20-120min).

[0145] S4.2 Model Training Process

[0146] Training dataset construction:

[0147] Collect 300 sets of samples, each set containing:

[0148] Input parameters: (95%–99%, with an interval of 0.5%) (85%–95%, with 1% intervals) (800-1200 r / min, with 50 r / min intervals);

[0149] Experimental Results: Single-factor experiments were conducted to test different ultrasound parameters (power 200–800 W, intervals of 100 W; time 20–120 min, intervals of 20 min), and the corresponding slice thicknesses were recorded. (Unit: nm, measured by AFM) and lamellar particle size (Unit: nm, DLS measurement) The optimal ultrasound parameters for each sample group were selected. (i.e., simultaneously satisfying) and (parameters).

[0150] Loss function and training process:

[0151] The loss function is defined as the weighted error between the layer metric and the target value:

[0152] (6)

[0153] in (Thickness weight, as thickness has a greater impact on performance) (Particle size weight), n=300 is the sample size.

[0154] Training was performed using the Adam optimizer:

[0155] Initial learning rate Iterations 800 times, batch size 32;

[0156] The validation set loss (30 samples) is calculated every 200 iterations. Training should be stopped immediately.

[0157] The final trained model has an L'=0.042, a sheet thickness prediction error ≤0.3nm, and a particle size prediction error ≤10nm.

[0158] S4.3 Model Application and Ultrasonic Operation Details

[0159] Example of parameter calculation for S4.3.1:

[0160] If S3 transmission , S2 transmission ,but:

[0161] (1) Substitute into formula (4) to calculate :

[0162] ;

[0163] (2) Substitute into formula (5) to calculate :

[0164] .

[0165] S4.3.2 Ultrasonic operation procedures:

[0166] (1) Equipment preparation: A JY92-IIN ultrasonic cell disruptor (Ningbo Xinzhi) with a probe diameter of 6mm was used to prepare the mixture of the exfoliated precursor obtained in S3 (by mass). Transfer the liquid to a 1000mL beaker, ensuring the liquid level is ≥5cm (to avoid probe idling).

[0167] (2) Parameter setting: Set power ,time The ultrasonic mode operates for 3 seconds followed by a 2-second interval (to prevent the solution from overheating), and the system temperature is controlled to be ≤35℃ using a constant temperature water bath (circulating water temperature 25℃).

[0168] (3) Ultrasonic process monitoring: Take 5 mL of sample every 5 min and use a laser particle size analyzer (Mastersizer 3000) to quickly measure the particle size change. If the particle size change is <5 nm within 5 min, the ultrasonic process can be terminated in advance.

[0169] S4.3.3 Delamination effect measurement and adjustment:

[0170] Layer thickness Measurement: Take 1 mL of dispersion, dilute it 100 times, drop it onto a mica sheet and let it dry naturally. Use AFM (Bruker Dimension Icon) in tapping mode to scan (scanning range 5 μm × 5 μm, resolution 512 × 512), and count the average thickness of 100 sheets.

[0171] Laminar particle size Determination: Take 0.5 mL of dispersion, dilute 50 times, and measure it using DLS (Malvern Zetasizer NanoZS) at 25℃ (scattering angle 90°, take the average value of 3 measurements).

[0172] like and (If measured) , If it is qualified, record it. , , , And transmit to S5;

[0173] If the standard is not met (e.g.) , If so, adjust the parameters: Increase by 50W (to 595.4W). Extend the time by 15 minutes (to 50.1 minutes), repeat the ultrasound measurement, and continue until the target is met.

[0174] S5: Co-optimization and adjustment of purification and delamination processes (co-optimization and molding process as follows) Figure 4 (As shown)

[0175] S5.1 Calculation of the Collaborative Optimization Index K

[0176] Based on S2 transmission , S3 transmission S4 transmission , Calculate the collaborative optimization index:

[0177] ;

[0178] Substitute the weighting coefficients , , , :

[0179] ;

[0180] Since K=0.59<0.85, parameter feedback update needs to be initiated.

[0181] S5.2 Model Parameter Dynamic Update

[0182] The purification parameters are updated according to the formula, the linear optimization model is optimized, and the model weights are dynamically adjusted based on the peeling effect.

[0183] Purification parameter linear optimization model weight update: ;

[0184] in , , The gradient of the loss function (assuming) If the gradient is 0.02, then... ;

[0185] Dynamically adjust model weight updates to achieve layer-by-layer peeling effect: ;

[0186] Assumption The gradient is 0.05, then .

[0187] S5.3 Iterative Optimization Operation

[0188] Recalculate using the updated purification parameters and linear optimization model. , Return to S2 for secondary purification to obtain ;

[0189] based on The S3 intercalation reaction was repeated to obtain ;

[0190] Calculate using the updated peeling effect dynamic control model Return to S4 for a second ultrasound to obtain... ;

[0191] Recalculate K = 0.92 ≥ 0.85, optimization is successful, and... , , , and the quality of the optimized dispersion Transmitted to S6.

[0192] S6: Stripped montmorillonite stabilization treatment

[0193] S6.1 Stabilizer Selection and Concentration Calculation

[0194] According to S5 transmission (3nm < For samples within the ≤5nm range, polyvinyl alcohol (PVA, molecular weight 1750±50, Sinopharm Group) was selected as the stabilizer, and its mass fraction was calculated using the formula:

[0195] ;

[0196] Calculate the required PVA quality: .

[0197] S6.2 Stabilization Reaction Process

[0198] PVA solution preparation: Take 11.78g of PVA and put it into a beaker. Add 500mL of deionized water and stir in a 90℃ constant temperature water bath until completely dissolved (200r / min). Cool to 40℃ for later use.

[0199] Mixing reaction: The optimized stripping montmorillonite dispersion (3800g) was transferred into a 5L reactor, and the temperature was set at 45℃ (3nm < (≤5nm corresponds to the intermediate temperature), stir at 200 rpm, slowly add PVA solution (completely add within 30 minutes), continue stirring for 2 hours (according to...). (Scope defined).

[0200] S6.3 Stability Testing and Validation

[0201] Standing test: Take 50 mL of the stabilized dispersion and put it into a stoppered graduated cylinder. Mark the liquid level and let it stand in a constant temperature environment of 25℃ for 72 h. Observe whether there is a precipitate layer (a height ≥ 0.5 cm is considered unqualified). In this case, there was no obvious precipitation after 72 h, and the stability was qualified.

[0202] Data transmission: This will transmit the mass of the stabilized montmorillonite dispersion. stabilizer concentration Transfer still photos to S7.

[0203] S7: Material Forming and Drying

[0204] S7.1 Casting Molding Parameter Settings

[0205] According to S5 transmission (100nm < (≤200nm range), select casting equipment (ZLY-100 type), set parameters:

[0206] Casting speed: 1m / min (due to the large particle size, the speed needs to be reduced to ensure uniformity);

[0207] Doctor blade height: 0.2mm (corresponding to a cast film thickness of 0.2mm);

[0208] Environmental control: molding chamber temperature 25±1℃, relative humidity 40±5% (to prevent the film surface from drying and cracking too quickly).

[0209] S7.2. Molding and Drying Operations

[0210] Pretreatment of the material: The stabilized montmorillonite dispersion (4300g) was filtered through a 1μm filter membrane (to remove possible agglomerates) and degassed under vacuum for 30min (vacuum degree -0.09MPa).

[0211] Casting process: Pour the degassed liquid into the casting machine trough, start the equipment, and the base belt (PET film, 50μm thick) carries the liquid through the doctor blade to form a continuous film preform. It is initially dried on the conveyor belt for 30 minutes (hot air temperature 40℃).

[0212] Drying process: The pre-dried film blank (with the PET base tape peeled off) is transferred to a forced-air drying oven, according to... (Approximately 0.3%) Set the drying temperature to 60℃ and the drying time to 6 hours (due to the relatively long time corresponding to a casting speed of 1m / min). Turn the film blank over every 2 hours (to ensure uniform drying).

[0213] S7.3. Billet Performance Testing

[0214] Density determination: The density of the dried green body was determined using the water displacement method (Archimedes' principle). Take three 1cm×1cm samples, test them, and take the average value. (Qualified within the range of 1.2-1.5 g / cm³).

[0215] Data transmission: transferring the quality of the billet (Moisture content <1%) The preform appearance photo is transmitted to S8.

[0216] S8: Material Sintering and Performance Characterization

[0217] S8.1 Sintering Process Parameter Settings

[0218] According to S5 transmission (3nm < (≤5nm range), set sintering parameters:

[0219] Sintering equipment: Box-type muffle furnace (SX2-12-10 type);

[0220] Sintering temperature: 600℃;

[0221] Heating curve: room temperature → 300℃ (rate 5℃ / min, hold for 1h, PVA removal) → 600℃ (rate 10℃ / min, hold for 3h);

[0222] Cooling method: After naturally cooling to 300℃, cool to room temperature with the furnace (to avoid rapid cooling and cracking).

[0223] Comprehensive Performance Characterization of S8.2

[0224] Phase analysis: Interlayer spacing was determined using XRD. The test conditions are the same as S2, and the Bragg equation is used. (λ=0.154nm) Calculated and measured (Qualified within the range of 1.8-2.2nm).

[0225] Microstructure: Observation was performed using TEM (JEOL JEM-2100) with an accelerating voltage of 200kV. The powder sample was ultrasonically dispersed in ethanol and dropped onto a copper grid for observation. The sheets were uniformly dispersed with no obvious agglomeration.

[0226] Mechanical properties: Tensile strength was tested using a universal tensile testing machine (Instron 5967). The sample size was 50mm × 10mm × 0.1mm, the tensile rate was 2mm / min, and the average value of 5 samples was taken. (Qualified within the range of 15-20MPa).

[0227] S8.3 Final Product Confirmation and Archiving

[0228] All performance indicators meet the standards, and the quality of the final material is recorded. (Weight loss after sintering: 8.1%) , Generate a product qualification report, including a full-process parameter traceability table (with key data and graphs for each step), and complete the preparation process.

[0229] In some embodiments, a surface fitting algorithm is introduced to construct a nonlinear mapping relationship between purification-stripping parameters and final performance. The surface fitting optimization model receives the output parameters from the purification parameter linear optimization model and the stripping effect dynamic control model, generates a performance prediction surface, and the result feeds back into the parameter updates of the purification parameter linear optimization model and the stripping effect dynamic control model, forming a three-level collaborative mechanism of "linear prediction-nonlinear correction-parameter feedback".

[0230] Centrifugal purification of montmorillonite

[0231] Purification of parameters: Linear optimization model output parameter expansion: Based on the original output... (Optimize centrifuge speed) Based on the optimized centrifugation time, the following output parameters are added:

[0232] centrifugal separation coefficient (unit The ratio of centrifugal acceleration to gravitational acceleration is defined as:

[0233] ;

[0234] Where r = 0.15m (centrifugal radius) and g = 9.8m / s² (gravitational acceleration).

[0235] Example: hour, .

[0236] The basic data transferred to the surface fitting optimization model:

[0237] Will , (Purification level) and corresponding raw material characteristics ,

[0238] Packaged as a dataset The surface fitting optimization model is transmitted to S5 as one of the input variables for surface fitting.

[0239] Montmorillonite ultrasonic stripping treatment

[0240] Dynamic adjustment of peeling effect model output parameters expansion: based on the original output (Optimize ultrasonic power) Based on the optimized ultrasound time, the following output parameters are added:

[0241] Ultrasonic energy density (unit ), defined as the ratio of total ultrasonic energy to the volume processed:

[0242] ;

[0243] Where V is the processing volume (in mL), in the example , If V = 3900 mL, then .

[0244] Basic data transferred to the surface fitting optimization model

[0245] Will , (Laminar thickness) (Layer size) packaged into a dataset The surface fitting optimization model is transmitted to S5 as another set of input variables for surface fitting.

[0246] Synergistic optimization and adjustment of purification and stripping

[0247] Algorithm 3: Construction and Application of Surface Fitting Optimization Model

[0248] 1. Model Building Details

[0249] Input variables:

[0250] Purification characteristic parameters: (Centrifugal separation coefficient, ), (Purity, %)

[0251] Peeling feature parameters: (ultrasound energy density, ), (Layer thickness, nm);

[0252] Output variables:

[0253] Performance prediction: Predicted tensile strength (MPa), predicted interlayer spacing (nm);

[0254] Model formula (ternary quadratic surface):

[0255] ;

[0256] ;

[0257] in: - , - The coefficients are the surface fitting coefficients (obtained through training, where u corresponds to tensile strength and v corresponds to interlayer spacing).

[0258] 2. The entire model training process

[0259] Training dataset construction:

[0260] Collect 500 sets of experimental data, each set containing:

[0261] Input parameters: (100~300m / s²) (95%~99%) (200~400J / mL) (2-5nm);

[0262] Output parameters: Measured tensile strength (15-20MPa), measured interlayer spacing (1.8-2.2nm).

[0263] Parameter estimation methods:

[0264] The coefficients are solved using the least squares method, and the objective function is to minimize the sum of squared residuals between the predicted and measured values.

[0265] ;

[0266] ;

[0267] Solving for coefficients using matrix operations, for example , , , , .

[0268] Training effect verification:

[0269] Mean absolute error of the validation set (100 groups): Prediction error ≤ 0.3 MPa Prediction error ≤ 0.05 nm, goodness of fit R² ≥ 0.96.

[0270] 3. Model Application and Interaction of Three Algorithms

[0271] Performance prediction example:

[0272] Transmit S2 , and S4 transmission , Substitute into the formula:

[0273] ;

[0274] ;

[0275] Interaction mechanism with the purification parameter linear optimization model / exfoliation effect dynamic control model:

[0276] 1. Calculate the performance deviation index :

[0277] ;

[0278] Target value: , The weights are 0.6 and 0.4, respectively.

[0279] Example: .

[0280] 2. Based on Correcting purification parameters in a linear optimization model / dynamically adjusting stripping effect model parameters:

[0281] Purification parameter linear optimization model weight correction: ( (The updated weights in S5)

[0282] Example: ;

[0283] Dynamic adjustment of model weights to improve peeling effect: ( (The updated weights in S5)

[0284] Example: .

[0285] 3. Iterative optimization: using the corrected version and Recalculate , Parameters are returned to S2 / S4 for processing until... (After 2 iterations in the example) (meets the requirements).

[0286] Material sintering and performance characterization (verification and feedback of surface fitting optimization model)

[0287] 1. Comparison of Actual Performance Tests and Predictions

[0288] Actual performance of the final product: , The relative errors between the predicted values ​​and those of the surface fitting optimization model were 1.6% and 0%, respectively, verifying the effectiveness of the model.

[0289] 2. Feedback data to the surface fitting optimization model

[0290] Compare the final measured value with the corresponding input parameters New samples are created and added to the training dataset of the surface fitting optimization model. The model is retrained periodically (every 50 batches), and the coefficients are updated. and This ensures the model's adaptability.

[0291] Core Logic and Contributions of the Integration of Three Algorithms

[0292] 1. Collaborative Path:

[0293] Purification parameter linear optimization model → Output → Input to the surface fitting optimization model;

[0294] Dynamic control model for peeling effect → Output → Input to the surface fitting optimization model;

[0295] Surface fitting optimization model → Output →Modify purification parameters linear optimization model / dynamically adjust stripping effect model parameters.

[0296] 2. Core Contributions:

[0297] To address the fitting bias of the linear model to the nonlinear relationship between centrifugal intensity, ultrasonic energy, and performance, the performance prediction error is reduced to a minimum.

[0298] pass The degree of performance deviation is quantified exponentially, reducing the blindness of collaborative optimization;

[0299] Dynamic correction of parameters throughout the entire process improves product qualification rate.

[0300] Experimental data and process records

[0301] Step S1: Experimental Materials and Basic Parameters

[0302] Raw material batch: Montmorillonite ore from Xinghe County, Inner Mongolia (No. XH-2023-068);

[0303] Initial parameter determination:

[0304] Initial particle size distribution Measurements were taken using a Malvern Mastersizer 3000. Three repeated tests yielded results of 248 μm, 252 μm, and 250 μm, with the mean value being... ;

[0305] Initial impurity content XRF analysis results (mass fraction): Quartz 10.2%, Calcite 1.5%, Feldspar 0.3%, Total impurities ;

[0306] Powder quality after pretreatment (68.5% yield of material under 200-300 mesh sieve).

[0307] S2: Centrifugal purification of montmorillonite

[0308] 1. Parameter calculation of the linear optimization model for purification parameters

[0309] Input parameters: , ;

[0310] Model calculation:

[0311] ;

[0312] because Correction factor After correction

[0313] .

[0314] 2. Centrifugation Experiment Procedure

[0315] Equipment: GL-21M high-speed centrifuge;

[0316] Dispersion preparation: Pre-treated powder, at a solid-liquid ratio of 1:10 (because...) Add 6850mL of deionized water and add 1.0% sodium hexametaphosphate (68.5g).

[0317] Centrifugation parameters: rotation speed 1027 r / min, time 73.2 min, centrifugation radius 15 cm, acceleration 172 m / s². 2 ;

[0318] Separation results: The volume of the upper suspension was 5800 mL, and the mass was... (4.8% solids content).

[0319] 3. Purity testing

[0320] XRD test (Bruker D8 Advance): Intensity of montmorillonite characteristic peak (2θ=5.8°) Total intensity of impurity peaks ;

[0321] Purity calculation: (≥95% qualified);

[0322] Data transmission: , .

[0323] S3: Preparation of Montmorillonite stripping precursor

[0324] 1. Intercalating agent preparation

[0325] Montmorillonite quality: ;

[0326] Intercalating agent ratio (Because 95%≤) <97%), required CTAB mass = 284.16 ÷ 6 ≈ 47.36g;

[0327] Prepare 500 mL of CTAB solution (concentration 9.47%).

[0328] 2. Intercalation response

[0329] Equipment: 2L three-necked flask, constant temperature water bath (70℃), mechanical stirrer (300r / min);

[0330] Reaction process: CTAB solution was added dropwise at a rate of 2 mL / min (completed in 4 hours), and stirring was continued for another 4 hours, with the pH maintained at 7.0 ± 0.2;

[0331] Intercalation rate detection: measured by ultraviolet spectroscopy (220 nm) , , ;

[0332] Data transmission: Mixture quality .

[0333] S4: Montmorillonite ultrasonic stripping treatment

[0334] 1. Calculation of parameters for dynamic control model of peeling effect

[0335] Input parameters: , , .

[0336] Model calculation:

[0337] ;

[0338] .

[0339] 2. Ultrasonic Experiment Procedure

[0340] Equipment: JY92-IIN ultrasonic cell disruptor (6mm probe);

[0341] Operating parameters: power 545.4W, time 35.1min, working time 3s / intermittent time 2s, water temperature control 25±1℃;

[0342] Ultrasonic energy density calculation (Formula 11): .

[0343] 3. Delamination effect detection

[0344] AFM test (Bruker Dimension Icon): Average thickness of 100 layers ;

[0345] DLS test (Malvern Zetasizer): Volume average particle size ;

[0346] Data transmission: , , .

[0347] S5: Collaborative Optimization Adjustment (Three Algorithms Fusion)

[0348] 1. Calculation of the Co-optimization Index K

[0349] ;

[0350] Since K < 0.85, the surface fitting optimization model is initiated.

[0351] 2. Performance Prediction of Surface Fitting Optimization Model

[0352] Input parameters: , , , ;

[0353] Prediction results: , ;

[0354] Performance Deviation Index: .

[0355] 3. Model parameter correction and iteration

[0356] Purification parameter linear optimization model weight correction: ,like ;

[0357] Dynamic adjustment of model weights to improve peeling effect: ,like ;

[0358] Secondary optimization results: , Ultimately, K=0.92. ;

[0359] Data transmission: , , , .

[0360] S6-S8: Stabilization, shaping and sintering:

[0361]

[0362] Algorithm fusion effect verification data:

[0363]

[0364] Experimental conclusions

[0365] 1. The three-algorithm fusion system optimizes centrifugation parameters through a linear optimization model for purification parameters, controls ultrasonic parameters through a dynamic control model for exfoliation effect, and corrects nonlinear relationships through a surface fitting optimization model, thereby achieving intelligent coordination of the entire process of montmorillonite purification-exfoliation.

[0366] 2. Experimental data show that after introducing the surface fitting algorithm, the stability of product performance indicators (tensile strength and interlayer spacing) is significantly improved, with the standard deviation decreasing from ±1.5MPa to ±0.3MPa and from ±0.15nm to ±0.05nm, respectively.

[0367] 3. Algorithm iteration shortened the preparation cycle from 48 hours to 24 hours using traditional methods, increasing production efficiency by 50%, thus verifying the feasibility and superiority of the method.

[0368] The embodiments described above are for illustrative purposes only and are not intended to limit the invention. Therefore, any changes in numerical values ​​or substitutions of equivalent elements should still fall within the scope of this invention.

[0369] The above detailed description will enable those skilled in the art to understand that the present invention can indeed achieve the aforementioned objectives and has complied with the provisions of the Patent Law.

[0370] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention. The above descriptions are merely preferred embodiments of the invention and are not intended to limit the invention. It should be noted that any modifications, equivalent substitutions, and improvements made within the spirit and principles of the invention should be included within the scope of protection of the invention.

[0371] It should be noted that the above description of the process is for illustrative purposes only and does not limit the scope of this specification. Those skilled in the art can make various modifications and changes to the process under the guidance of this specification. However, these modifications and changes remain within the scope of this specification.

[0372] The basic concepts have been described above. Obviously, for those skilled in the art who have read this application, the above disclosure is merely illustrative and does not constitute a limitation of this application. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and corrections to this application. Such modifications, improvements, and corrections are suggested in this application, and therefore, such modifications, improvements, and corrections still fall within the spirit and scope of the exemplary embodiments of this application.

[0373] Furthermore, this application uses specific terms to describe its embodiments. For example, "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic related to at least one embodiment of this application. Therefore, it should be emphasized and noted that "an embodiment," "one embodiment," or "an alternative embodiment" mentioned twice or more in different positions in this specification do not necessarily refer to the same embodiment. In addition, certain features, structures, or characteristics in one or more embodiments of this application can be appropriately combined.

Claims

1. A material preparation method based on the synergistic optimization of montmorillonite purification and exfoliation, characterized in that, Includes the following steps: S1: Pre-treat the natural montmorillonite raw material, determine the average initial particle size distribution and initial impurity content, and obtain the pre-treated montmorillonite. S2: Based on the initial average particle size distribution and initial impurity content, the optimized centrifugation speed and optimized centrifugation time were calculated using a linear optimization model of purification parameters; the pretreated montmorillonite was centrifuged and purified according to the optimized centrifugation speed and optimized centrifugation time to obtain a purified montmorillonite suspension and determine its purity. S3: Prepare an intercalating agent solution according to the purity, and carry out an intercalation reaction between the intercalating agent solution and the purified montmorillonite suspension to obtain a mixture of exfoliation precursors and determine its intercalation rate. S4: Based on purity, intercalation rate and optimized centrifugation speed, the optimized ultrasonic power and optimized ultrasonic time were calculated using a dynamic control model for the delamination effect. The delamination precursor mixture was ultrasonically delaminated according to the optimized ultrasonic power and optimized ultrasonic time to obtain the initial delaminated montmorillonite dispersion and to determine its initial laminated thickness and initial laminated particle size. S5: Calculate the synergistic optimization index based on purity, intercalation rate, initial lamella thickness, and initial lamella particle size. If the synergistic optimization index is less than the preset threshold, update the weight parameters of the linear optimization model of purification parameters and the dynamic control model of exfoliation effect respectively, and return to perform centrifugation purification, intercalation reaction, or ultrasonic exfoliation again until the synergistic optimization index is not less than the preset threshold, and obtain the optimized exfoliated montmorillonite dispersion. At the same time, measure its optimized lamella thickness and optimized lamella particle size. S6: Prepare a stabilizer solution according to the optimized lamella thickness, and carry out a stabilization reaction between the stabilizer solution and the optimized stripped montmorillonite dispersion to obtain a stabilized stripped montmorillonite dispersion. S7: Based on the optimized lamellae particle size, set the casting parameters, cast and dry the stabilized exfoliated montmorillonite dispersion to obtain a material blank; S8: Set sintering parameters according to the optimized layer thickness, sinter the material blank, and obtain a montmorillonite-based material.

2. The material preparation method based on the synergistic optimization of montmorillonite purification and exfoliation according to claim 1, characterized in that, The training process of the linear optimization model for purification parameters in S2 includes: collecting 300 training datasets containing the mean initial particle size distribution of raw materials, initial impurity content, corresponding optimal centrifugation speed, optimal centrifugation time, and purification purity; using the mean initial particle size distribution and initial impurity content as independent variables, and the optimal centrifugation speed and optimal centrifugation time as dependent variables; fitting the linear equation using the least squares method; and minimizing the mean square error between the predicted centrifugation parameters and the actual optimal centrifugation parameters using the gradient descent method until the loss value is less than 0.01, thus completing the model training.

3. The material preparation method based on the synergistic optimization of montmorillonite purification and exfoliation according to claim 1, characterized in that, The training process of the dynamic control model for delamination effect in S4 includes: collecting 300 training datasets containing purification purity, intercalation rate, optimized centrifugation speed, corresponding optimal ultrasonic power, optimal ultrasonic time, initial lamella thickness, and initial lamella particle size; using purification purity, intercalation rate, and optimized centrifugation speed as independent variables, and optimal ultrasonic power and optimal ultrasonic time as dependent variables; training the nonlinear regression equation using the Adam optimizer; and adjusting the parameters by minimizing the weighted error until the loss value is less than 0.05, thus completing the model training.

4. The material preparation method based on the synergistic optimization of montmorillonite purification and exfoliation according to claim 1, characterized in that, If the purity measured after centrifugation purification in S2 is less than 95%, keep the optimized centrifugation time unchanged, increase the optimized centrifugation speed by 10%, and repeat the centrifugation purification operation. Measure the purity of the purified montmorillonite suspension again. Repeat the adjustment and measurement steps until the purity is not less than 95%. Then, transmit the qualified purity, the corresponding optimized centrifugation speed, and the optimized centrifugation time to S3 and S5.

5. The material preparation method based on the synergistic optimization of montmorillonite purification and exfoliation according to claim 1, characterized in that, If the initial lamella thickness is greater than 5 nm or the initial lamella particle size is greater than 200 nm after ultrasonic delamination in S4, the ultrasonic power is increased by 50 W and the ultrasonic time is extended by 15 min before ultrasonic delamination is performed again. The initial lamella thickness and initial lamella particle size of the initially delaminated montmorillonite dispersion are measured again. The adjustment and measurement steps are repeated until the initial lamella thickness is no greater than 5 nm and the initial lamella particle size is no greater than 200 nm. Then the qualified initial lamella thickness and initial lamella particle size are transmitted to S5.

6. The material preparation method based on the synergistic optimization of montmorillonite purification and exfoliation according to claim 1, characterized in that, In S6, the stabilizer solution is polyvinyl alcohol. The mass fraction of the stabilizer solution is set according to the optimized sheet thickness: if the optimized sheet thickness is no more than 3nm, the mass fraction is 0.3%; if the optimized sheet thickness is greater than 3nm but no more than 5nm, the mass fraction = 0.3% + 0.05% × (optimized sheet thickness - 3nm).

7. The material preparation method based on the synergistic optimization of montmorillonite purification and exfoliation according to claim 1, characterized in that, The casting parameters in S7 are set as follows: if the optimized sheet particle size is no greater than 100nm, the casting speed is 2m / min and the casting film thickness is 0.1mm; if the optimized sheet particle size is greater than 100nm but no greater than 200nm, the casting speed is 1m / min and the casting film thickness is 0.2mm.

8. The material preparation method based on the synergistic optimization of montmorillonite purification and exfoliation according to claim 1, characterized in that, The sintering parameters in S8 are set as follows: if the optimized layer thickness is no more than 3nm, the sintering temperature is set to 500℃ and the holding time is set to 2h; if the optimized layer thickness is greater than 3nm but no more than 5nm, the sintering temperature is set to 600℃ and the holding time is set to 3h.

Images