A Membrane Electrode Design Optimization Method Based on Variable Resolution Catalytic Layer 3D Reconstruction

Abstract

This invention relates to a membrane electrode design optimization method based on three-dimensional reconstruction of a catalytic layer with variable resolution, comprising the following steps: constructing a numerical reconstruction model of the catalytic layer based on existing experimental characterization results; validating the effectiveness of the numerical reconstruction model of the catalytic layer; if the verification is successful, proceeding to the next step; otherwise, adjusting the growth probability of each phase and returning to the previous step to adjust the numerical reconstruction model of the catalytic layer; adjusting the parameters of the numerical reconstruction model of the catalytic layer, and analyzing the catalyst utilization and transport characteristics within the catalytic layer after parameter adjustment; and determining electrode parameters based on the analysis results to guide the fabrication of the membrane electrode. Compared with existing technologies, this invention has advantages such as reduced membrane electrode fabrication costs and strong method versatility.

Description

Technical Field

[0001] This invention relates to the field of membrane electrode fabrication technology, and in particular to a membrane electrode design optimization method based on three-dimensional reconstruction of a variable resolution catalytic layer. Background Technology

[0002] Calculating the properties of the catalyst layer of a proton exchange membrane electrode using reconstruction methods is complex, and currently, no universal algorithm has been proposed that can be widely applied to the reconstruction of the catalyst layer of a proton exchange membrane electrode. In the fields of proton exchange membrane fuel cells and proton exchange membrane electrolyzers, there are two main reconstruction methods for analyzing the microstructure of their catalyst layers. The first is reconstruction analysis based on experimental characterization results. Based on the Nano-CT or FIB / SEM cross-sectional characterization results of the corresponding catalyst layer, researchers use interpolation methods to reconstruct the microstructure of the catalyst layer, thereby optimizing the preparation process to obtain the best performance and optimized transport characteristics. However, the method of reconstructing the catalyst layer based on experimental characterization requires very high equipment resolution; currently, the highest resolution of Nano-CT is only 50 nm, which cannot distinguish between ionomers and catalyst particles. FIB / SEM technology is destructive sample preparation and requires a large amount of characterization data, making it time-consuming, labor-intensive, and extremely costly. The second method is computer-based reconstruction using algorithms. This method does not require actual membrane electrode preparation, consumes very little time and cost, and can quickly obtain the catalyst layer structure with the target loading in a short period. However, current computer reconstruction methods can only determine the anisotropic parameters in the reconstruction method through the macroscopic morphology information of the non-porous parts in the catalyst layer, and the minimum resolution for reconstruction is the same for all substances. It is impossible to set a corresponding appropriate voxel resolution for each phase, which often leads to substances with a large unit volume appearing as unattainably small in the reconstruction results, affecting the accuracy of the model. In addition, computer models also have problems such as not being able to consider the effective phase loss caused by catalyst or support agglomeration, and lacking the versatility between multiple proton exchange membrane base membrane electrode catalytic layers.

[0003] In existing technologies, CN111986736A proposes a method for estimating fuel cell catalyst layer parameters based on process three-dimensional reconstruction. While this method introduces the concept of catalyst layer reconstruction, its primary purpose is to calculate the gas diffusion coefficient within the catalyst layer. Furthermore, this method sets different components in the catalyst layer (carbon spheres, Pt particles, ionomers, and pores) to the same voxel resolution for reconstruction growth, significantly reducing model accuracy. Typically, carbon supports have a diameter of 30-50 nm, while platinum spheres have a diameter of 2-4 nm. This leads to a severe underestimation of the unit size of the carbon support in the generated reconstructed catalyst layer, causing the model to lose accuracy and preventing the study of the influence of carbon support size on the physicochemical parameters in the reconstruction model. Additionally, this model is programmed using Matlab, resulting in high time costs. CN113506895A proposes a method for analyzing the performance of fuel cell catalyst layers based on the influence of relative humidity. While this method proposes reconstruction methods for catalyst layers with different unit sizes, its primary purpose is to calculate the local oxygen transport resistance near the catalytic active sites, without considering the catalyst utilization rate within the catalyst layer or investigating the impact of different catalyst layer preparation processes on the utilization rate of precious metals in the catalyst layer. CN111929338A proposes a three-dimensional reconstruction method for fuel cell catalyst layers based on simulated annealing algorithms. Although this method makes the catalyst layer reconstruction results more realistic through annealing algorithms, the reconstruction method itself is feedforward-oriented, requiring the actual catalyst layer to be prepared and a digital model to be obtained based on a large number of cross-sectional characterization results. This results in the algorithm itself lacking general applicability, only able to analyze existing preparation results, and unable to provide feedforward guidance for PEM membrane electrode device fabrication processes. Furthermore, the large number of cross-sectional characterization results makes this method extremely time-consuming and resource-intensive. Summary of the Invention

[0004] The purpose of this invention is to provide a membrane electrode design optimization method based on three-dimensional reconstruction of the catalyst layer with variable resolution. This method addresses the issues of cost, time consumption, and measurement result deviations caused by actual preparation and online electrochemical testing of the physicochemical properties of the catalyst layer. It obtains the transport and power performance limits of proton exchange membrane-based electrodes using traditional membrane electrode preparation processes at a target loading with extremely low computational cost. Simultaneously, it determines important process parameters such as optimal catalyst layer porosity, optimal carbon support size, catalyst slurry dispersion, and ionomer loading, enabling membrane electrode performance prediction and optimization. This method is applicable to the optimization and feedforward prediction of catalyst utilization and mass transfer characteristics in proton exchange membrane-based electrodes, significantly reducing production costs.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A membrane electrode design optimization method based on three-dimensional reconstruction of a catalytic layer with variable resolution includes the following steps:

[0007] S1. Construct a numerical reconstruction model of the catalytic layer based on existing experimental characterization results;

[0008] S2. Verify the effectiveness of the numerical reconstruction model of the catalyst layer. If the verification is successful, proceed to step S3. Otherwise, adjust the growth probability of each phase and return to step S1 to adjust the numerical reconstruction model of the catalyst layer.

[0009] S3. Adjust the parameters of the numerical reconstruction model of the catalyst layer and analyze the catalyst utilization and transport characteristics in the catalyst layer after adjusting the parameters;

[0010] S4. Based on the analysis results of S3, feedforward to determine the electrode parameters to guide the fabrication of the membrane electrode.

[0011] Step S1 includes the following steps:

[0012] S11. Identification and statistics of phase material classification characteristics in the catalyst layer: Based on existing data, the phase material classification characteristics in the catalyst layer under the catalyst layer preparation process are determined. Based on existing characterization results, the Sobel operator edge recognition algorithm is used to determine the classification characteristics without existing data. At the same time, based on the relevant parameters of the catalyst layer in actual production needs, the parameters of the catalyst layer numerical reconstruction model are determined.

[0013] S12, the support is generated in the catalyst layer;

[0014] S13, Catalytic active phase particles loaded in the catalyst layer;

[0015] S14, Ionomer Loading.

[0016] The phase material classification characteristics in the catalyst layer include the anisotropic distribution of support clusters, the size distribution of the support, the average thickness of the ionomer, the pore size distribution in the catalyst layer, and the electrochemically effective area in the catalyst layer.

[0017] The parameters of the numerical reconstruction model of the catalyst layer include the length, width, and height of the reconstructed cuboid space of the catalyst layer, the support size, the size of the active phase of the catalyst, the minimum size of the ionomer resin, the volume fraction of the support, the volume fraction of the active phase of the catalyst, the volume fraction of the ionomer resin, and the surface distribution density of the active phase of the catalyst on the support surface.

[0018] The support formation in the catalyst layer includes the following steps:

[0019] S121. Based on the carrier size, divide the space of the catalyst layer into grids, and assign each grid a value of 0, treating it as empty;

[0020] S122. In the generated grid, determine the number of carrier clusters based on the set dispersion degree of the slurry in the catalyst layer, and use this number as the random seed number core1. Randomly assign the value of core1 grids in the generated grid space from 0 to 1, which is considered as full.

[0021] S123. Set the growth thresholds for the face direction, edge direction, and corner direction of the seed neighboring mesh respectively, and randomly preset the mesh assignments for the face direction, edge direction, and corner direction of the seed neighboring mesh. If the neighboring mesh assignment is less than the growth threshold of the corresponding direction, then assign the neighboring mesh a value of 1, which is considered full; otherwise, reset the neighboring mesh to 0, which is considered empty.

[0022] S124. As growth continues, the newly filled grid is regarded as a new seed. The above steps S122-S123 are continuously repeated, and the volume fraction of the grid that has been filled in the reconstructed space is constantly counted until it converges to the preset carrier volume fraction.

[0023] The loading of catalytic active phase particles in the catalyst layer includes the following steps:

[0024] S131. Based on the particle size of the active phase of the catalyst, expand the number of grids in the length, width and height directions within the determined catalyst layer space, and inherit the grids that have been filled by the support to the current grid according to the scaling ratio. At the same time, according to the current grid size, empty grids that are less than or equal to the support size from the grids that have been filled by the support are assigned a value of 1 and are considered as full, thus completing the voxel resolution conversion.

[0025] S132. Assign a value in the range of 0-1 to the neighboring grids of the loaded support grid and perform a search. If the assigned value is less than or equal to the surface distribution density of the catalyst active phase on the support surface, then assign a value of 2 to the grid, that is, it is filled by the active phase; otherwise, reset the grid to 0, and consider it empty.

[0026] S133. After one growth cycle, the growth is stopped. At this point, the loading density of the active phase on the surface of the support reaches the preset surface distribution density.

[0027] The ionomer loading includes the following steps:

[0028] S141. Based on the minimum generated size of the ionomer resin, expand the number of grids in the length, width and height directions within the determined catalyst layer space, and inherit the grids that have been previously filled by the support and catalytic active phase into the current grid according to the scaling ratio. At the same time, according to the current grid size, empty grids that are less than or equal to the size of the support-filled grids are assigned a value of 1 and are considered to be filled by the support. Empty grids that are less than or equal to the radius of the active phase are assigned a value of 2 and are considered to be filled by the active phase particles, thus completing the voxel resolution conversion.

[0029] S142. Assign a value in the range of 0-1 to the adjacent grids of the loaded carrier grid, and search. If the assigned value is less than or equal to the preset first carrier surface coverage rate, then assign a value of 2 to the grid, that is, it is filled with ionomer resin; otherwise, reset the grid to 0, and consider it empty.

[0030] S143. Assign a value in the range of 0-1 to the neighboring grids of the loaded catalytic active phase particle grid, and search. If the assigned value is less than or equal to the preset second carrier surface coverage, then assign a value of 2 to the grid, that is, it is filled by the ionomer resin. The ionomer resin grid generated in this process is regarded as a growth seed.

[0031] S144. Set the growth thresholds for the face direction, edge direction, and corner direction of the seed neighboring grid, and randomly preset the grid values ​​for the face direction, edge direction, and corner direction of the seed neighboring grid. If the neighboring grid value is less than the growth threshold of the corresponding direction, then assign the neighboring grid value to 3, which is considered to be filled by the ionomer resin; otherwise, reset the neighboring grid to 0, which is considered to be empty.

[0032] S145. As growth continues, the newly filled grid is regarded as a new seed. The above steps S142-S144 are continuously repeated, and the volume fraction of the grid filled in the reconstructed space is constantly counted until it converges to the preset ionomer resin volume fraction.

[0033] The validity verification of the numerical reconstruction model of the catalyst layer is specifically carried out by: statistically analyzing the pore size distribution results and the effective electrochemical active area in the numerical reconstruction model of the catalyst layer, and comparing them with the reference results that have been statistically analyzed and experimentally tested in the actual catalyst layer. If the deviation does not exceed the preset percentage, the catalytic layer reconstruction model is considered accurate and passes the validity verification.

[0034] In step S3, the analysis process of catalyst layer utilization and transport characteristics is as follows:

[0035] S31. Catalyst layer utilization analysis:

[0036] Based on the catalytic layer structure simulated by the numerical reconstruction model of the catalytic layer, it is proposed that the upper side of the catalytic layer is connected to the PTL and the lower side is in contact with the proton exchange membrane.

[0037] Using a clustering algorithm, all carriers connected to the first layer of support on the upper side of the catalyst layer are marked as connected carriers, all ionomers connected to the first layer of ionomers on the lower side of the catalyst layer are marked as connected ionomers, and the pores connected to the upper side of the catalyst layer, i.e. the diffusion layer, are marked as connected pores.

[0038] Traverse all catalytic active phases. If a phase is located next to a connecting ionomer, a connecting support, and a connecting pore, mark it as an effective active phase. Count the number of all effective active phases and divide by the total number of active phases to obtain the utilization rate of the active phase in the catalyst layer, i.e., the utilization rate of the catalyst layer.

[0039] S32. Analysis of the transport characteristics of the catalyst layer:

[0040] By statistically analyzing the pore size distribution data in the catalyst layer, and based on the pore size distribution and porosity in the catalyst layer, the effective Knudsen diffusion coefficient for oxygen transport in the catalyst layer is calculated, and the oxygen transport performance of the catalyst layer is determined.

[0041] Step S3 further includes predicting the utilization rate and transport characteristics after the catalyst layer decays, specifically:

[0042] Based on the areal density of the active phase on the support surface in the TEM characterization results after decay, the active sites generated in the initial state are randomly deleted, that is, the corresponding grid is reset to 0, until the areal density of the active phase on the support surface in the numerical reconstruction model of the catalyst layer reaches the areal density after decay. The utilization rate of the active phase and the oxygen transport performance in the current catalyst layer are recalculated to obtain the predicted value of the battery performance after decay.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] (1) This invention solves the problems of inaccurate accuracy and low versatility in traditional catalytic layer reconstruction methods. By using independent unit voxel resolution, each phase component in the catalytic layer is reconstructed according to its unit voxel size. This solves the problem that the previous reconstruction model could not be reconstructed for the smallest size of each phase, which led to model inaccuracy. This further improves the accuracy of the model. In addition, the voxel resolution of each component in the catalytic layer can be adjusted. The method has strong versatility and can be well used in all three-dimensional reconstruction problems of proton exchange membrane base membrane electrode catalytic layers.

[0045] (2) The method of the present invention can reconstruct the catalytic layer by identifying the feedforward nature of the cross-sectional characterization data of the typical catalytic layer without the physical membrane electrode, and screen and guide the membrane electrode preparation process, which greatly reduces the necessary process optimization cost and experimental time cost in the production process.

[0046] (3) The present invention can calculate the utilization rate of noble metal catalyst and oxygen transport diffusion coefficient in membrane electrode catalytic layer, realize the transport performance and catalyst utilization rate prediction and calculation of catalytic layer at different times under initial process and decay conditions, obtain the physicochemical properties and decay characteristics in catalytic layer in a feedforward manner, and select appropriate support size, ionomer loading, pore-forming process and slurry dispersion process according to the preparation target, greatly optimize the catalyst utilization rate and mass transfer characteristics of membrane electrode catalytic layer, thereby reducing cost. Attached Figure Description

[0047] Figure 1 This is a flowchart of the method of the present invention;

[0048] Figure 2 Generate a flowchart for the carrier;

[0049] Figure 3 Flowchart for grid expansion;

[0050] Figure 4 This is a flowchart of the ionomer loading process;

[0051] Figure 5 A flowchart for calibrating the anisotropic growth probability of the carrier and ionomer;

[0052] Figure 6 This addresses the issues of pore size distribution and cluster size distribution deviations in existing reconstruction methods.

[0053] Figure 7 The characteristics of the catalytic layer generated by the existing reconstruction method, the characteristics of the catalytic layer generated after the improved reconstruction method, and the actual characteristics of the catalytic layer are compared.

[0054] Figure 8 This is a schematic diagram illustrating the verification of the reconstruction result of the present invention in one embodiment;

[0055] Figure 9 This is a schematic diagram of the reconstruction result of the present invention in one embodiment. Detailed Implementation

[0056] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0057] In the optimization design of the catalyst layer for proton exchange membrane-based electrodes, the high cost of fabricating membrane electrodes and catalyst layers containing noble metals (such as Pt and Ir) makes direct optimization of membrane electrode performance based on catalyst utilization and mass transfer performance through batch experiments extremely expensive. Furthermore, experimental measurements and physical characterization methods for catalyst utilization and mass transfer performance are limited by technology, failing to accurately reflect the actual conditions within the catalyst layer. This often results in substandard membrane electrode performance in industrial production, where the utilization of noble metals and mass transfer characteristics cannot reach their maximum potential. To address the aforementioned issues, this invention provides a membrane electrode design optimization method based on variable resolution catalytic layer 3D reconstruction. Using a variable resolution 3D random reconstruction method based on a four-parameter algorithm, the optimal values ​​of four key process parameters most critical in traditional catalytic layer preparation schemes—catalytic layer porosity, slurry dispersion, optimal carbon support size, and ionomer loading—are obtained before catalytic layer design and preparation. This maximizes the utilization rate of the noble metal catalyst in the catalytic layer. Simultaneously, mass transfer performance indicators such as gas diffusion resistance are calculated, yielding the catalytic layer design optimization direction based on optimal mass transfer performance and optimal catalyst utilization. This method enables the acquisition of design and optimization directions for proton exchange membrane-based membrane electrode catalytic layers with extremely low computational cost.

[0058] Specifically, such as Figure 1 As shown, this embodiment provides a membrane electrode design optimization method based on three-dimensional reconstruction of a variable resolution catalytic layer, including the following steps:

[0059] S1. Construct a numerical reconstruction model of the catalytic layer based on existing experimental characterization results;

[0060] S2. Verify the effectiveness of the numerical reconstruction model of the catalyst layer. If the verification is successful, proceed to step S3. Otherwise, adjust the growth probability of each phase and return to step S1 to adjust the numerical reconstruction model of the catalyst layer.

[0061] S3. Adjust the parameters of the numerical reconstruction model of the catalyst layer and analyze the catalyst utilization and transport characteristics in the catalyst layer after adjusting the parameters;

[0062] S4. Based on the analysis results of S3, feedforward to determine the electrode parameters to guide the fabrication of the membrane electrode.

[0063] Step S1 includes the following steps:

[0064] S11. Identification and Statistical Analysis of Phase Classification Characteristics in the Catalyst Layer: Based on reported relevant studies, statistical analysis was conducted on the phase classification characteristics in the catalyst layer under a specific catalyst layer preparation process, including the anisotropic distribution of support clusters ζ0, support size distribution L0, average thickness of ionomers θ0, and pore size distribution d in the catalyst layer. i The effective electrochemical area S in the catalyst layer iFor those not explicitly reported in the literature, corresponding characterization results from the literature were sought. Phase separation was performed using the Sobel operator edge detection algorithm. The ratio of the major and minor axes of the circumscribed ellipse of the separated phases was defined as the anisotropic fractal characteristic data, and the average length of the major and minor axes of the circumscribed ellipse was defined as the average cluster length. Simultaneously, based on parameters such as catalyst type, catalyst loading, and membrane electrode area required in actual production, the length, width, and height (L, W, H) of the reconstructed catalytic layer cuboid space, as well as the support dimensions, were determined. Catalyst active phase size χ, minimum size of ionomer resin ψ, carrier volume fraction ε1, catalyst active phase volume fraction ε2, ionomer resin volume fraction ε3, and surface distribution density of catalyst active phase on carrier surface λ.

[0065] Specifically, for the anisotropic distribution data ζ0 of carrier clusters, the first priority is to search for reports of the same system in the literature. If not, based on the cross-sectional cluster morphology characterization results of the same system, image segmentation is first performed using pixel enhancement algorithms and Sobel operator edge recognition algorithms. Then, the ratio of the major axis to the minor axis of the circumscribed ellipse of the clusters in the separated region is defined as the carrier anisotropic classification feature, and the anisotropic distribution law of carrier clusters ζ0 is statistically analyzed. The average length of the major and minor axes of the circumscribed ellipse is defined as the average cluster length, and the carrier size distribution law L0 is statistically analyzed. The carrier size is determined according to actual production needs. The dimensions of the reconstructed space—length L, width W, and height H—are determined by the catalyst active phase size χ, the minimum size of the ionomer resin ψ, the volume fraction of the support ε1, the volume fraction of the catalyst active phase ε2, the volume fraction of the ionomer resin ε3, and the surface distribution density of the catalyst active phase on the support surface λ.

[0066] S12, Generation of support in the catalyst layer:

[0067] like Figure 2 As shown, it includes the following steps:

[0068] S121. Based on the average size of the carrier Within the defined catalyst layer space, the number of grid cells in the length, width, and height directions are N, respectively. L N W N H (The number of grids in the three directions is the maximum integer of the reconstructed space size in the corresponding direction divided by the unit size of the carrier), and each grid is assigned a value of 0, which is considered empty;

[0069] S122. In the generated grid, determine the number of carrier clusters based on the set dispersion degree of the slurry in the catalyst layer, and use this number as the random seed number core1. Randomly assign the value of core1 grids in the generated grid space from 0 to 1, which is considered as full.

[0070] S123. Set the growth thresholds p for the seed neighbor mesh direction, edge direction, and corner direction, respectively. face p edge p point All of these values ​​are between 0 and 1, and the values ​​of the neighboring meshes in the direction of the seed face, the direction of the edge, and the direction of the corner are randomly preset and assigned to v. face v edge v point The values ​​are all between 0 and 1, representing the growth probability. If the value assigned to a neighboring mesh is less than the growth threshold in the corresponding direction, then the neighboring mesh is assigned a value of 1, considered full; otherwise, the neighboring mesh is reset to 0, considered empty. Specifically, the growth threshold p for the seed neighboring mesh in the face direction, edge direction, and corner direction is set. face p edge p point After the reconstructed model is generated, the parameters are set, and an image recognition algorithm is used to statistically analyze the genotypic feature distribution ζ of the carrier clusters after generation. i and average size L i The growth thresholds of adjacent grid cells in different directions are compared with the statistical fractal characteristics of the carrier, including the anisotropic distribution of carrier clusters ζ0 and the carrier size distribution L0. If they do not match, the growth thresholds of adjacent grid cells in different directions are adjusted until the fractal characteristics and size distribution match the preset values. At this point, the growth thresholds in different directions are fixed.

[0071] S124. As growth continues, the newly filled grid is regarded as a new seed. The above steps S122-S123 are continuously repeated, and the volume fraction of the grid that has been filled in the reconstructed space is constantly counted until it converges to the set carrier volume fraction ε1.

[0072] S13, Catalytic active phase particle loading in the catalyst layer:

[0073] S131, Mesh expansion: such as Figure 3 As shown, based on the particle size of the catalyst active phase, the number of grids expanded in the length, width, and height directions within the determined catalyst layer space are M, respectively. L M W M H (The number of grids in the three directions is the largest integer calculated by dividing the reconstruction space size in the corresponding direction by the unit size of the active phase.) Grids previously filled by the carrier are inherited into the current grid based on the scaling ratio. Simultaneously, grids with a distance less than or equal to the carrier-filled grid are searched according to the current grid size. Empty grid cells are all assigned a value of 1, which is considered full, thus completing the voxel resolution conversion;

[0074] S132. Assign a value in the range of 0-1 to the neighboring grids of the loaded support grid and perform a search. If the assigned value is less than or equal to the surface distribution density λ of the catalyst active phase on the support surface, then assign a value of 2 to the grid, that is, it is filled by the active phase; otherwise, reset the grid to 0, and consider it empty.

[0075] S133. After one growth cycle, the growth is stopped. At this point, the loading density of the active phase on the surface of the support reaches the preset surface distribution density λ.

[0076] S14, Ionomer Loading:

[0077] like Figure 4 As shown, it includes the following steps:

[0078] S141. Based on the minimum formation size of the ionomer resin, the number of grids expanded in the length, width, and height directions within the determined catalyst layer space are N respectively. L N W N H (The number of grids in the three directions is the maximum integer of the reconstruction space size divided by the minimum size of the ionomer resin in the corresponding direction), and the grids previously filled by the support and catalytic active phase are inherited into the current grid according to the scaling ratio. At the same time, according to the current grid size, grids that are less than or equal to the support size are searched. Empty grids are all assigned a value of 1, which is considered to be filled by the carrier. Empty grids whose distance from the grids already filled by the active phase is less than or equal to the radius of the active phase are all assigned a value of 2, which is considered to be filled by the active phase particles, thus completing the voxel resolution conversion.

[0079] S142. Assign a value between 0 and 1 to the adjacent grids of the loaded carrier grid, and search. If the assigned value is less than or equal to the preset carrier surface coverage θ1, then assign a value of 2 to the grid, that is, it is filled with ionomer resin; otherwise, reset the grid to 0 and consider it empty.

[0080] S143. Assign a value in the range of 0-1 to the neighboring grids of the loaded catalytic active phase particle grid, and search. If the assigned value is less than or equal to the preset carrier surface coverage θ2, then assign a value of 2 to the grid, that is, it is filled by the ionomer resin. The ionomer resin grid generated in this process is regarded as a growth seed.

[0081] S144. Set the growth threshold p for the seed neighbor mesh face direction, edge direction, and corner direction. face_i p edge_i p point_i All of these values ​​are between 0 and 1, and the values ​​of the neighboring meshes in the direction of the seed face, the direction of the edge, and the direction of the corner are randomly preset and assigned to v. face_i v edge_i v point_iThe values ​​are all between 0 and 1; if the value assigned to a neighboring mesh is less than the growth threshold in the corresponding direction, then the neighboring mesh is assigned a value of 3, which is considered as being filled by the ionomer resin; otherwise, the neighboring mesh is reset to 0, which is considered as empty; among them, the growth thresholds for the seed neighboring mesh in the face direction, edge direction, and corner direction are p. face_i p edge_i p point_i Using image recognition methods, the average size θ of the ionomer film after generation was statistically analyzed. i If the average thickness θ0 of the ionomer film is not matched with the preset statistical value, the growth thresholds of adjacent grids in different directions are adjusted until the average thickness θ0 is reached. i It matches the preset value θ0 (deviation less than 3%).

[0082] S145. As growth continues, the newly filled grid is regarded as a new seed. The above steps S142-S144 are continuously repeated, and the volume fraction of the grid that has been filled in the reconstructed space is constantly counted until it converges to the preset ionomer resin volume fraction ε3.

[0083] In step S2, the validity verification of the numerical reconstruction model of the catalyst layer specifically involves: statistically analyzing the pore size distribution results d in the numerical reconstruction model of the catalyst layer. r (1-500nm), and the pore size distribution results characterized by BET and MIP. i A comparison is made; if the deviation is less than 3%, the reconstructed model is considered accurate in terms of pore characteristics. The number of catalyst sites simultaneously located on continuous proton and electron transport channels is counted, and the effective electrochemically active area (ECSA) within the catalyst layer in the reconstructed model is calculated. r If the effective electrochemical active area ECSA0 in the previously obtained solid catalyst layer is compared with that of the model, and the deviation is not more than 3%, then the reconstructed model is considered to have an accurate understanding of the three-phase point generation mechanism and has passed the validity verification.

[0084] Figure 5 The process of parameter calibration (i.e. validity verification and model adjustment) of the reconstructed model in step S2 is demonstrated.

[0085] In step S3, the adjustable parameters provided by the model include porosity, catalyst support size, ionomer resin volume fraction ε3, and catalyst support surface loaded particle density, wherein porosity = 1 - support volume fraction ε1 - catalyst active phase volume fraction ε2 - ionomer resin volume fraction ε3.

[0086] The analysis process for catalyst layer utilization and transport characteristics is as follows:

[0087] S31. Catalyst layer utilization analysis:

[0088] Based on the catalytic layer structure simulated by the numerical reconstruction model of the catalytic layer, it is proposed that the upper side of the catalytic layer is connected to the PTL and the lower side is in contact with the proton exchange membrane.

[0089] Using a clustering algorithm, all carriers connected to the first layer of support on the upper side of the catalyst layer are marked as connected carriers, all ionomers connected to the first layer of ionomers on the lower side of the catalyst layer are marked as connected ionomers, and the pores connected to the upper side of the catalyst layer, i.e. the diffusion layer, are marked as connected pores.

[0090] Traverse all catalytically active phases. If a phase is simultaneously located next to a connecting ionomer, a connecting support, and a connecting pore, mark it as an effective active phase. Count the number N of all effective active phases. Pt_E And divided by the total number of active phases N Pt_T The utilization rate of the active phase in the catalyst layer is obtained, i.e., the utilization rate of the catalyst layer.

[0091] S32. Analysis of the transport characteristics of the catalyst layer:

[0092] By statistically analyzing the pore size distribution data in the catalyst layer, and based on the pore size distribution and porosity in the catalyst layer, the effective Knudsen diffusion coefficient for oxygen transport in the catalyst layer is calculated, and the oxygen transport performance of the catalyst layer is determined.

[0093] In addition, step S3 also includes prediction of the utilization rate and transport characteristics after the catalyst layer decays, specifically:

[0094] Based on the areal density of the active phase on the support surface in the TEM characterization results after decay, the active sites generated in the initial state are randomly deleted, that is, the corresponding grid is reset to 0, until the areal density of the active phase on the support surface in the numerical reconstruction model of the catalyst layer reaches the areal density after decay. The utilization rate of the active phase and the oxygen transport performance in the current catalyst layer are recalculated to obtain the predicted value of the battery performance after decay.

[0095] The following details how this method works:

[0096] 1) The impact of using the same unit voxel resolution to reconstruct the catalyst layer for different components in the catalyst layer on the reconstruction results.

[0097] Figure 6This indicates that using the same reconstructed voxel resolution for components of different unit sizes in the catalyst layer will lead to deviations in pore size distribution and support cluster size distribution. Take a fuel cell catalyst layer as an example. Since carbon particles are used as the support and Pt particles as the active phase in the fuel cell catalyst layer, their volumes differ by nearly 1000 times. To achieve sufficient accuracy in the model voxel resolution, traditional catalyst layer reconstruction methods directly reconstruct the large-diameter carbon support using the voxel resolution of Pt particles (typically 2 nm). This results in inaccuracies in the catalyst layer support framework, which is mainly composed of support clusters, leading to numerous small pores. This increases the frequency of small pore size distribution, affecting the reconstruction analysis results and causing a lower predicted value for transport performance in the catalyst layer. Simultaneously, this unreasonable voxel resolution results in a large number of support clusters smaller than the actual size of a single support. The appearance of these clusters is completely unreasonable, leading to decreased electron transport path connectivity during catalyst layer analysis and lower catalyst utilization.

[0098] 2) The principle and effect of generating different phases at different reconstructed voxel resolutions to solve the problem.

[0099] The catalyst layer is composed of the support, active phase, binder resin, and pores. The support has a relatively large unit size compared to other catalysts; for example, in the commonly used Pt / C catalyst for fuel cells, the volume of a unit carbon sphere support is 1000 times the volume of the active phase Pt spheres. Figure 7 As shown, in traditional catalytic layer reconstruction methods, the same voxel resolution is used for both the support and the active phase. However, in actual catalytic layer formation, the principle of cluster framework formation within the catalytic layer is that the support preferentially forms clusters, which is largely unrelated to the clustering behavior of the active phase. However, the mechanisms of cluster formation for particles of different sizes are completely different. Smaller particles, due to excessively high packing density during reconstruction, suffer from poor continuity and increased dead zones. This leads to an excessive number of dead pores within the catalytic layer due to unreasonable unit size selection, coupled with low continuity of the electron conduction framework, resulting in low catalyst utilization under the same set parameters. By distinguishing the voxel resolution of different phases, the reconstruction method proposed in this invention incorporates the framework formation structure within the catalytic layer caused by support aggregation behavior, making the generated analytical model closer to actual experimental results. The model's accuracy and its feedforward guidance for experiments are significantly enhanced.

[0100] 3) Basis for successful setting of various parameters in this invention

[0101] In principle, the catalyst layer generation method employed in this invention is a process-based porous model generation algorithm, whose basic idea is consistent with the mechanism of porous material growth. In practice, the model employs rigorous verification methods to ensure the successful setting of each parameter. Specifically, taking fuel cell catalyst layer reconstruction as an example, the model's parameters are adjusted against the actual catalyst layer under the same preparation process (same porosity, ionomer loading, catalyst type, slurry dispersion, etc.) to determine characteristics such as cluster anisotropy, cluster size distribution, ionomer thickness distribution, pore size distribution, and effective electrochemical area, until all characteristics are consistent with reality.

[0102] Figure 8 A schematic diagram is provided to verify the reconstruction results. The experimental characterization results of the catalytic layer structure with consistent membrane electrode parameters are compared with those of the reconstructed structure. The reconstructed results show a high degree of agreement with the experimental structure in terms of pore structure, ionomer thickness, and agglomerate anisotropy. BET testing results show that the most integrable pore size in the catalytic layer is 38 nm, which deviates from the reconstructed result of 41 nm by less than 10%, and the distribution of larger pore sizes is also consistent. AFM testing results show that the average thickness of the ionomers in the catalytic layer is approximately 14 nm, consistent with the reconstructed result of 16 nm, with a deviation of less than 5%. TEM testing results show that the anisotropy of the agglomerate structure in the catalytic layer is distributed in the range of 1-3, and more than 90% of the agglomerates in the reconstructed results are distributed in the range of 1-3, consistent with the characterization results. Figure 9 This is a reconstruction result diagram from one embodiment of the present invention. The results visually compare the two-dimensional and three-dimensional effects of the catalyst layer with the experimental characterization results. The reconstructed catalyst layer macroscopically has the same thickness as the catalyst layer in the experimental characterization results, both being 2 micrometers. Furthermore, the bulk phase resolution within the reconstructed catalyst layer is accurate to the size of a single active phase (4 nm in this embodiment), and it is highly consistent with the characterization results visually. Therefore, according to… Figure 8 , Figure 9 The results of the reconstruction demonstrate that the reconstruction method is effective.

[0103] In summary, the innovative points of this invention are as follows:

[0104] 1) A method for three-dimensional reconstruction of the catalyst layer of a proton exchange membrane base electrode with variable resolution and calculation of catalyst utilization and oxygen transport performance in the catalyst layer are proposed. The algorithm for three-dimensional reconstruction of different phases in the catalyst layer using the same voxel resolution is improved into a process-based reconstruction algorithm for three-dimensional reconstruction of the specific phase unit size. The utilization and oxygen transport performance of catalyst layers prepared by different processes are calculated.

[0105] 2) This method can significantly reduce the cost of membrane electrode fabrication. The model can optimize key fabrication parameters in the process flow without the need for catalyst layer and device fabrication, such as ionomer loading, porosity, catalyst type and support size, and slurry dispersion, thereby obtaining a device with superior performance in one step and greatly reducing the mechanical process optimization cost.

[0106] 3) In this method, the voxel resolution of each component in the catalyst layer can be adjusted. Therefore, the algorithm is highly versatile and, after model verification and parameter tuning, it can be applied to the three-dimensional reconstruction and performance analysis of the catalyst layer of all proton exchange membrane base membrane electrodes.

[0107] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for optimizing membrane electrode design based on three-dimensional reconstruction of a catalytic layer with variable resolution, characterized in that, Includes the following steps: S1. Construct a numerical reconstruction model of the catalytic layer based on existing experimental characterization results; S2. Verify the effectiveness of the numerical reconstruction model of the catalyst layer. If the verification is successful, proceed to step S3. Otherwise, adjust the growth probability of each phase and return to step S1 to adjust the numerical reconstruction model of the catalyst layer. S3. Adjust the parameters of the numerical reconstruction model of the catalyst layer and analyze the catalyst utilization and transport characteristics in the catalyst layer after adjusting the parameters; S4. Based on the analysis results of S3, feedforward to determine the electrode parameters to guide the fabrication of the membrane electrode; Step S1 includes the following steps: S11. Identification and statistics of phase material classification characteristics in the catalyst layer: Based on existing data, the phase material classification characteristics in the catalyst layer under the catalyst layer preparation process are determined. Based on existing characterization results, the Sobel operator edge recognition algorithm is used to determine the classification characteristics without existing data. At the same time, based on the relevant parameters of the catalyst layer in actual production needs, the parameters of the catalyst layer numerical reconstruction model are determined. S12, the support is generated in the catalyst layer; S13, Catalytic active phase particles supported in the catalyst layer; S14, Ionomer loading; The phase material classification characteristics in the catalyst layer include the anisotropic distribution of support clusters, the size distribution of support, the average thickness of ionomers, the pore size distribution in the catalyst layer, and the electrochemically effective area in the catalyst layer. The parameters of the numerical reconstruction model of the catalyst layer include the length, width and height of the reconstructed cuboid space of the catalyst layer, the support size, the size of the catalyst active phase, the minimum size of the ionomer resin, the volume fraction of the support, the volume fraction of the catalyst active phase, the volume fraction of the ionomer resin, and the surface distribution density of the catalyst active phase on the support surface. The support formation in the catalyst layer includes the following steps: S121. Based on the carrier size, divide the space of the catalyst layer into grids, and assign each grid a value of 0, treating it as empty; S122. In the generated grid, determine the number of carrier clusters based on the set dispersion degree of the slurry in the catalyst layer, and use this number as the random seed number core1. Randomly assign the value of core1 grids in the generated grid space from 0 to 1, which is considered as full. S123. Set the growth thresholds for the face direction, edge direction, and corner direction of the seed neighboring mesh respectively, and randomly preset the mesh assignments for the face direction, edge direction, and corner direction of the seed neighboring mesh. If the neighboring mesh assignment is less than the growth threshold of the corresponding direction, then assign the neighboring mesh a value of 1, which is considered full; otherwise, reset the neighboring mesh to 0, which is considered empty. S124. As growth continues, the newly filled grid is regarded as a new seed. The above steps S122-S123 are continuously repeated, and the volume fraction of the grid that has been filled in the reconstructed space is constantly counted until it converges to the preset carrier volume fraction.

2. The membrane electrode design optimization method based on three-dimensional reconstruction of a variable resolution catalytic layer according to claim 1, characterized in that, The loading of catalytic active phase particles in the catalyst layer includes the following steps: S131. Based on the particle size of the active phase of the catalyst, expand the number of grids in the length, width and height directions within the determined catalyst layer space, and inherit the grids that have been filled by the support to the current grid according to the scaling ratio. At the same time, according to the current grid size, empty grids that are less than or equal to the support size from the grids that have been filled by the support are assigned a value of 1 and are considered as full, thus completing the voxel resolution conversion. S132. Assign a value in the range of 0-1 to the neighboring grids of the loaded support grid and perform a search. If the assigned value is less than or equal to the surface distribution density of the catalyst active phase on the support surface, then assign a value of 2 to the grid, that is, it is filled by the active phase; otherwise, reset the grid to 0, and consider it empty. S133. After one growth cycle, the growth is stopped. At this point, the loading density of the active phase on the surface of the support reaches the preset surface distribution density.

3. The membrane electrode design optimization method based on three-dimensional reconstruction of a variable resolution catalytic layer according to claim 1, characterized in that, The ionomer loading includes the following steps: S141. Based on the minimum generated size of the ionomer resin, expand the number of grids in the length, width and height directions within the determined catalyst layer space, and inherit the grids that have been previously filled by the support and catalytic active phase into the current grid according to the scaling ratio. At the same time, according to the current grid size, empty grids that are less than or equal to the size of the support-filled grids are assigned a value of 1 and are considered to be filled by the support. Empty grids that are less than or equal to the radius of the active phase are assigned a value of 2 and are considered to be filled by the active phase particles, thus completing the voxel resolution conversion. S142. Assign a value in the range of 0-1 to the adjacent grids of the loaded carrier grid, and search. If the assigned value is less than or equal to the preset first carrier surface coverage rate, then assign a value of 2 to the grid, that is, it is filled with ionomer resin; otherwise, reset the grid to 0, and consider it empty. S143. Assign a value in the range of 0-1 to the neighboring grids of the loaded catalytic active phase particle grid, and search. If the assigned value is less than or equal to the preset second carrier surface coverage, then assign a value of 2 to the grid, that is, it is filled by the ionomer resin. The ionomer resin grid generated in this process is regarded as a growth seed. S144. Set the growth thresholds for the face direction, edge direction, and corner direction of the seed neighboring grid, and randomly preset the grid values ​​for the face direction, edge direction, and corner direction of the seed neighboring grid. If the neighboring grid value is less than the growth threshold of the corresponding direction, then assign the neighboring grid value to 3, which is considered to be filled by the ionomer resin; otherwise, reset the neighboring grid to 0, which is considered to be empty. S145. As growth continues, the newly filled grid is regarded as a new seed. The above steps S142-S144 are continuously repeated, and the volume fraction of the grid filled in the reconstructed space is constantly counted until it converges to the preset ionomer resin volume fraction.

4. The membrane electrode design optimization method based on three-dimensional reconstruction of a variable resolution catalytic layer according to claim 1, characterized in that, The validity verification of the numerical reconstruction model of the catalyst layer is specifically carried out by: statistically analyzing the pore size distribution results and the effective electrochemical active area in the numerical reconstruction model of the catalyst layer, and comparing them with the reference results that have been statistically analyzed and experimentally tested in the actual catalyst layer. If the deviation does not exceed the preset percentage, the catalytic layer reconstruction model is considered accurate and passes the validity verification.

5. The membrane electrode design optimization method based on three-dimensional reconstruction of a variable resolution catalytic layer according to claim 1, characterized in that, In step S3, the analysis process of catalyst layer utilization and transport characteristics is as follows: S31. Catalyst layer utilization analysis: Based on the catalytic layer structure simulated by the numerical reconstruction model of the catalytic layer, it is proposed that the upper side of the catalytic layer is connected to the PTL and the lower side is in contact with the proton exchange membrane. Using a clustering algorithm, all carriers connected to the first layer of support on the upper side of the catalyst layer are marked as connected carriers, all ionomers connected to the first layer of ionomers on the lower side of the catalyst layer are marked as connected ionomers, and the pores connected to the upper side of the catalyst layer, i.e. the diffusion layer, are marked as connected pores. Traverse all catalytic active phases. If a phase is located next to a connecting ionomer, a connecting support, and a connecting pore, mark it as an effective active phase. Count the number of all effective active phases and divide by the total number of active phases to obtain the utilization rate of the active phase in the catalyst layer, i.e., the utilization rate of the catalyst layer. S32. Analysis of the transport characteristics of the catalyst layer: By statistically analyzing the pore size distribution data in the catalyst layer, and based on the pore size distribution and porosity in the catalyst layer, the effective Knudsen diffusion coefficient for oxygen transport in the catalyst layer is calculated, and the oxygen transport performance of the catalyst layer is determined.

6. The membrane electrode design optimization method based on three-dimensional reconstruction of a variable resolution catalytic layer according to claim 5, characterized in that, Step S3 further includes predicting the utilization rate and transport characteristics after the catalyst layer decays, specifically: Based on the areal density of the active phase on the support surface in the TEM characterization results after decay, the active sites generated in the initial state are randomly deleted, that is, the corresponding grid is reset to 0, until the areal density of the active phase on the support surface in the numerical reconstruction model of the catalyst layer reaches the areal density after decay. The utilization rate of the active phase and the oxygen transport performance in the current catalyst layer are recalculated to obtain the predicted value of the battery performance after decay.

Images