A fuel cell gradient cathode catalyst layer and a method for designing and optimizing the same

Abstract

The present application belongs to the field of fuel cell catalyst layer design, and specifically discloses a fuel cell gradient cathode catalyst layer and a design and optimization method thereof. The method obtains local functional requirements of the cathode catalyst layer through regional sensitivity analysis, and guides targeted platinum and ionomer content gradient distribution strategies. Through parameter sensitivity analysis and platinum and ionomer gradient interaction analysis, the key parameters of the distribution strategy, the influence mechanism and the refined value range are determined. Based on the key parameters and their ranges, a data set is generated, a proxy model is constructed, a candidate solution set is obtained, an optimal solution is screened and verified. Based on the method, the gradient cathode catalyst layer is designed, which can reduce the dependence on experience and trial and error, convert the complex component distribution into a low-dimensional parameter optimization problem with clear physical meaning, fully play the synergistic effect of platinum and ionomer distribution, take into account the requirements of proton conduction, oxygen transmission and catalytic reaction, improve the net output power of the fuel cell and improve the reaction uniformity of the cathode catalyst layer.

Description

Technical Field

[0001] This invention relates to the field of fuel cell catalyst layer design, and more specifically, to a gradient cathode catalyst layer for fuel cells and its design and optimization method. Background Technology

[0002] Driven by global energy transition and carbon emission reduction demands, proton exchange membrane fuel cells (PEMFCs) have become promising electrochemical energy conversion devices due to their high energy conversion efficiency, rapid low-temperature start-up, and clean operation. However, the large-scale application of PEMFCs is still limited by factors such as performance, durability, and cost. The core of these issues lies in the cathode catalyst layer of the membrane electrode assembly. The cathode catalyst layer contains a relatively slow-kinetic oxygen reduction reaction and disperses expensive platinum-based catalysts; therefore, its structural design has a significant impact on improving fuel cell performance, extending its lifespan, and reducing its cost.

[0003] As a key component of the cathode catalyst layer, platinum-based catalysts provide active sites for the oxygen reduction reaction, while ionomers construct proton conduction channels. Reducing platinum content is a crucial design direction for cathode catalyst layers to lower fuel cell costs. However, under low platinum conditions, the reduction in active sites exacerbates local oxygen transport resistance and reaction inhomogeneity, limiting fuel cell performance and durability. While increasing ionomer content is beneficial for proton conduction, it also reduces porosity and restricts oxygen transport. Therefore, a rational design of the cathode catalyst layer is necessary to achieve a balance between reactant transport and catalytic reaction, thereby improving overall performance.

[0004] The existing patent with publication number CN115000471A discloses a machine learning-based method for predicting, analyzing, and optimizing fuel cell catalyst layers. It uses a machine learning approach to build a mathematical model, generate a dataset, perform data-driven model preprocessing, and establish the model. Combined with a genetic algorithm, it performs multi-objective and multi-parameter optimization of the catalyst layer, enabling rapid prediction and quantitative sensitivity analysis of catalyst layer structural parameters. This solves the problems of traditional methods, which are difficult to achieve global analysis and multi-objective optimization in catalyst layer structural parameter optimization, and the low accuracy of single-variable methods in problems with strong nonlinear relationships. However, it cannot achieve nonlinear fine design of gradient catalyst layers, coupling mechanism analysis, and interpretable optimization. Another patent, CN118918979A, provides a method for predicting the volume fraction and gradient distribution of ionomers in a cathode catalyst layer. It combines an agglomeration model with a neural network model, trains the method using polarization curves and maximum power density, derives a fast prediction formula for limiting current density and a dimensionless output voltage correlation, and uses a bisection method to iteratively obtain the optimal ionomer volume fraction and gradient distribution. It establishes a correlation between the ionomer gradient and the average volume fraction, solving the problem of low accuracy in predicting ionomer volume fraction and gradient distribution in data-driven models. This achieves highly accurate cathode catalyst layer design. However, it only focuses on predicting the volume fraction and linear gradient of a single ionomer component in the cathode catalyst layer. While combining an agglomeration model with a neural network improves prediction accuracy, it only pursues an increase in maximum power density.

[0005] Existing research has shown that by controlling the spatial distribution of key components such as platinum and ionomers, it is possible to better adapt to the non-uniformity of oxygen, proton, and water distribution within the cathode catalyst layer, thus improving catalyst utilization and reaction uniformity. However, current studies mostly employ relatively simple distributions such as linear or step gradients, relying primarily on empirical judgment or trial-and-error optimization. They lack a deep understanding of the local mechanisms of action within the cathode catalyst layer, making it difficult to adapt to the complex functional requirements of internal nonlinearity. Furthermore, insufficient consideration is given to the interactions and coupling effects between platinum and ionomer distributions, easily leading to the deterioration of other transport processes after optimizing one component. Simultaneously, screening gradient structures through experiments or extensive trial-and-error schemes is typically costly and time-consuming, and it is difficult to reveal the underlying reasons for performance improvement. With the development of electrode fabrication technology, more flexible and refined component distributions are gradually becoming achievable. Therefore, a mechanism-guided, systematic method is urgently needed for the rational and efficient gradient design of the cathode catalyst layer. Summary of the Invention

[0006] To address the aforementioned issues, this invention provides a gradient cathode catalyst layer for fuel cells and its design and optimization method. This method addresses the difficulty in rationally determining the platinum and ionomer gradient design within the cathode catalyst layer and the lack of mechanistic support. It quantifies the impact of local variations in platinum and ionomer content to formulate a targeted gradient distribution strategy based on this functional requirement. To clarify and leverage the synergistic effect of platinum and ionomer distributions, and avoid direct black-box optimization, it first identifies key parameters with clear physical meaning and their influencing mechanisms. Based on the interaction analysis of these two parameters, it further refines the parameter ranges and elucidates their mechanisms. Based on the key parameters and their ranges of the platinum and ionomer gradient determined through mechanistic analysis, it generates a dataset, constructs a surrogate model, obtains a candidate solution set, selects the optimal solution, and verifies it. This transforms the high-dimensional, complex component distribution space into a small number of key parameter optimization problems with clear physical meaning, improving optimization efficiency while retaining the ability to explain the reasons for the final gradient design and the improvement mechanism.

[0007] To achieve the above objectives, the present invention provides a method for designing and optimizing a gradient cathode catalyst layer for a fuel cell, comprising the following steps: S1. Determine the baseline and design objectives for the cathode catalyst layer of the fuel cell; S2. Divide the cathode catalyst layer into regions; S3. By adjusting the local platinum content and ionomer content in each region, regional sensitivity analysis is conducted to obtain the impact of changes in local platinum content and ionomer content on the design objectives. S4. Based on the local functional requirements obtained from regional sensitivity analysis, determine the gradient distribution strategy for platinum content and ionomer content; S5. Perform parameter sensitivity analysis on the platinum and ionomer gradient distribution strategies respectively to determine the key parameters affecting the design objectives and their influence mechanisms. S6. Perform interaction analysis on key parameters of platinum gradient under a preset ionomer distribution, and perform interaction analysis on key parameters of ionomer gradient under a preset platinum distribution. S7. Determine whether there is a coupling effect between platinum and the ionomer gradient; if there is a coupling effect, proceed to S8; otherwise, proceed to S9. S8. Refine the value range of key parameters of platinum and ionomer gradient and determine their influence mechanism; S9. Generate a dataset based on the determined key parameters of the platinum and ionomer gradient and their value ranges; S10. Based on the dataset, construct a surrogate model to characterize the nonlinear mapping relationship between key parameters of platinum and ionomer gradient and design objectives; S11. Obtain a candidate solution set through an optimization algorithm. The candidate solution set is a non-dominated solution set that is not simultaneously superior to other schemes under multiple design objectives. S12. Select the optimal compromise solution from the candidate solution set and verify the design objective to obtain the gradient cathode catalyst layer of the fuel cell.

[0008] Furthermore, in S1, the design objective includes at least one of a performance objective, a durability objective, and a cost objective; the performance objective is net output power, the durability objective is reaction uniformity, and the cost objective is the amount of platinum-based catalyst used.

[0009] Furthermore, in S2, the region division method includes at least one of the following: region division along the thickness direction of the catalyst layer, region division along the flow direction of the reaction gas, and region division along the width direction of the flow channel and the rib.

[0010] Furthermore, in S3, the regional sensitivity analysis includes: increasing or decreasing the local platinum content or ionomer content in the target region while keeping other regional design parameters unchanged, and quantifying the gain or reduction on net output power and reaction uniformity.

[0011] Furthermore, in S4, based on the local functional requirements of different regions for improving net output power and reaction uniformity, the regions where the platinum and ionomer content needs to be increased, decreased, or maintained are determined, and targeted gradient distribution strategies for platinum and ionomer content are formed accordingly.

[0012] Furthermore, in S5, the key parameters of the gradient distribution strategy include at least one of the following: the average content of platinum and ionomer, the degree of content difference between different regions, and the degree of concentration of components in local regions, which are used to characterize the overall level, overall change, and local shape of the gradient distribution, respectively.

[0013] Furthermore, in S6 and S7, if the preset ionomer gradient distribution causes a change in the influence of the key parameters of the platinum gradient on the design target, or if the preset platinum gradient distribution causes a change in the influence of the key parameters of the ionomer gradient on the design target, then it is determined that there is a coupling effect between platinum and the ionomer gradient.

[0014] Furthermore, in S10, the proxy model is a backpropagation neural network model.

[0015] Furthermore, in S11, the optimization algorithm is a non-dominated sorting genetic algorithm, which obtains multiple candidate gradient catalytic layer design schemes. In S12, a multi-attribute decision method is used to select the optimal compromise scheme. The multi-attribute decision method is an approximation ideal solution sorting method, and it is verified whether the net output power and reaction uniformity of the optimal scheme are improved relative to the benchmark.

[0016] On the other hand, the present invention also provides a gradient cathode catalyst layer for a fuel cell, which is obtained by the gradient cathode catalyst layer design and optimization method for fuel cells as described above.

[0017] Compared with the prior art, the present invention has the following beneficial effects: This invention identifies the local functional requirements of platinum and ionomers in different regions of the cathode catalyst layer through region sensitivity analysis. This overcomes the limitations of simple trial-and-error or simplistic linear gradient design, allowing the gradient distribution strategy to be guided by the transport and reaction mechanisms within the catalyst layer. The resulting gradient design not only explains why different regions need to increase, decrease, or maintain the content of corresponding components, but also provides a clear physical basis for subsequent optimization.

[0018] Furthermore, this invention simultaneously considers decoupling and interaction analyses of the platinum gradient and the ionomer gradient. This not only clarifies the independent effects of the two key components on the design objective but also identifies the coupling effects between them, avoiding imbalances in local reactions, proton conduction, or oxygen transport caused by optimizing platinum or ionomer distributions individually. By refining the influence mechanisms and value ranges of key parameters, this invention improves the targeting, reliability, and design efficiency of subsequent optimizations.

[0019] Furthermore, this invention transforms the physical insights gained from regional mechanism analysis into key parameters of the platinum and ionomer gradients with clear meaning. Based on these parameters, a surrogate model is constructed and optimization design is carried out, thereby compressing the complex high-dimensional component distribution problem into an interpretable and controllable low-dimensional optimization problem. Compared with direct black-box optimization, this invention not only obtains gradient catalytic layer schemes that meet the design objectives but also reveals the specific reasons why the scheme improves the design objectives during the design process.

[0020] Furthermore, this invention can fully leverage the specific effects and interactions of platinum and ionomer distribution, taking into account the needs of proton conduction, oxygen transport, and catalytic reaction within the cathode catalyst layer, thereby coordinating ohmic loss and concentration loss and improving the net output power of the battery; and by adjusting the availability of reactants in different regions, it suppresses extreme local reaction areas, improves the overall reaction uniformity of the cathode catalyst layer, and enhances the operational stability and durability of the fuel cell.

[0021] Furthermore, this invention establishes a systematic process from identifying local functional requirements, determining gradient strategies, analyzing coupling effects, to optimization verification, exhibiting good versatility and scalability. This method can formulate corresponding targeted gradient strategies based on different fuel cell structures, operating conditions, manufacturing capabilities, and design goals, providing a generalizable analysis and optimization path for the design of cathode catalyst layers and membrane electrode assemblies in low-platinum fuel cells. Attached Figure Description

[0022] To more intuitively and clearly illustrate the embodiments of this specification, the embodiments will be briefly described below with accompanying drawings. Obviously, the following drawings are only some embodiments of this specification. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0023] Figure 1 This is a schematic diagram of a gradient cathode catalyst layer design and optimization method for a fuel cell according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the region division along the thickness direction of the cathode catalyst layer in an embodiment of the present invention; Figure 3 This invention relates to a regional sensitivity analysis, which adjusts the local platinum loading and carbon separation ratio in each region of the reference cathode catalyst layer and obtains their effects on the battery's net output power and standard deviation of bulk current density. Figure 4 The present invention relates to a linear gradient design of the cathode catalyst layer along the thickness direction and a piecewise gradient distribution strategy for platinum loading and carbon separation ratio determined based on local functional requirements obtained from regional sensitivity analysis. Figure 5 These are parameter sensitivity analysis cases of the platinum loading segmented gradient distribution strategy in embodiments of the present invention; the impact of these cases on net output power and standard deviation of bulk current density; the net power changes caused by these cases are analyzed by activation, ohmic and concentration loss analysis; and the reaction uniformity changes caused by these cases are analyzed by local bulk current density analysis along the center line of the catalyst layer. Figure 6 These are parameter sensitivity analysis cases of the carbon fraction segmented gradient distribution strategy in embodiments of the present invention; the impact of these cases on net output power and standard deviation of bulk current density; the net power changes caused by these cases are analyzed by activation, ohmic and concentration loss analysis; and the reaction uniformity changes caused by these cases are analyzed by local bulk current density analysis along the center line of the catalyst layer. Figure 7 This invention presents a regional sensitivity analysis of platinum content under a preset ionomer distribution. The local platinum loading in each region of the catalyst layer with a segmented gradient carbon ionomer ratio distribution was adjusted and its influence on the battery net output power and the standard deviation of the bulk current density was obtained. Figure 8 In the interaction analysis of key parameters of platinum gradient under preset ionomer distribution and the interaction analysis of key parameters of ionomer gradient under preset platinum distribution of the present invention, the influence of these cases on net output power and standard deviation of bulk current density is analyzed by activation, ohm and concentration loss analysis of the net power change caused by these cases, and by local bulk current density analysis along the center line of the catalyst layer analysis of the reaction uniformity change caused by these cases. Figure 9 The optimal compromise scheme selected in the embodiments of the present invention includes the platinum loading and carbon separation ratio distribution, as well as the net output power, standard deviation of volume current density, activation loss, ohmic loss, concentration loss, and local volume current density along the center line of the catalyst layer of the fuel cell under the uniform, linear, and optimal scheme. Detailed Implementation

[0024] The specific technical solutions in the embodiments of this specification will be analyzed in detail below with reference to the accompanying drawings, providing a more comprehensive description. Obviously, the following description is only a part of the embodiments and not a description of all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0025] It is hereby stated that, unless otherwise defined, all technical and scientific terms used in the embodiments and accompanying drawings of this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. Furthermore, the terms "comprising" and "having," and any variations thereof, used in the embodiments and accompanying drawings of this specification, indicate a non-exclusive inclusion relationship.

[0026] This invention provides a method for designing and optimizing the gradient cathode catalyst layer of a fuel cell, such as... Figure 1 As shown, it includes the following steps: Step 1: Determine the baseline and design objectives for the fuel cell cathode catalyst layer; Step 2: Divide the cathode catalyst layer into regions; Step 3: Perform regional sensitivity analysis by adjusting the local platinum content and ionomer content in each region to obtain the impact of changes in local platinum content and ionomer content on the design objectives; Step 4: Based on the local functional requirements obtained from the regional sensitivity analysis, determine the gradient distribution strategy for platinum content and ionomer content; Step 5: Perform parameter sensitivity analysis on the platinum and ionomer gradient distribution strategies respectively to determine the key parameters affecting the design objectives and their influence mechanisms; Step 6: Perform interaction analysis on key parameters of platinum gradient under a preset ionomer distribution, and perform interaction analysis on key parameters of ionomer gradient under a preset platinum distribution. Step 7: Determine whether there is a coupling effect between platinum and the ionomer gradient; if there is a coupling effect, proceed to step 8; otherwise, proceed to step 9. Step 8: Refine the value range of key parameters of platinum and ionomer gradient and determine their influence mechanism; Step 9: Generate a dataset based on the determined key parameters of the platinum and ionomer gradient and their value ranges; Step 10: Construct a proxy model based on the dataset. The proxy model is used to represent the nonlinear mapping relationship between key parameters and design objectives. Step 11: Obtain a candidate solution set through an optimization algorithm. The candidate solution set is preferably a non-dominated solution set that is not simultaneously superior to other schemes under multiple design objectives. Step 12: Select the optimal compromise solution from the candidate solution set and verify the design objective; wherein, steps 2 to 5 are decoupling gradient analysis process, steps 6 to 8 are synergistic effect analysis process, and steps 9 to 12 are multi-objective optimization process.

[0027] Preferably, the present invention employs Figure 1 The method for designing and optimizing the gradient cathode catalyst layer (CCL) of a fuel cell, combined with three-dimensional multiphase numerical simulation of a proton exchange membrane fuel cell (PEMFC), presents a complete process for designing and optimizing the CCL. To eliminate unnecessary structural and design parameters, the carbon volume fraction in the CCL is fixed at 0.25%, and the CCL thickness is fixed at 10 μm. The platinum loading is varied by adjusting the platinum-carbon mass percentage to represent changes in platinum content, while the ionomer content is varied by adjusting the ionopolymer mass ratio.

[0028] Specifically, step 1 focuses on the three major challenges of PEMFC performance, durability, and cost. To reduce costs, a platinum loading of 0.1 mg / cm³ is used. 2 Using low platinum CCL as a benchmark, and in order to improve performance and durability, the design goals are to increase the net output power of the battery and reduce the standard deviation of the CCL volume current density, which respectively represent improving the overall performance of PEMFC and the uniformity of CCL reaction.

[0029] Figure 2 This is a schematic diagram of the region division along the thickness direction of the cathode catalyst layer in an embodiment of the present invention. In step 2, the cathode catalyst layer is divided into five regions, namely region 1, region 2, region 3, region 4 and region 5, from the proton exchange membrane (PEM) to the microporous layer (MPL). Protons and oxygen enter the CCL from the membrane side region and the microporous layer side region, respectively.

[0030] Perform step 3. Figure 3 This invention relates to a regional sensitivity analysis, which adjusts the local platinum loading and carbon separation ratio in each region of the reference cathode catalyst layer and obtains their effects on the battery's net output power and the standard deviation of the bulk current density. Figure 3Figures (a) and (b) show that increasing the platinum load in any region increases the net power output, but applying any local platinum load perturbation increases the standard deviation of the volume current density. Furthermore, a highly nonlinear relationship is revealed between local platinum load variations and power output; the net power is most sensitive to changes in the PEM-side regions, with region 1 being the most significant, followed by region 2, while the effects in other regions are significantly weakened and almost negligible. Unlike perturbations of the platinum load, Figure 3 Figures (c) and (d) show that adjusting the carbon-to-catalyst ratio only in specific regions can improve performance and even simultaneously improve uniformity. For example, increasing the carbon-to-catalyst ratio only in regions 1 and 2 increases net output power, and increasing the content in region 1 can even simultaneously reduce the standard deviation of the bulk current density, while increasing the carbon-to-catalyst ratio in regions 3, 4, and 5 leads to a decrease in net power. Similarly, decreasing the carbon-to-catalyst ratio only in regions 4 and 5 yields a net power increase with relatively little impact on reaction uniformity, while decreasing it in other regions leads to a decrease in power. These results collectively indicate that conventional linear gradient distributions cannot fully utilize the effects of platinum and ionomers in all regions, thus requiring a more complex and flexible nonlinear gradient strategy, namely, increasing the platinum and ionomer content in the PEM-side regions (especially region 1) while decreasing their content in the MPL-side regions.

[0031] Perform step 4. Figure 4 The present invention relates to a linear gradient design of the cathode catalyst layer along the thickness direction and a piecewise gradient distribution strategy for platinum loading and carbon separation ratio determined based on local functional requirements obtained from region sensitivity analysis. The distribution function is as follows:

[0032] In the formula: P local These are local design parameter values, such as platinum loading and carbon separation ratio; P PEM , P MPL and P mean These represent the boundary values ​​and average values ​​on the PEM side and the MPL side, respectively. y norm For the normalized thickness direction coordinates in CCL; n Represents the power index. Figure 4 The gradient magnitude Δ shown in (a) is P mean and P PEM or P MPL The absolute value of the difference between them represents the overall change in the gradient distribution. For example... Figure 4 As shown in (b), by adjusting nThis piecewise gradient distribution strategy can precisely adjust the local shape of the distribution curve while ensuring that the total integral remains constant and consistent with a uniform and linear distribution. n When the concentration is >1, higher platinum or ionomer contents are distributed in regions 1 and 2 (especially region 1), while the contents near the MPL side (especially region 5) decrease, which is consistent with... Figure 3 Design requirements derived from sensitivity analysis of the mid-region.

[0033] Perform step 5. Figure 5 These are examples of parameter sensitivity analysis of the segmented gradient distribution strategy for platinum loading in embodiments of the present invention; the impact of these examples on net output power and standard deviation of bulk current density; the net power changes caused by these examples analyzed through activation, ohmic, and concentration loss analysis; and the reaction uniformity changes caused by these examples analyzed through local bulk current density analysis along the centerline of the catalyst layer. The segmented gradient distribution strategy parameters corresponding to these examples are shown in Table 1. The average platinum loading of all examples remains the same as the baseline CCL, i.e., the average value of the platinum loading distribution is fixed at 0.1 mg / cm³. 2 The average value of the carbon separation ratio distribution was included in the analysis as a key variable.

[0034] Table 1. Case Study of Parametric Sensitivity Analysis of Platinum Loading Gradient Distribution

[0035] Depend on Figure 5 As shown in (b), the net output power increases continuously with the increase of the power exponent, while the standard deviation of the current density also increases. A similar trend is observed with the increase of the gradient magnitude, indicating that the power exponent and the gradient magnitude are both key parameters for piecewise gradient platinum load distribution and need to be carefully weighed. Figure 5 (c) shows that increasing the power exponent and gradient magnitude can enhance net power by reducing ohmic loss and concentration loss, but the reduction in ohmic loss is more significant, indicating that the platinum gradient distribution mainly improves net output power by enhancing proton conduction. Figure 5 In (d), it is shown that, compared with the baseline, the linear distribution (Case 2) enhances the reaction rate near the PEM / CCL interface while suppressing the reaction rate on the MPL side, thereby worsening the reaction homogeneity. The piecewise gradient platinum distribution further amplifies this trend. This critical homogeneity issue must be addressed in subsequent processes.

[0036] Figure 6These are examples of parameter sensitivity analysis of the segmented gradient distribution strategy for carbon fraction in embodiments of the present invention; the impact of these examples on net output power and standard deviation of bulk current density; the net power changes caused by these examples analyzed through activation, ohmic, and concentration loss analysis; and the reaction uniformity changes caused by these examples analyzed through local bulk current density analysis along the center line of the catalyst layer. The segmented gradient distribution strategy parameters corresponding to these examples are shown in Table 2.

[0037] Table 2. Case Study of Parameter Sensitivity Analysis of Carbon Displacement Gradient Distribution

[0038] Depend on Figure 6 As shown in (b), the gradient amplitude, mean, and power law exponent are all key parameters of the piecewise gradient carbon ratio distribution, and they need to be finely adjusted to achieve an ideal balance between performance and uniformity. Figure 6 (c) shows that ohmic and concentration loss are still the main factors affecting performance changes. However, unlike the platinum gradient, the reduction of concentration loss is the dominant factor for the ionomer gradient, indicating that net power is mainly improved by improving oxygen transport capacity. Figure 6 Figure (d) shows that when the power exponent of the ionomer distribution exceeds 1.0, it leads to a significant "S"-shaped current density distribution, which stems from the coupling effect of proton and oxygen availability. As the power exponent and gradient amplitude increase, this "S"-shaped distribution becomes more pronounced, resulting in a severe deterioration in uniformity. Therefore, the primary challenge for subsequent optimization is to mitigate the adverse effects of this "S"-shaped current distribution caused by the piecewise gradient carbon-to-proton ratio distribution.

[0039] Perform step 6. Figure 7 This invention presents a region-sensitivity analysis of platinum content under a preset ionomer distribution. The local platinum loading in each region of the catalyst layer with a segmented gradient carbon-to-metal ratio distribution was adjusted, and its impact on the battery's net output power and volume current density standard deviation was obtained. The figure shows that the presence of the ionomer gradient significantly alters the effect of local platinum loading on net power and current standard deviation. Specifically, the effect of platinum loading is weakened in region 2, while the effect is amplified in region 4. Furthermore, reducing the platinum content on the MPL side reduces the negative impact on current uniformity, while increasing the platinum loading on the PEM side can even improve uniformity. This indicates that the ionomer gradient significantly alters the optimal platinum distribution, especially requiring a reduction in the power exponent of the segmented gradient platinum loading distribution to adjust the local distribution shape and adapt to changes in local functional requirements under the coupling of platinum and the ionomer gradient.

[0040] To further clarify this interaction, Figure 8The present invention presents the interaction analysis of key parameters of platinum gradient under a preset ionomer distribution (the interaction mainly affects the power exponent of platinum loading distribution, as shown in Table 3) and the interaction analysis of key parameters of ionomer gradient under a preset platinum distribution (the interaction mainly affects the average value of carbon-ion ratio distribution, as shown in Table 4). The impact of these cases on net output power and standard deviation of bulk current density is analyzed by activation, ohmic and concentration loss analysis of the net power change caused by these cases, and by local bulk current density analysis along the center line of the catalyst layer analysis of the reaction uniformity change caused by these cases.

[0041] Table 3. Case studies of interactive analysis of platinum gradient parameters under preset carbon separation ratio distribution.

[0042] Table 4. Case studies of interactive analysis of ionomer gradient parameters under preset platinum loading distribution.

[0043] Perform step 7. Figure 8 Figure (a) shows that, unlike the independent platinum load distribution, both net power and current uniformity initially improve and then deteriorate with increasing power exponent, with the optimal values ​​for both parameters occurring at a power exponent of 1.0 (linear distribution). Figure 8 (b) and Figure 8 As shown in (c), by suppressing ohmic losses and mitigating the "S"-shaped bulk current density distribution caused by the individual ionomer distribution, a piecewise gradient platinum distribution with a power exponent of approximately 1.0 can improve the performance and uniformity defects caused by the independent ionomer distribution. The interaction between platinum and the ionomer gradient also significantly affects the average value of the piecewise gradient ionomer distribution, such as... Figure 8 As shown in (d), as the average carbon separation ratio increases, the net power first increases and then decreases, while the uniformity continues to improve, but the average value for achieving the optimal net power decreases from 0.70 to 0.60. Figure 8 Figure (f) shows that the improvement in volume current density uniformity is due to the fact that increasing the average carbon-to-water ratio alleviates the "S"-shaped current distribution, although this improvement weakens when the carbon-to-water ratio exceeds 0.60. Figure 8 As shown in (e), increasing the average carbon-to-ionomer ratio reduces ohmic loss but exacerbates concentration loss. However, since the gradient platinum distribution has significantly mitigated ohmic loss, the overall requirement for ionomer content is reduced. This indicates a significant coupling effect between platinum and the ionomer gradient. Therefore, in step 8, subsequent optimizations will refine the power-law range of the platinum loading gradient to approximately 1.0, while refining the average range of the carbon-to-ionomer ratio gradient to around 0.60.

[0044] After executing step 9, the key parameters and their value ranges were listed in Table 5, and a dataset containing 243 samples was finally established.

[0045] Step 10 was executed. To avoid overfitting and ensure generalization ability, the dataset was divided into a training set (70%), a validation set (15%), and a test set (15%). A single-hidden-layer backpropagation neural network with 10 neurons was trained and used as a surrogate model. The prediction results for net output power and standard deviation of volume current density on the test set achieved a determination coefficient greater than 0.99 and less than 1 × 10⁻⁶. -4 The mean square error.

[0046] Execute step 11, integrating the surrogate model with the second-generation non-dominated sorting genetic algorithm, setting the initial population size to 100, the maximum number of iterations to 400, and both the function and constraint tolerances to 1×10. -6 The Pareto front was obtained using this genetic algorithm, which aims to improve net output power and reduce the standard deviation of volume current density.

[0047] Table 5. Input parameters and levels for generating the dataset.

[0048] Step 12 is executed, and the optimal compromise solution is selected from the candidate solution set obtained by the genetic algorithm using the approximation ideal solution sorting method. Figure 9 This refers to the platinum loading and carbon separation ratio distribution of the optimal compromise scheme selected in the embodiments of the present invention, as well as the net output power, standard deviation of volume current density, activation loss, ohmic loss, concentration loss, and local volume current density along the center line of the catalyst layer of the fuel cell under the uniform, linear, and optimal scheme. Figure 9 As shown in Figure (a), the specific parameters of this optimal scheme include: the gradient magnitude and power exponent of the platinum loading distribution are 0.0598 mg·cm⁻¹. -2 With a value of 0.941, the average value, power exponent, and gradient magnitude of the ionomer distribution are 0.624, 1.418, and 0.523, respectively, indicating that the power exponent of the ionomer distribution greater than 1 and the power exponent of the platinum distribution slightly less than 1 can achieve the best synergistic effect.

[0049] Figure 9 Figure (b) shows that the optimal design improves net output power by 5.26% compared to the baseline uniform CCL, significantly outperforming the 3.52% gain achieved by linear CCL, while only increasing the current standard deviation by 2.75% at the same voltage. To fairly analyze the impact of these designs, Figure 9(c) compares ohmic, concentration, and activation losses at the same output current density. Compared to the baseline, the linear design reduces both ohmic and concentration losses, while the optimal design employs a more sophisticated strategy: compared to the linear design, this design further reduces the concentration loss from 0.180V to 0.151V, while strategically accepting an increase in ohmic loss from 0.095V to 0.115V. This indicates that the optimal design prioritizes enhancing ionomer gradient-driven oxygen transport capabilities rather than platinum gradient-driven proton conduction capabilities. Figure 9 Figure (d) shows that, at the same current density, the linear design improves uniformity by 4.40% compared to the baseline, while the optimal design further increases this gain to 5.65%. This demonstrates that the interaction between platinum and the ionomer gradient successfully alleviates the defect of the “S”-shaped current distribution caused by the ionomer gradient, thereby achieving a simultaneous improvement in net power and uniformity.

[0050] In summary, the optimal solution fully utilizes the interaction between platinum and the ionomer gradient, not only improving the net output power of the fuel cell but also enhancing reaction uniformity in low-platinum scenarios. These findings fully validate the rationality and effectiveness of the gradient catalyst layer structure obtained through this fuel cell gradient cathode catalyst layer design and optimization method. Based on the above design and optimization, this invention can also provide a fuel cell gradient cathode catalyst layer based on this design and optimization method.

[0051] The preferred embodiments of the present invention have been described above by way of illustration only and are not intended to limit the present invention. Those skilled in the art can make various modifications and variations to the foregoing embodiments. Therefore, any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for designing and optimizing a gradient cathode catalyst layer in a fuel cell, characterized in that, Includes the following steps: S1. Determine the baseline and design objectives for the cathode catalyst layer of the fuel cell; S2. Divide the cathode catalyst layer into regions; S3. By adjusting the local platinum content and ionomer content in each region, regional sensitivity analysis is conducted to obtain the impact of changes in local platinum content and ionomer content on the design objectives. S4. Based on the local functional requirements obtained from regional sensitivity analysis, determine the gradient distribution strategy for platinum content and ionomer content; S5. Perform parameter sensitivity analysis on the platinum and ionomer gradient distribution strategies respectively to determine the key parameters affecting the design objectives and their influence mechanisms. S6. Perform interaction analysis on key parameters of platinum gradient under a preset ionomer distribution, and perform interaction analysis on key parameters of ionomer gradient under a preset platinum distribution. S7. Determine whether there is a coupling effect between platinum and the ionomer gradient; If a coupling effect exists, execute S8; otherwise, execute S9. S8. Refine the value range of key parameters of platinum and ionomer gradient and determine their influence mechanism; S9. Generate a dataset based on the determined key parameters of the platinum and ionomer gradient and their value ranges; S10. Based on the dataset, construct a surrogate model to characterize the nonlinear mapping relationship between key parameters of platinum and ionomer gradient and design objectives; S11. Obtain a candidate solution set through an optimization algorithm. The candidate solution set is a non-dominated solution set that is not simultaneously superior to other schemes under multiple design objectives. S12. Select the optimal compromise solution from the candidate solution set and verify the design objective to obtain the gradient cathode catalyst layer of the fuel cell.

2. The method for designing and optimizing the gradient cathode catalyst layer of a fuel cell according to claim 1, characterized in that, In S1, the design objective includes at least one of a performance objective, a durability objective, and a cost objective; the performance objective is net output power, the durability objective is reaction uniformity, and the cost objective is the amount of platinum-based catalyst used.

3. The method for designing and optimizing the gradient cathode catalyst layer of a fuel cell according to claim 1, characterized in that, In S2, the region division method includes at least one of the following: region division along the thickness direction of the catalyst layer, region division along the flow direction of the reaction gas, and region division along the width direction of the flow channel and the rib.

4. The method for designing and optimizing the gradient cathode catalyst layer of a fuel cell according to claim 1, characterized in that, In S3, the regional sensitivity analysis includes: increasing or decreasing the local platinum content or ionomer content in the target region while keeping other regional design parameters unchanged, and quantifying the gain or reduction on net output power and reaction uniformity.

5. The method for designing and optimizing the gradient cathode catalyst layer of a fuel cell according to claim 4, characterized in that, In S4, based on the local functional requirements of different regions for improving net output power and reaction uniformity, the regions where the platinum and ionomer content needs to be increased, decreased, or maintained are determined, and targeted gradient distribution strategies for platinum and ionomer content are formed accordingly.

6. The method for designing and optimizing the gradient cathode catalyst layer of a fuel cell according to claim 1, characterized in that, In S5, the key parameters of the gradient distribution strategy include at least one of the following: the average content of platinum and ionomer, the degree of content difference between different regions, and the degree of concentration of components in local regions. These parameters are used to characterize the overall level, overall change, and local shape of the gradient distribution, respectively.

7. The method for designing and optimizing the gradient cathode catalyst layer of a fuel cell according to claim 1, characterized in that, In S6 and S7, if the preset ionomer gradient distribution causes a change in the influence of the key parameters of the platinum gradient on the design target, or if the preset platinum gradient distribution causes a change in the influence of the key parameters of the ionomer gradient on the design target, then it is determined that there is a coupling effect between platinum and the ionomer gradient.

8. The method for designing and optimizing the gradient cathode catalyst layer of a fuel cell according to claim 1, characterized in that, In S10, the proxy model is a backpropagation neural network model.

9. The method for designing and optimizing the gradient cathode catalyst layer of a fuel cell according to claim 1, characterized in that, In S11, the optimization algorithm is a non-dominated sorting genetic algorithm, which obtains multiple candidate gradient catalytic layer design schemes. In S12, a multi-attribute decision method is used to select the optimal compromise scheme. The multi-attribute decision method is an approximation ideal solution sorting method, and it is verified whether the net output power and reaction uniformity of the optimal scheme are improved relative to the benchmark.

10. A gradient cathode catalyst layer for a fuel cell, characterized in that, The method for designing and optimizing the gradient cathode catalyst layer of a fuel cell as described in any one of claims 1-9 is adopted.

Images