High-entropy alloy electrocatalyst screening method based on machine learning and preparation method of electrocatalyst with grain boundary segregation microstructure

Abstract

The invention discloses a high-entropy alloy electrocatalyst screening method based on machine learning and a preparation method of an electrocatalyst with a grain boundary segregation microstructure, and aims to solve the problems that an existing electrocatalytic oxygen evolution / full water electrolysis electrode is poor in conductivity and difficult to practically apply and stably work. The high-entropy alloy electrocatalyst screening method comprises the following steps: 1, collecting element property data of a high-entropy alloy electrocatalyst; 2, constructing a data set by taking the high-entropy alloy electrocatalyst element property data as an input value and taking the conductivity as an output value; 3, training a random forest RF model through the data set; (4) taking the FeaNibMocCodCueAlf high-entropy alloy as a screening target; and 5, predicting the conductivity of the screening target through the trained RF model. According to the method, a machine learning method is utilized, high-entropy alloy components containing specific elements are rapidly screened, and the copper element is introduced to form grain boundary segregation, so that the activity of the nickel element is improved, and meanwhile, the conductivity of the alloy is improved.

Description

Technical Field

[0001] The present invention relates to a screening method for high-entropy alloy electrocatalytic materials based on machine learning, and a high-entropy alloy fiber electrocatalyst with a grain boundary segregation microstructure is prepared. Background Art

[0002] The large-scale development and utilization of hydrogen energy is imminent. As an important link to improve the hydrogen energy production rate, water electrolysis catalysis has also become a research hotspot in the current new energy field. In terms of the reaction mechanisms of hydrogen evolution and oxygen evolution, the reaction process of hydrogen evolution catalysis is relatively simple and has been basically clarified. However, for the oxygen evolution catalytic reaction, there are many reaction steps, complex intermediate products, and a high reaction barrier, and no clear explanation has been obtained for the mechanism yet.

[0003] As a new material system, high-entropy alloys have shown good development prospects. High-entropy alloys have four typical effects. Among them, the cocktail effect means that different elements in the alloy can play more excellent performance characteristics through synergistic effects. Currently, the transition metal elements most commonly used in industry are Fe and Ni. However, in practical applications, there are also problems such as poor performance and poor stability under high industrial current densities. Summary of the Invention

[0004] The present invention aims to solve the problem that the current electrocatalytic oxygen evolution / water electrolysis electrodes have poor conductivity at industrial current densities and are difficult to work stably in practical applications, and proposes a screening method for high-entropy alloy electrocatalysts based on machine learning and a preparation method for high-conductivity high-entropy alloy electrocatalysts.

[0005] The screening method for high-entropy alloy electrocatalysts based on machine learning of the present invention is implemented according to the following steps:

[0006] 1. Collect element property data of high-entropy alloy electrocatalysts;

[0007] 2. Using the element conductivity, atomic radius, electronegativity, first ionization energy, first affinity, and covalent radius in the element property data of high-entropy alloy electrocatalysts as input values, and the conductivity of high-entropy alloy electrocatalysts as the output value, construct a data set, and perform normalization processing on the data set to obtain a preprocessed data set;

[0008] 3. Train a random forest RF model with the preprocessed data set to obtain a trained RF model;

[0009] 4. Fe composed of Fe, Co, Ni, Mo, Al, and Cu elements a Ni b Mo c Co d Cu e Al fHigh-entropy alloys are used as screening targets;

[0010] V. Predict the conductivity of the screening target through the trained RF model to obtain the atomic percentage content (range) of each element in the high-entropy alloy under high conductivity, thereby completing the screening method for high-entropy alloy electrocatalysts. a Ni b Mo c Co d Cu e Al f The preparation method of the electrocatalyst with a grain boundary segregation microstructure of the present invention is realized according to the following steps:

[0011] I. Predict the conductivity of the high-entropy alloy of Fe

[0012] Ni a Mo b Co c Cu d Al e through the trained RF model to optimize the atomic percentage content of each element in the high-entropy alloy of Fe f Ni a Mo b Co c Cu d Al e in the high-entropy alloy, and then weigh each elemental metal raw material according to the optimized atomic percentage content and mix them evenly to obtain a mixed metal raw material; f II. Use a high-vacuum arc melting furnace to melt the mixed metal raw material into a metal ingot, and then melt and suction-cast the metal ingot into a rod-shaped master alloy;

[0013] III. Evacuate the chamber of the melt spinning quenching equipment through a mechanical pump and a molecular pump, introduce a protective gas, start the rotation of the copper wheel, turn on the power supply of the induction coil to heat and melt the rod-shaped master alloy, and the molten metal liquid sprays onto the rotating copper wheel and is quickly cooled into fibrous shape to obtain a high-entropy alloy electrocatalyst with a grain boundary segregation microstructure.

[0014] The application of the electrocatalyst with a grain boundary segregation microstructure of the present invention is to braid the high-entropy alloy electrocatalyst with a copper grain boundary segregation microstructure and use it as the cathode and anode of an anion exchange membrane electrolyzer (AEM) used in industry for electrolytic water catalytic reactions.

[0015] In the composition design of the high-entropy alloy of the present invention, through the use of machine learning methods, the composition of the high-entropy alloy is preliminarily screened, greatly improving the alloy composition design efficiency. At the same time, by using a rapid solidification preparation method, a large number of high-quality high-entropy alloy fibers with a stable crystal structure are prepared.

[0016] ​

[0017] The present invention has prepared a high-entropy alloy with a special copper grain boundary segregation structure, which exhibits superior performance to commercial noble metal-based electrodes under different working conditions in actual industrial water electrolysis applications. Due to the large number of constituent elements and complex composition of the high-entropy alloy, a large amount of time is required for component screening at the front end of composition design. Through machine learning methods, the composition ratios of high-entropy alloys containing fixed elements are rapidly screened; structure regulation is an effective strategy to improve the catalytic performance of high-entropy alloys. To fully utilize the advantages of the alloy bulk material, the in-situ generated compact structure can significantly improve performance and stability. The present invention also introduces copper elements with a relatively negative mixing enthalpy with active elements, forming smaller precipitation phases on the high-entropy alloy matrix phase. At the same time, copper elements form grain boundary segregation, enriching with active element nickel at the grain boundary, regulating the adsorption energy of nickel, accelerating the electron transfer rate, and the segregation of copper elements improves the electrical conductivity of the alloy. As a result, the catalyst has good activity and stability at a large current density. At the same time, stability tests are carried out using an AEM electrolytic cell commonly used in industrial alkaline water electrolysis. The high-entropy alloy with this special structure enables the catalyst to well adapt to different industrial working conditions and has good stability. At the same time, due to its good mechanical properties, it can also be used as a self-supporting electrode.

[0018] The method for screening high-entropy alloy electrocatalysts based on machine learning and the preparation method of electrocatalysts with a grain boundary segregation microstructure in the present invention include the following beneficial effects:

[0019] 1. By using machine learning methods to rapidly screen the compositions of high-entropy alloys containing specific elements, the efficiency of high-entropy alloy composition design is improved.

[0020] 2. By utilizing the cocktail effect of high-entropy alloys and further improving the catalytic performance of electrocatalytic active elements through doping methods, a special high-entropy alloy fiber with precipitation phases is successfully prepared by introducing copper elements to form grain boundary segregation. The segregation of copper and nickel elements at the grain boundary improves the activity of nickel elements while enhancing the electrical conductivity of the alloy. The in-situ generated composite structure is tightly combined and has good self-supporting ability.

[0021] 3. The high-entropy alloy copper grain boundary segregation structure fiber of the present invention has good oxygen evolution catalytic performance and overall water splitting performance. In terms of oxygen evolution catalysis, its overpotential at 10 mA / cm 2 is 238 mV, and the overpotential at 3000 mA / cm 2 is only 497 mV. In an AEM electrolytic cell, a current density of 500 mA / cm 2 can be achieved when the cell voltage is 1.90 V, which is suitable for industrial use.

[0022] 4. The electrocatalyst with the grain boundary segregation microstructure of the present invention has good electrochemical stability and can stably operate for more than 40 hours at an ultra-high current density of 5000 mA / cm 2 ; when used as both the cathode and anode in an AEM electrolyzer, it can stably operate for 100 hours at a current density of 500 mA / cm 2 .

[0023] 5. The preparation cost of the high-entropy alloy fiber is low, and it can replace the noble metal-based catalyst used in industry in terms of performance, and the simulation test of the industrial electrolyzer has been completed.

[0024] 6. By designing the structure of the high-entropy alloy and regulating the structure of the alloy using the preparation method, the high-entropy alloy electrode has better catalytic performance suitable for industrial applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a bar chart of the fitting degree of different machine learning fitting methods obtained in Example 1;

[0026] Figure 2 It is a screening diagram of the high-entropy alloy composition using the machine learning method in Example 1;

[0027] Figure 3 It is a transmission diagram of the high-entropy alloy fiber Fe 20 Ni 20 Mo 20 Co 20 Cu 20 with a duplex grain boundary segregation structure;

[0028] Figure 4 It is an oxygen evolution performance curve diagram of a series of high-entropy alloys Fe a Ni b Mo c Co d Cu e obtained in Example 1;

[0029] Figure 5 It is a hydrogen evolution performance curve diagram of a series of high-entropy alloys Fe a Ni b Mo c Co d Cu e obtained in Example 1;

[0030] Figure 6 It is the high-entropy alloy Fe 20 Ni 20 Mo 20 Co 20 Cu 15Performance curve of the overall hydrolysis of the anion exchange membrane electrolyzer with Al5;

[0031] Figure 7 The high-entropy alloy Fe obtained in Example 3 20 Ni 20 Mo 20 Co 20 Cu 20 Stability curve at an ultra-high current density of 5000 mA / cm 2 Specific implementation mode

[0032] Specific implementation mode 1: The screening method of high-entropy alloy electrocatalyst based on machine learning is implemented according to the following steps:

[0033] 1. Collect the element property data of high-entropy alloy electrocatalysts;

[0034] 2. Construct a data set with the element conductivity, atomic radius, electronegativity, first ionization energy, first affinity, and covalent radius in the element property data of high-entropy alloy electrocatalysts as input values and the conductivity of high-entropy alloy electrocatalysts as the output value, and perform normalization processing on the data set to obtain the preprocessed data set;

[0035] 3. Train the random forest RF model through the preprocessed data set to obtain the trained RF model;

[0036] 4. The Fe a Ni b Mo c Co d Cu e Al f High-entropy alloy composed of is used as the screening target;

[0037] 5. Predict the conductivity of the screening target through the trained RF model to obtain the atomic percentage content (range) of each element in the Fe a Ni b Mo c Co d Cu e Al f High-entropy alloy, thus completing the screening method of high-entropy alloy electrocatalysts.

[0038] In this implementation mode, the conductivity is set as the output value of machine learning, and the physical property parameters affecting the conductivity, such as conductivity, atomic radius, electronegativity, first ionization energy, first affinity, and covalent radius, are used as input values.

[0039] In Step 4 of this embodiment, for the selection of the high-entropy alloy catalyst elements, since Fe, Co, Ni, and Mo are common active elements for water electrolysis catalysis, the Al element is used for dealloying to increase the specific surface area, and the Cu element is used to enhance the conductivity and form a grain boundary segregation structure.

[0040] Embodiment 2: The difference between this embodiment and Embodiment 1 is that in Step 1, the high-entropy alloy electrocatalyst is composed of 5 to 10 elements.

[0041] Embodiment 3: The difference between this embodiment and Embodiment 1 or 2 is that in Step 2, the preprocessed data set is divided into a training set and a test set.

[0042] Embodiment 4: The difference between this embodiment and any one of Embodiments 1 to 3 is that in Step 3, the number of forest trees in the random forest RF model is 96, and the deepest depth of the tree is 4.

[0043] Embodiment 5: The preparation method of the electrocatalyst with a grain boundary segregation microstructure in this embodiment is realized according to the following steps:

[0044] 1. Predict the conductivity of the high-entropy alloy by the trained RF model for Fe a Ni b Mo c Co d Cu e Al f to optimize the atomic percentage content of each element in the high-entropy alloy of Fe a Ni b Mo c Co d Cu e Al f Then weigh each elemental metal raw material according to the optimized atomic percentage content and mix them evenly to obtain a mixed metal raw material;

[0045] 2. Use a high-vacuum arc melting furnace to melt the mixed metal raw material into a metal ingot, and then melt and suction-cast the metal ingot into a rod-shaped master alloy;

[0046] 3. Evacuate the chamber of the melt spinning quenching equipment through a mechanical pump and a molecular pump, introduce a protective gas, start the rotation of the copper wheel, turn on the power supply of the induction coil to heat and melt the rod-shaped master alloy, and the molten metal liquid sprays onto the rotating copper wheel and is quickly cooled into fibers to obtain a high-entropy alloy electrocatalyst with a grain boundary segregation microstructure.

[0047] Embodiment 6: The difference between this embodiment and Embodiment 5 is that the process of training the RF model in Step 1 is as follows:

[0048] a. Collect the elemental property data of the high-entropy alloy electrocatalyst;

[0049] b. Using the element conductivity, atomic radius, electronegativity, first ionization energy, first affinity, and covalent radius in the element property data of the high-entropy alloy electrocatalyst as input values, and the conductivity of the high-entropy alloy electrocatalyst as the output value, construct a data set, and perform normalization processing on the data set to obtain a preprocessed data set;

[0050] c. Train the random forest RF model with the preprocessed data set to obtain a trained RF model.

[0051] Specific Embodiment Seven: The difference between this embodiment and Specific Embodiment Five is to optimize Fe a Ni b Mo c Co d Cu e Al f In the high-entropy alloy, 75 ≤ a + b + c + d ≤ 80, 5 ≤ e ≤ 25, 0 ≤ f ≤ 25.

[0052] Specific Embodiment Eight: The difference between this embodiment and one of Specific Embodiments Five to Seven is that in Step Two, the metal ingot is melted and suction cast into a rod-shaped master alloy with a diameter of 1 cm and a length of 4 cm.

[0053] Specific Embodiment Nine: The difference between this embodiment and one of Specific Embodiments Five to Eight is that in Step Three, the rotation speed of the copper wheel is controlled to be 2000 - 2400 r / min.

[0054] Specific Embodiment Ten: The difference between this embodiment and one of Specific Embodiments Five to Nine is that in Step Three, the high-entropy alloy electrocatalyst with a grain boundary segregation microstructure is fibrous, and the fiber diameter is 20 - 40 μm.

[0055] Example One: The screening method of high-entropy alloy electrocatalyst based on machine learning is implemented according to the following steps:

[0056] 1. Collect the element property data of the high-entropy alloy electrocatalyst;

[0057] 2. Using the element conductivity, atomic radius, electronegativity, first ionization energy, first affinity, and covalent radius in the element property data of the high-entropy alloy electrocatalyst as input values, and the conductivity of the high-entropy alloy electrocatalytic material as the output value, construct a data set, perform normalization processing on the data set to obtain a preprocessed data set, and the data set is randomly divided into a training set (85%) and a test set (15%);

[0058] III. Use the preprocessed dataset to train the SVR_poly (Support Vector Regression), SVR_rbf (Support Vector Regression), MLP (Multi-Layer Perceptron), RF (Random Forest), Bagging (Bootstrap Aggregating), Ads (Autonomous Decentralized System), and XGB model (Gradient Boosting Model) respectively to obtain the trained models;

[0059] IV. The high-entropy alloy composed of elements Fe, Co, Ni, Mo, Al, and Cu is used as the screening target; a Ni b Mo c Co d Cu e Al f

[0060] V. Use the trained models to predict the conductivity of the screening target to obtain the atomic percentage content (range) of each element in the high-entropy alloy of Fe a Ni b Mo c Co d Cu e Al f at high conductivity, thus completing the screening method for high-entropy alloy electrocatalysts.

[0061] In step one of this embodiment, the elemental property data of high-entropy alloy electrocatalysts are collected through existing literature and materials. The property data of common metal elements are shown in Table 1 below. The output values such as Cr 20 Mn 20 Fe 20 Co 20 Ni 20 has a conductivity of 377000 (cm*Ω) -1 、Fe 40 Mn 40 Co 10 Cr 10 has a conductivity of 226000 (cm*Ω) -1 .

[0062] Table 1

[0063]

[0064]

[0065] This embodiment Figure 1 gives the prediction results of the optimized maximum likelihood model. Among them, the RF algorithm with n_estimators = 96 and max_depth = 4 shows the best performance, achieving the highest R 2 value and the lowest RMSE for both the training and test datasets. The R value exceeds 0.85​2 The value indicates a strong correlation between the predicted and experimental conductivity values, confirming the high accuracy of the optimized RF model. Figure 2 The predicted conductivity results of the optimized RF model are shown. To achieve high performance and stability at ultra-high current densities, HEA catalysts must have multiple active sites and maximize conductivity. Based on the predictions of the RF model, Figure 2 the composition range within the black circle in 75-80 Cu 15-25 Al 0-25 ((FeCoNiMo)

[0066] Example 2: The preparation method of the electrocatalyst with a grain boundary segregation microstructure in this example is realized according to the following steps:

[0067] 1. Predict the conductivity of the Fe a Ni b Mo c Co d Cu e high-entropy alloy through the trained RF model, and weigh the raw metal materials of each element to obtain a mixed metal raw material; 20 Ni 20 Mo 20 Co 20 Cu 20 Mix them evenly.

[0068] 2. Use a high-vacuum arc melting furnace to melt the mixed metal raw material 5 times, 2.5 minutes each time, to ensure the uniform mixing of different elements in the alloy, and then melt and suction-cast 50 g of the metal ingot into a rod-shaped master alloy with a diameter of 1 cm and a length of 4 cm.

[0069] 3. Evacuate the chamber of the melt spinning quenching equipment through a mechanical pump and a molecular pump, introduce argon gas, start the rotation of the copper wheel at a speed of 2200 r / min, turn on the power supply of the induction coil to heat and melt the rod-shaped master alloy, and the molten metal liquid sprays onto the rotating copper wheel and quickly cools into fibrous shape to obtain a high-entropy alloy electrocatalyst with a grain boundary segregation microstructure.

[0070] Figure 4 and Figure 5 HEA-Cu1 in 20 Ni 20 Mo 20 Cu 20 Al 20 represents Fe 20 Ni 20 Mo 20 Co 20 Cu 20, HEA-Cu3 represents Fe 20 Ni 20 Mo 20 Co 20 Cu5Al 15 and HEA-Cu4 represents Fe 20 Ni 20 Mo 20 Co 20 Cu 15 Al5.

[0071] Example 3: The difference between this example and Example 2 is that in Step 1, the optimized Fe 20 Ni 20 Mo 20 Co 20 Cu 15 Al5 is weighed for each elemental metal raw material, and the mixed metal raw materials are obtained by mixing evenly.

[0072] The electrochemical workstation model in this example is Shanghai Chenhua CHI760E. The Shanghai Chenhua current amplifier is used for current amplification to explore the catalytic performance of the fiber at high current densities. During the test, the electrolyte is 1.0 mol / L KOH, the reference electrode is a mercury-mercuric oxide electrode, and the counter electrode is a platinum plate electrode. The high-entropy alloy fiber is placed in the platinum plate electrode clip. By placing almost all of the high-entropy alloy electrocatalyst in the electrolyte and the platinum electrode not contacting the electrolyte, after calculating the surface area of the fiber, the electrochemical workstation software is opened. The oxygen evolution reaction is performed with 100 CV cycles at 0 V - 1.2 V, and then linear sweep voltammetry LSV is used to quantify the catalytic performance of the counter electrode. The measured curve is as Figure 4 shown. The abscissa is the reversible hydrogen potential, and the ordinate is the current density. When dealing with the overpotential of the oxygen evolution reaction, the reversible hydrogen potential at the corresponding current density needs to be subtracted by the potential barrier of the oxygen evolution reaction, 1.23 V. After processing, it is found that the overpotential at 3000 mA / cm 2 is only 497 mV, as Figure 4 , Figure 5 shown.

[0073] This example uses an electrochemical workstation and a current amplifier, and uses the chronoamperometry method of the electrochemical workstation to measure the stability of the current density under different industrial conditions.

[0074] The stability test of this example at different current densities is as Figure 7 shown. The abscissa is the test time, and the ordinate is the reversible hydrogen potential at different current densities. Under the fluctuating current density, the working voltage of the electrode only changes slightly. The stability test of this example at an ultra-high current density of 5000 mA / cm 2 is as Figure 6As shown, at an ultra-high current density, the electrode can operate stably for over 40 hours.

[0075] In this embodiment, an AEM electrolyzer is used to test the stability of the catalyst. The fibers are made into a woven body using a weaving tool and placed on a titanium current collector plate so that the electrolyte can come into full contact with the woven body catalyst. The high-entropy alloy grain boundary segregation catalyst is used as the catalyst for both the cathode and the anode. The anode and cathode parts are ion-exchanged with an anion exchange membrane. A peristaltic pump is used to make the electrolyte flow smoothly between the cathode and the anode. The flow rate of the electrolyte is 2.5 mL / min. In Example 3, Fe 20 Ni 20 Mo 20 Co 20 Cu 15 The stability performance test of the Al5 electrode for overall water splitting is as Figure 6 shown. The abscissa is time and the ordinate is the cell voltage. At a current density of 1000 mA / cm 2 , the cell voltage is only 1.90 V, and a current density of 500 mA / cm 2 can be achieved, far exceeding the industrial application standard.

[0076] This embodiment proposes a special method for regulating the microstructure of high-entropy alloys, that is, by regulating the mixing enthalpy of the alloy. By introducing Cu elements with a more negative mixing enthalpy, grain boundary segregation can be formed at the position of the precipitated phase, which can adjust the adsorption energy of the active sites and promote electron transfer. At ultra-high current densities, the conductivity of the catalyst becomes the main factor affecting its performance. The Cu elements segregated at the grain boundaries can provide good conductivity for the catalyst, thereby enabling it to obtain good catalytic activity at high current densities.

[0077] This embodiment proposes using a machine learning method to preliminarily screen the alloy composition, and at the same time using a preparation method based on rapid solidification to generate this structure at one time. The prepared electrocatalyst with a grain boundary segregation microstructure has good self-supporting ability and also exhibits good oxygen evolution and overall water splitting catalytic performance.

Claims

1. A method for screening high entropy alloy electrocatalysts based on machine learning, characterized in that The high entropy alloy electrocatalyst screening method based on machine learning is implemented in the following steps:

1. Collect elemental property data of high entropy alloy electrocatalysts; Second, using the element conductivity, atomic radius, electronegativity, first ionization energy, first affinity and covalent radius in the element property data of the high entropy alloy electrocatalyst as input values, and the conductivity of the high entropy alloy electrocatalyst as output value, construct a data set, normalize the data set, and obtain a preprocessed data set; 3. Train the random forest RF model using the preprocessed data set to obtain a trained RF model; 4. Fe composed of Fe, Co, Ni, Mo, Al and Cu elements a Ni b Mo c Co d Cu e Al f High entropy alloys as screening targets; 5. The conductivity of the screening target is predicted by the trained RF model to obtain the Fe a Ni b Mo c Co d Cu e Al f The atomic percentage of each element in the high entropy alloy is determined, thereby completing the high entropy alloy electrocatalyst screening method.

2. The method for screening high entropy alloy electrocatalysts based on machine learning according to claim 1 is characterized in that In step 1, the high entropy alloy electrocatalyst is composed of 5 to 10 elements.

3. The method for screening high entropy alloy electrocatalysts based on machine learning according to claim 1, characterized in that The preprocessed data set in step 2 is divided into a training set and a test set.

4. The method for screening high entropy alloy electrocatalysts based on machine learning according to claim 1, characterized in that In step 3, the number of forest trees in the random forest RF model is 96, and the maximum depth of the tree is 4.

5. A method for preparing an electrocatalyst having a grain boundary segregation microstructure, characterized in that The preparation method of the electrocatalyst having a grain boundary segregation microstructure is achieved by the following steps:

1. Fe through the trained RF model a Ni b Mo c Co d Cu e Al f Prediction of electrical conductivity of high entropy alloys to optimize Fe a Ni b Mo c Co d Cu e Al f The atomic percentage of each element in the high entropy alloy, then weighing each single metal raw material according to the optimized atomic percentage, and mixing them evenly to obtain a mixed metal raw material; Second, using a high vacuum arc melting furnace to melt the mixed metal raw materials into metal ingots, and then melting the metal ingots and casting them into rod-shaped master alloys; 3. The chamber of the melt spinning equipment is evacuated by mechanical pumps and molecular pumps, protective gas is introduced, the copper wheel is started to rotate, and the induction coil power supply is turned on to heat and melt the rod-shaped master alloy. The molten metal is sprayed onto the rotating copper wheel and quickly cooled into fibers to obtain a high-entropy alloy electrocatalyst with a grain boundary segregation microstructure.

6. The method for preparing an electrocatalyst having a grain boundary segregation microstructure according to claim 5, characterized in that The process of training the RF model in step 1 is as follows: a. Collect elemental property data of high entropy alloy electrocatalysts; b. Using the element conductivity, atomic radius, electronegativity, first ionization energy, first affinity and covalent radius in the element property data of the high entropy alloy electrocatalyst as input values ​​and the conductivity of the high entropy alloy electrocatalyst as output values, constructing a data set, normalizing the data set, and obtaining a preprocessed data set; c. Train the random forest RF model using the preprocessed data set to obtain a trained RF model.

7. The method for preparing an electrocatalyst having a grain boundary segregation microstructure according to claim 5, characterized in that Optimizing Fe a Ni b Mo c Co d Cu e Al f In high entropy alloys, 75≤a+b+c+d≤80, 5≤e≤25, 0≤f≤25.

8. The method for preparing an electrocatalyst having a grain boundary segregation microstructure according to claim 5, characterized in that In step 2, the metal ingot is melted and suction-cast into a rod-shaped master alloy with a diameter of 1 cm and a length of 4 cm.

9. The method for preparing an electrocatalyst having a grain boundary segregation microstructure according to claim 5, characterized in that In step 3, the rotation speed of the copper wheel is controlled to be 2000-2400 r / min.

10. The method for preparing an electrocatalyst having a grain boundary segregation microstructure according to claim 5, characterized in that In step 3, the high entropy alloy electrocatalyst with grain boundary segregation microstructure is fibrous, and the fiber diameter is 20 to 40 μm.

Images