Theoretical Study On Structure-activity Relationship Of Atomically Dispersed Electrocatalysts For Nitrogen Reduction

Electronic nitrogen reduction reaction（e NRR）is a potential alternative to the Haber-Bosch process,which has high energy consumption,high pollution and high dependence on non-renewable resources such as natural gas.However,the large-scale industrial application of electrochemical nitrogen reduction is still limited by its low reaction rate and low Faraday efficiency（FE）.Therefore,it is very important to search for excellent electrochemical nitrogen reduction catalysts.Atomic dispersion systems,such as single atom catalysts（SACs）and single cluster catalysts（SCCs）,have attracted wide attention due to their advantage,such as high atom utilization.At the same time,a series of theoretical frameworks have been proposed:scaling relationship and active volcano relationship model based on single intermediate adsorption energy（single descriptor）,d-band center theory and reaction dynamics based on maximum energy barrier model.However,the reliability of these theoretical models has been questioned in SACs and SCCs.A new theoretical model is urgently needed to guide the design,modification and screening of e NRR catalysts.The main research contents and achievements of this paper are as follows:1.Build a more accurate volcano relationship model based on combinatorial descriptorsThe catalytic paths of a series of SCCs were calculated using density Functional theory（DFT）,and the adsorption energies of the intermediates were obtained.The results show that the scaling relationship based on single descriptor has bad correlation,and the prediction of volcano relationship model is not accurate enough.The reason is that the adsorption modes of the two stages of the electrochemical nitrogen reduction catalytic process are different:Before the cleavage of nitrogen-nitrogen bond,two nitrogen atoms are adsorbed to active center.Only one nitrogen atom is adsorbed to the active center after the cleavage of nitrogen-nitrogen bond.Therefore,the sum of adsorption energies of intermediates in two stages(ΔG_*NH2+ΔG_*NNH)is taken as a new descriptor（combinatorial descriptor）.The scaling relationship based on combinatorial descriptors has better correlation and the corresponding volcano relationship model has more accurate prediction.The idea of combinatorial descriptors was applied to the data set of adsorption energy of various intermediates in electrochemical nitrogen reduction reactions obtained from references.The results also prove that the combinatorial description’s scaling relationship has better correlation and the volcano relationship model can predict more accurately.And the value corresponding to the vertex of its volcanic model are fixed and not dependent on the catalytic system.Based on the fact that the PDS is the first or last step of proton-electron coupled transfer（PECT）,the fixed vertex of volcano relationship of the combinatorial descriptors is explained.These combinatorial descriptors can better guide the design,modification and screening of catalysts.2.Construct the structure-activity relationship of combinatorial descriptors based on machine learning methodThe analysis of SACs data sets shows that the d-band center theory fails in atomic-scale dispersed catalytic systems,and a new structure-activity relationship is urgently needed.DFT was used to calculate the adsorption structure and electronic structure of*NH₂ and*NNH of a series of SACs,and a small SACs data set was constructed.Based on the data set,deep neural network model,random forest model and convolutional neural network model were constructed to train the functional relationship between the electronic structure characteristics,geometric structure characteristics and element properties of the central metal state density and combinatorial descriptors.Based on the SHapley Additive ex Planation（SHAP）model,the input features of each model’Shapley value were calculated,and the influence of each feature on the predicted value of the combinatorial descriptor was obtained.Based on the model above,the values of SACs combinatorial descriptors can be predicted.According to the shapley value of each characteristic value,a method was proposed to modify the SACs structure to adjust the adsorption capacity.This paper provides a theoretical guidance for the search of ideal electrochemical nitrogen reduction catalysts in a qualitative way.3.Construct a new reaction kinetics model based on energy span theoryThe maximum barrier model can’t explain some experimental phenomena well because it only considers the influence of the potential determination step（PDS）.Based on the SCCs analysis results and energy span theory,a new reaction kinetics model was constructed.The new dynamic model can explain the experimental phenomenon well.Based on this more accurate reaction kinetic model,a method to break through the existing theoretical limitations of catalytic activity is proposed.The results show that,for SCCs system,weakening the adsorption capacity of*NH_x and enhancing the adsorption capacity of*NHNH_x can break through the existing theoretical limit of catalytic activity.This provides a clear direction for the design of electrochemical nitrogen reduction catalysts.Based on this,some specific control schemes for adsorption capacity of intermediates are proposed,some of which have been applied and realized in experiments.