The development of statistical tools based on machine learning(ML)and deep networks is actively sought for materials design problems.While structure-property relationships can be accurately determined using quantum me...The development of statistical tools based on machine learning(ML)and deep networks is actively sought for materials design problems.While structure-property relationships can be accurately determined using quantum mechanical methods,these firstprinciples calculations are computationally demanding,limiting their use in screening a large set of candidate structures.Herein,we use convolutional neural networks to develop a predictive model for the electronic properties of metal halide perovskites(MHPs)that have a billions-range materials design space.We show that a well-designed hierarchical ML approach has a higher fidelity in predicting properties of the MHPs compared to straight-forward methods.展开更多
Alloying has been proposed to circumvent scaling relations between the adsorption energies thus allowing for the complete optimization of multistep reactions.Herein the fidelity of scaling rules on high-entropy alloy(...Alloying has been proposed to circumvent scaling relations between the adsorption energies thus allowing for the complete optimization of multistep reactions.Herein the fidelity of scaling rules on high-entropy alloy(HEA)surfaces is assessed focusing on hydrogen-containing molecules,^(*)AH_(x) for A=C and N(x=0,1,2,3),A=S(x=0,1,2)and A=O(x=0,1).Using an adsorbate-and site-specific deep learning model to rapidly compute the adsorption energies on CoMoFeNiCu HEA surfaces,the energies of ^(*)AH_(x) and ^(*)A are shown to be linearly correlated if ^(*)A and ^(*)AH_(x) have identical adsorption site symmetry.However,a local linear dependence emerges between the configuration-averaged adsorption energies irrespective of the site symmetry.Although these correlations represent a weaker form of the scaling relationships,they are sufficient to prohibit the optimization of multistep reactions.The underpinning of this behavior is twofold(1)the nearsightedness principle and(2)the narrow distribution of the adsorption energies around the mean-field value.While the nearsightedness is general for all electronic systems,the second criterion applies in HEAs with relatively strong reactive elements.The present findings strongly suggest that alloys may not generally enable the breaking of scaling relationships.展开更多
基金W.A.S.acknowledges the financial support from the National Science Foundation(Award No.DMR-1809085)。
文摘The development of statistical tools based on machine learning(ML)and deep networks is actively sought for materials design problems.While structure-property relationships can be accurately determined using quantum mechanical methods,these firstprinciples calculations are computationally demanding,limiting their use in screening a large set of candidate structures.Herein,we use convolutional neural networks to develop a predictive model for the electronic properties of metal halide perovskites(MHPs)that have a billions-range materials design space.We show that a well-designed hierarchical ML approach has a higher fidelity in predicting properties of the MHPs compared to straight-forward methods.
文摘Alloying has been proposed to circumvent scaling relations between the adsorption energies thus allowing for the complete optimization of multistep reactions.Herein the fidelity of scaling rules on high-entropy alloy(HEA)surfaces is assessed focusing on hydrogen-containing molecules,^(*)AH_(x) for A=C and N(x=0,1,2,3),A=S(x=0,1,2)and A=O(x=0,1).Using an adsorbate-and site-specific deep learning model to rapidly compute the adsorption energies on CoMoFeNiCu HEA surfaces,the energies of ^(*)AH_(x) and ^(*)A are shown to be linearly correlated if ^(*)A and ^(*)AH_(x) have identical adsorption site symmetry.However,a local linear dependence emerges between the configuration-averaged adsorption energies irrespective of the site symmetry.Although these correlations represent a weaker form of the scaling relationships,they are sufficient to prohibit the optimization of multistep reactions.The underpinning of this behavior is twofold(1)the nearsightedness principle and(2)the narrow distribution of the adsorption energies around the mean-field value.While the nearsightedness is general for all electronic systems,the second criterion applies in HEAs with relatively strong reactive elements.The present findings strongly suggest that alloys may not generally enable the breaking of scaling relationships.
