| Surface adsorption and interface system are very normal in nature,which exerts a strong influence on the performance of materials.The atoms at the surface or interface are located on the boundary of the materials.Since the chemical environment of them is different with that of the bulk phase,the novel atomic structures of surface adosrption or interface have a strong influence on the physical and chemical properties of materials.There are many physical and chemical reaction processes on the surface,such as corrosion,redox reaction and catalysis.Interface is closely related to some mechanical processes of materials,including stretch and crack.Surface adsorption and interface are always important research fields in material science..It is well known that atomistic surface structure for understanding the physical and chemical properties,as wel as for future design of better photocatalyst is essential.While several experimental techniques(e.g.,low energy electron diffractions,X-ray diffraction and scanning tunneling microscopy)can be used to study the structures of materials,the atomic structures often remain unsolvable due to technical limitations.The alternative of solving structures is theoretical methods by building models manually or using the molecular dynamics simulations in earlier studies.To a great extent,these methods rely on the initial structures,sosometimes can't overcome the high energy barrier of potential energy surface and depend on the chemical experience of researchers.As studies of surface and interface are becoming a hot topic,the global structure optimization method comes to light.In recent years,atomistic structure prediction techniques have been widely developed and applied to resolve many important structural problems.In previous work,our team has developed a swarm-intelligence-based CALYPSO(Crystal structure AnaLYsis by Particle Swarm Optimization)method for structure prediction,and implemented in a CALYPSO software(see http://www.calypso.cn),which has been widely applied to predict predict the structure for 3D crystals,isolated clusters or molecules,surface reconstructions,2D layer materials,and designfunctional materials(e.g.,superhard materials and semiconductors).Here we report a generalized CALYPSO method to predict the structures of surface adsorption and interface.The main results are listed below:1.We have generalized the CALYPSO methodology for surface adsorption structure prediction and coded in the CALYPSO software.The structure of surface adsorption has a 2-dimisional periodicity.In some particular cases the adsorbed atoms are known to stay at specific positions on the substrate.We implemented several structure processing techniques based on these structural characteristics.(1)In the steps of generating and updating structures,the positions of adsorbed atoms are limited at the fixed adsorption site to produce more physically justified structure.(2)The symmetry and minimal interatomic distance constraints are employed to generate the random structures to reduce the dimension of potential energy surface.(3)A set of symmetry functions are adopted to fingerprint the structures in our method.The similarity between two structures can be given by the difference of their atomic symmetry functions.(4)Based on the fixed adsorption site technique,a constrained PSO algorithm is developed for update the structures.These techniques have been tested and proved to improve the efficiency of structure searching.Now this surface adsorption structure prediction method can be applied to the systems whose adsorbate are atoms,moleculers or functional groups,and the systems that atoms adsorbed on single surface or both surfaces.The developed method is benchmarked using two typical systems(hydrogenated and oxidized graphene),successfully reproducing their known stable structures.Besides,the energetically most stable structures are predicted for single-sided hydrogenated graphene with various contents of hydrogen,where band gaps can be tuned by changing the hydrogen content in these stable structures.Furthermore,our method reveals the energetically best structure reported thus far for oxidized graphene.In general,our method is a promising approach for the smart prediction and designing of structures of atoms adsorbed on 2D layered materials.2.We investigated O adsorption on a Zr(0001)surface using our newly developed surface adsorption structure prediction method.A novel structural prototype with a unique combination of surface face-centered cubic(SFCC)and surface hexagonal close-packed(SHCP)O adsorption sites was predicted using a single-layer adsorption model(SLAM)for a 0.5 and 1.0 monolayer(ML)O coverage.First-principles calculations based on the SLAM revealed that the new predicted structures are energetically favorable compared with the well-known SFCC structures for a low O coverage(0.5 and 1.0 ML).Furthermore,on basis of our predicted SFCC+ SHCP structures,a new structure within multi-layer adsorption model(MLAM)was proposed to be more stable at the O coverage of 1.0 ML,in which adsorbed O atoms occupy the SFCC + SHCP sites and the substitutional octahedral sites.The calculated result of trend of work functions with variation in O coverage agrees with the experimental data.Therefore,our proposed structures are optimal candidates for O-adsorbed Zr(0001)surfaces.Our finding provides new structural prototypes for O chemisorption on metal surface systems,especially for hexagonal close-packed metal(e.g.Mg,Ti)surfaces.3.We have developed an efficient method for predicting the interface structures based on the CALYPSO methodology.The interface structures usually have low symmetry,and the bond formed by the atoms at interface is similar with that in the bulk phase.Besides,the lattice mismatch will produce the stress at the interface and make the structure unstable.We have implemented several techniques to deal with these problems.(1)Considering the complexity of lattice match,we develop a toolkit to scan the suitable surpercells with the small lattice mismatch full automatically.(2)We develop a structure generation method based on the constraints of coordination number and interatomic distance.(3)We adopt a constrained 2-D PSO algorithm for updating the structure,and regard the rigid-body displacement as a freedom of structure in the process of structure evolution.The results of tests demonstrate the efficiency of these techniques.We have benchmarked our method using the well-studied systems including graphene armchair-zigzag grain boundary and TiO2 rutile ?3(112)twin boundary.The previous proposed structures of these grain boundaries have been reproduced successfully.We have applied our method to investigate the TiO2 ?5(210)twin boundary.Two new nonstoichiometric structures have been predicted,which can be stable in different chemical environments respectively.In these two structures,the O defect increasing with the decreased O atoms at the interface.Based on the calculation of electronic properties,we found that the band gap can be tuned by the concentration of O defect at interface,providing a potential way for improve the efficiency of rutile.These results demonstrate that our approach is an efficient method for predicting interface structures and provide a new tool for designing the novel interface materials. |
PDF Full Text Request