Free to use for registered users. Commercial usage is not allowed.
Please cite qCat if you publish or present results from qCat.
Register with your business email, and send your information to contact email. You will get an email from us after we upgrade your authority.
The qBaandi project was supported by Nano·Material Technology Development Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (Grant. NRF-2016M3A7B4024131).
The computing resource is provied by Computational Science Research Center at KIST
In this page, we will construct several samples with VLatoms in Modeling Lad as a practice.
Pt-based catalysts have been considered the most efficient catalysts for the oxygen reduction reaction (ORR) but exorbitant cost of Pt and designing DOS profiles for efficiency have leaded to alloy catalyst materials research. Moreover, the research has been accelerated with recent result from nanosize effect that offers a chance to find alloy systems that are the immiscible type under ambient conditions in bulk system. Sample Analysis Lab in qCat provides GUI for calculating optimized structure and electronic properties for catalytic materials.
For adsorption energy calculation, we do not have to consider every single atom. Contribution of the atoms covered by others would be much smaller than the surface ones, which are exposed therefore very likely to be adsorption sites. It would be much efficient to deal with only surface atoms.
In order to extract surface atoms on an arbitrary structure, i.e. crystal structures with periodic boundary condition (PBC), nanoparticles in vacuum, even amorphous and wavy 2D structures, we have applied a self-structure analysis using the Voronoi Tessellation in the 3 x 3 x 3 supercell. The periodicity in 3 axes, the maximum internal distance, angles between two neighbor atoms, and distances to the atom over the cell boundary if the neighbor is beyond it were analyzed. Moreover, principal component analysis (PCA) assisted by fast Fourier transformation (FFT) was also applied to detect a local 2-dimensional structure which should be classified as surface atoms.
The surface atom extraction results are visualized in the figure below. Green and red indicate surface atoms and blue represents not surface atoms, i.e. atoms located below the surface.
The current limitation:
Our algorithm has difficulties in detecting large-sized pores of which diameter is ~10 atoms. Some atoms at the innermost layer of multi-walled carbon nanotubes and nanoporous materials could not be detected as surface. (See the figure below of multi-walled tubes.) There was a controversy to classify atoms as surface or not at the development stage, finally we decided to leave it as is. If one has such structures which does not work properly, please request to the developers with the sample structures.
It would be reasonable to assume that the adsorption sites of a given surface can be classified by the local geometry around the considered surface atoms. To screen the unique surface atoms from the all surface atoms, the code makes the list of the chemical species and the bond lengths of the nearest neighbor (NN) atoms from every surface atoms. When we search the NN atoms, we divide NN among surface atoms (NN from surface) and NN among bulk atoms (NN from bulk). As a result, we can obtain the NN information as shown in below example. If two surface atoms have same chemical species and bond lengths are same in descending order of bond lengths, we consider those two atoms have identical local geometries. We confirm that a change of the bond lengths and order can depict the change of the other local geometry features such as angles between bonds, coordination number, etc.
As a result, all surface atoms can be classifies in groups that members of each group have same local geometry. Only one surface atom representing each group is chosen to search the adsorption site. The code selects the representing atom that has highest z-axis position for convenience. The right image of above figure shows an example of the screened 25 unique surface atoms from 328 surface atoms from TiO2 nanoparticle.
Once positions of the unique surface atoms are obtained, the code search three types of possible molecular adsorption site as below:
1) On-top sites : On-top sites are easily defined from the position of unique surface atoms.
2) Bridge sites : From the information of ‘NN from surface’ of each unique surface atoms, the code search all the bridges between the considered surface atom and its NN surface atoms. Then the unique bridges are screened that each screened bridges have different local geometry. The center of a given bridge is used as the bridge site.
3) Hollow sites : Starting from the bridges already screened, the 3-ring and 4-ring connections between NN atoms are searched for each bridges. The center of 3-ring or 4-ring atoms is defined as the hollow site. Again the unique hollow sites are screened that consisting atoms have different local geometry.
The below figures show the example of screened bridges and hollows.
For each adsorption sites, the adsorption vector (direction and length) must be decided. To determine the most perpendicular vector from the various types of surface (nanoparticles, wavy thin film, amorphous, etc), the code calculate the adsorption unit vector as shown in below figure.
For the length of adsorption vector, the code use the sum of covalent radii of the surface atom and edge atom of adsorbent. For each type of sites below criteria is used to roughly consider the different local geometry of on-top, bridge, and hollow sites.
- On-top site : (ra+rb)*0.95
- Bridge site : (ra+rb)*0.75
- Hollow site : (ra+rb)*0.55
, where ra and rb are covalent radii of two adjacent atomns. For orientation of the adsorbed molecule, the code automatically rotate the molecule with respect to the adsorption vector. For ORR, initial orientations of the considered molecules are set to make bond with oxygen. Below figures show the example adsorption structures.
In this page, we will simulate oxide reduction reaction of Pt (111) slab with Modeling, Sample Ananlysis and Activity Lab.
In this page, we will simulate oxide reduction reaction of Pt3Ni (111) slab with Modeling, Sample Ananlysis and Activity Lab. Most of the process is similar with Example 1 so check Example 1; Pt (111) surface.
In this page, we will simulate Pt3Co 1nm particle with Modeling and Stability Lab.
Except molecules and sub-nm particles, materials consists of vast amount of atoms, so it is not able to take every individual atom into account to the calculations. To solve this problem in atomistic modelling, people conventionally assumes that a group of atoms repeats periodically to form a large object, similar to single crystal in crystallography, and this is called the periodic boundary condition (PBC). In PBC, there is a unit cell which is defined by three cell vectors and atoms in it, and the unit cell is repeated in all three directions, just like a single crystal. However, even things that are not actually repetitive are repeated in PBC, such as atomic defect in solids or adsorbed chemicals on solid surfaces. The PBC is a mathematical tools that allows computation of large size of materials, comparing to atomic size, with limited computational resources.
Be careful! You have to consider whether your ‘periodic’ model structure is suitable to describe real material.
At least 10 Angstrom of vacuum between neighbor unit cell is neccessary, 15-20 Angstrom is recommended.
Suppose you are simulating surface of a solid, and the z-axis is the surface normal direction in your atomistic model, you may want to make half-infinite surface model that infinitely thick material and vacuum are facing with each other. However, such structure is impossible because every thing, including empty space, is periodic. Therefore, there is no choice but to make space between atoms in the neighbor unit cells so that interaction between them to be ignorable.
Regardless of checking x, y and z in "Periodic Boundary" in the Modeling Lab, the atomic structure is periodic. But, "Periodic Boundary" affects speed and accuracy of the calculations. If 'z' is un-checked, number of k-point in 'z' direction for structure stabilizer, DOS and optical calculation set to be 1. If it is checked for the direction having vacuum, the calculation results are accurate but it may takes longer time than it should be. If it is un-checked for the direction having no vacum, the results may not be accurate.