Transcript:
Hello, everyone today. In this video, I will demonstrate how to create random structure for multi-component system or high entropy alloys for demonstration purposes. I will follow this paper and reduce the results. So this is high entropial material, which contains six elements in equimolar composition, and we know it for maximum entropy. It exists for high entropy alloys. There are many methods to generate random structure, a short range order warrant, calling method, slave methods, But for demonstration purposes, I will use Mcsq’s approach, which is implemented in the attacked code, so you can. You can visit this website and read the paper how they have implemented this paper and the mathematics behind it, which exists, which correlates the objective correlation of the SQL cell with the random alloys and try to minimize the objective function in the paper. You can say this gamma is the the point. Correlation function and the implementation of the algorithm and Q over is the objective function, which minimizes the correlation between our cell with the random cell, the rest of the details. You can read by yourself so to get started. We will go first to HTML and go to Mcsql’s approach. This is the man page where you can find the details, so we will need the input file over here. It is given input file. These are the Cartesian coordinates, and this is the letter’s vector and these are the atoms and the corresponding of the material and overhead command line. So we have Monte Carlo generator for special coincide random structure. So these are the details where dash N stands for number of atoms per unit cell dash, 2-3 or dash 4 These are pair triplet and quadruplets around a given atom, which we need to generate the cluster expansion and minimize the the point correlation function so as to mimic the perfect, infinite random alloy so to get started with. I will follow this paper and reproduce their results. The details are given in the supplementary paper. So this is the executable file first. We need the input file as we saw earlier. These are the cartesian copper that you can specify like in wasp format or in this bit. These are the cartesian coordinates or the orthogonal space. The rest three are the lattice vectors, which are the primitive vectors it’s best to use the primitive vector because SQS will explore whole space containing multiples of this one for FCC. We have one atoms with this composition equimolar so we have tried to do this way next. We will run the code to generate the clusters for FCC. We have 12 coordinations as you can see over here for BCC8. The nearest neighbor is zero point at a distance of 0.70 core dump invokes the MCSQ command and generate the cluster. It will read the random structure file, which I have named over here Random structure and when generating, we need to include pair triplets and quadruplets. So let’s go ahead and generate the file for pier. I have included the distance of 1.14 for triplets and quadruple one for good approximation. We should include more shells more. We can view the cluster’s file for the sake of simplicity 2 stand for pairs and the nearest neighbor, and these are the coordinates. The symmetric coordinates, the nearest number how many it detect is 6 12 the next one 12 we have over here 6 12 now to reproduce the SQS cell, we will use their approach so to generate the Mcqs type mcsqs you need to install the code before using it and minus and stand for the number of atoms in their paper In supplementary information, they have provided 108 items, so we will just type, and the code will start the calculation and reproduce produce the SQS cell, so we have to wait from here to get quality results. We need to run the code for many days so we can minimize the objective function as much as possible. The details you can find on the description box. I will provide it later. These are the FCC and SCP Structure. You can view this how they are packed. What are the nearest distance? Primitive cell, the coordination number, which is very trivial so now let’s generate this qsl. So the how many cells it have generated 2010 These are the lattice vector. For Dft calculation, it is best to use EQ equal length because of the sampling in the brilliant space in the K space. So I will just drop this code now. You can write your own script to generate equal length, but for Sakura purposes. I know that in their paper they have used the length of lettuce 300 in the X direction. This is in the y direction three. So this is the cell which comprise of three by three by three. So these are the multiples of the primitive vector over here three times equal now. I will go ahead now. We are forcing the system to accept, forcing the MCQS Mcsqs code or read this SQL cell and generate the atoms within this cell. So there is a command, MCS mcsqs RC and just generate press enter. Now it will read this file now. Over here, you will see that best correlation function and the corresponding best SQS, so for the correlation you can see as we expected. Two stands for the pairs the nearest distance and this third column stand for the point correlation function of the given cell. I will stop the code over here. For quality purposes, we need to run for many days or months to get quality results, but it depends on the problem. So the third column over here stands for the correlation function and fourth one stands for the perfect random alloy, which, by definition of covariance is zero, and the fifth column over here is. This is the difference between third and fourth. So for two, the nearest distance one. How many pairs? So now for three triplets? It includes two shells nearest hills, four quadruplets. We have two again. It shows that FCC is a densely packed system and, Lastly, we have the objective function, which shows, which compares the quality of your sqsl over here. You can see, we have objective function with this value and for the the other generation of M CS QSL, we have much more negative value, its best to use the much more negative value, which will entail that the quality of your squeeze is nice next. It also generate best SQS cell, which again over here three is the Cartesian coordinates. The rest are the lattice vector, and these are the coordinates, so you can go ahead and convert this file to postcard file, and this is also given in my Github page where I have implemented this python code. How to convert this file to postcard WASP file. And that’s it for this tutorial.