Perform Structural Relaxation¶
This tutorial explains how to run a structural relaxation using Density Functional Theory. Variable-cell relaxation consist in simultaneously minimizing the inter-atomic forces, whilst also optimizing the overall lattice geometry by minimizing its corresponding potential energy together with the components of its internal stress tensor.
Accessing the Functionality¶
In the present tutorial, we study the crystalline silicon distorted from its equilibrium cubic-diamond crystal structure and make use of the VASP simulation engine. We will investigate how to optimize the crystal structure geometry and atomic positions in the context of a Total Energy computation. Relaxation prior to a property calculation is generally-speaking a critical precaution to take in order to ensure an accurate final result in the material property being sought.
Generality of tutorial instructions
VASP version considered in this tutorial
The present tutorial is written for VASP at versions 5.3.5 or 5.4.4.
Silicon in its cubic-diamond crystal structure is the default material shown on new job creation, unless this default was changed by the user following account creation. If silicon is still the default choice, it will be automatically loaded at the moment of the opening of Job Designer.
Thereafter, in order to add structural relaxation as an Add-on to the total energy calculation workflow, the user should click the appropriate button within the Header Menu of Workflow Designer. The corresponding "Relaxation" option under this button should thus be chosen.
At the end of the insertion of the relaxation Add-on to the Total Energy Workflow, the user will notice that an additional "Variable-cell Relaxation" Subworkflow has been prepended to the overall computation order flowchart exhibited on the left-hand side of the Workflow Designer Interface.
Examine Unit Input Files¶
The user can now try to open the main "vc-relax" Execution Unit within the "Variable-cell Relaxation" Subworkflow by clicking it. The contents of the input files used for the structural relaxation study can in this way be inspected, towards the bottom of the unit editor interface.
Please note that the second total energy subworkflow reads the structural information output by the preliminary relaxation, instead of the parameters in its own input.
Specific example for VASP
The POSCAR file employed in the ensuing Total Energy subworkflow computation is just a placeholder, and during the course of its execution will be overwritten by a CONTCAR file obtained from the results of the relaxation. This behavior is triggered by the "prepare_restart" post-processor.
Before submitting the Job, the user should click the "Compute" tab of Job Designer and inspect the compute parameters included therein. Silicon is a small structure, so four CPU cores and one minute of calculation runtime should be sufficient.
Once the Job execution is finished, switching to the Results tab of Job Viewer will show the results of the computation, including the final optimized value of the total energy as well as additional information about each execution unit.
Optimized Structural Parameters¶
Finally, the user can also browse the actual output and input files that are part of the calculation under the Files tab of Job Viewer. In order to determine the structure geometry for which relaxation was achieved in the end, the POSCAR file can be downloaded and inspected.
We demonstrate the above-mentioned steps involved in the creation and execution of a structural relaxation study on a Total Energy workflow computation under the following animation, where we make use of the Quantum ESPRESSO simulation engine. The starting point is a crystal structure of silicon which has been slightly distorted from its equilibrium cubic-diamond lattice parameters and atomic positions.
As expected, the components of both the atomic forces and stress tensor shown at the end of the structural relaxation computation, under the interface of Results tab, have low values in proximity to zero, signalling successful relaxation and geometry optimization.