CP2K - Electron Transport based on Non-Equilibrium-Green's-Functions Method
eCSE08-009Key Personnel
PI/Co-Is:Lev Kantorovich, Matt Watkins
Technical: Sergey Chulkov
Relevant Documents
eCSE Technical Report: CP2K - Electron Transport based on Non-Equilibrium-Green's-Functions Method
Project summary
As microelectronic devices become smaller, sensors more optimised and materials design begins to be done in silico, it is very desirable to be able to routinely calculate the current through electrodes separated by some controllable media, such as a molecule or 2D film, like graphene.
The atomistic simulation code CP2K combines a large set of functionalities with computational efficiency, enabling the user to perform a wide range of simulations ranging from molecular dynamics to finding transition states, without the need to use another code or tool. It allows the user to tackle systems with a number of atoms an order of magnitude larger than ordinary plane-wave codes, while still retaining good accuracy.
One area that is currently missing in CP2K is an ability to perform quantum transport simulations. These include three important types of simulations:
- Calculation of the quantum conductance through a molecular junction as well as charge and electrostatic potential distribution across it.
- Calculation of Scanning Tunnelling Microscopy (STM) images
- Calculation of the quantum conductance of a nano-device constructed from 1D or 2D materials with 2D or 3D contacts. These are devices that have recently attracted strong interest from the nano-device community, due to the high electron mobility in materials such as graphene and black phosphorus and their possible application in the next generation of electronic devices.
Implementation of a non-self-consistent steady-state electron transport functionality, being able to perform calculations on devices with up to 3000 atoms.
This objective is complete. We still need to further optimize the code to allow calculations up to the full size listed with high quality basis sets.
Implementation of self-consistent steady-state electron transport, being able to perform calculations on devices with up to 3000 atoms.
This objective is complete, Scaling up to 1000s of atoms is possible, but further optimization is needed to make it routine.
Implementation of a method for modeling extended interface regions between a 2D material and 3D contacts, being able to perform calculations on devices with up to 3000 atoms.
Partially complete. The infrastructure and input structures are there along with a coupling to flexible Poisson solvers. A modest effort is still required to finish this part of the project.
A popular and accurate method to study quantum electron transport from first principles is the Non-Equilibrium Green's Functions (NEGF) method. While system types 1 and 2 are considered to be equivalent in the sense of practical application of the NEGF method, the third type is more complex. In the case of 2D materials with extended planar interface to 3D contacts, the standard NEGF method can become prohibitively expensive due to the need to include large interface regions into the calculation. In the future we will add additional functionality to speed up these types of calculation.
This project will add to CP2K a Density Functional Theory based Non-Equilibrium Green's Functions (NEGF) method for simulating quantum electron transport in nano-scale systems. It allows the use of full set of DFT functional and van der Waals dispersion interaction models currently available to CP2K.
Achievement of objectives
Summary of the software
All work in this project is made available as part of the CP2K software package. The code is in the development version of the code, 5.0, accessible via subversion or git from the main CP2K website. The code is freely available under the GNU GPL licence.
The software is available on ARCHER to all users of the system, the current latest build (as of 28/03/18) is the module cp2k/5.0.17657. Makefiles and libraries are accessible to users to build their own latest version, if required.