Installing and testing new DFTB parameter sets

Table of Contents

Installing DFTB parameters

Getting a DFTB license

The dftb.org website has a number of parameters for the DFTB method which can be used with ATK-SE. In order to download the parameters you will need to fill out their registration form. When your registration has been processed, you will receive a username and password to their website.

Downloading the parameters

On the DFTB download page you can get an overview of the different DFTB parameter sets. The primary ones are named mio, pbc, and matsci; the "mio" set can further be extended by a number of specialized sets listed farther below. The page also lists the papers that must be referenced when using the parameter sets.

Log into dftb.org and download the parameter sets of your interest. The downloaded files will be in a compressed tar ball (.tar.gz files).

Installing the parameters

ATK has a special folder for each of the three DFTB parameters sets. For instance, to install the mio parameter set, unpack the downloaded tar ball and copy the .skf files to the directory

atkpython/share/tightbinding/dftb/mio/

under the directory where ATK is installed. In Linux you can use the following commands (to be issued in the ATK installation directory, where it is also assumed the downloaded files reside)

tar zxvf mio-1-1.tar.gz
cp mio-1-1/* atkpython/share/tightbinding/dftb/mio

Similarly, the pbc and matsci parameter files should be copied to the dftb/pbc and dftb/matsci directories. Note that these directories already exist, but are empty.

[Note] Note

If you are using VNL on a laptop/workstation to set up the calculations, and then a separate installation of ATK on a cluster for running the calculations, then the parameter files must be installed on both computers separately.

Testing the installation

Setting up a bandstructure calculation

To test that the parameters are correctly installed you can perform the following bandstructure calculation for graphene. Open the database tool in VNL from the menu Tools>Database, and type graphene in the search field

Send the structure to the Script Generator using the “Send To” button .

In the Script Generator, add a

  • New Calculator block;

  • Bandstructure analysis block.

  • Change the output filename to graphene.nc

The Script Generator should now have the following settings

Now open the calculator block and

  • select the "ATK-SE: Slater-Koster" calculator,

  • change the k-point sampling to (5,5,1).

  • Go to the "Slater-Koster basis set" and check that the installed basis sets are in the basis set list. Select the "DFTB [mio]" basis set.

    [Note] Note

    If you do not see the "DFTB [mio]" basis set, something went wrong in the installation process. Check that the file atkpython/share/tightbinding/dftb/mio/C-C.skf exists in your ATK installation directory.

  • Uncheck the No SCF iteration box, to perform a self-consistent DFTB calculation.

[Note] Note

The default assumption is that a tight-binding model is non-self-consistent. It is the responsibility of the user to uncheck the No SCF iteration box for models which are self-consistent. Most DFTB models are self-consistent.

Running the test calculation

Transfer the script to the Job Manager using the “send-to”-button and start the calculation.

When the job has finished, locate the file graphene.nc in the file browser in the main VNL window and plot the band structure. You should get the result shown below

If the test went well then the DFTB parameters are properly installed. You will now be able to use DFTB parameters instead of DFT for some of the other tutorials available on the QuantumWise website (including quantum transport calculations), in order to save some time. Obviously this only works if the elements used are covered by the DFTB basis sets.

The next chapters illustrate how to perform spin polarized calculations and calculations of chemical reactions with the DFTB method.