How to generate amorphous structures of activated carbon in VNL/ATK

[reallusion]

Dear quantumwise staffs,
I want to know how to generate amorphous structures of activated carbon in VNL/ATK.
I know there are some tutorials on simulating amorphous silicon dioxide and Al2O3 ( http://www.quantumwise.com/publications/tutorials/item/845-generating-amorphous-structures), but i want to get more information on activated carbon. Would you give me some suggestion.
Thank you.

[Julian Schneider]

First of all, such porous carbon structures require considerably large system sizes, which is why you very likely need to use classical potentials to model the system.
You could follow the procedure outlined in the paper
http://homepages.rpiscrews.us/~shiy2/publications/jcp.porous.pdf

Although we do not have the exact potential of this paper implemented, you could use one of our Tersoff or Brenner potentials for Carbon.
Then, you essentially need to create a random arrangement of carbon atoms at the desired density.
You can do this e.g. by taking a diamond cell, repeating it to obtain the desired number of atoms, then increasing the lattice constants to obtain the desired atomic density (while keeping the fractional coordinates constant). As a new plugin in ATK-2015, you can also use the new Amorphous prebuilder plugin after downloading the SCAITools AddOn.

This initial structure has to be annealed at very high temperatures (in the paper they use 21 000 K) in a NVT thermostat e.g. Langevin or NVT Nose Hoover Chain to randomize the atomic arrangement. Such high temperatures require a very small time step e.g. 0.1 fs or smaller.
Then you can basically follow the instructions in the paper, by cooling the system to lower temperatures. In ATK-2015 this can easily be achieved in the NVT Nose Hoover Chain thermostat by choosing an initial temperature (e.g. 21 000 K and a final temperature, e.g. 1000 K). In ATK-2014 you need to run a series of constant temperature simulations, e.g. with the Langevin thermostat, as described in the amorphous systems tutorial.
Choose a sufficiently large number of steps, ideally resulting in several nanoseconds simulation time, to achieve a relatively slow quenching. Remember also to set the log interval parameter to something like 1000 or even larger, otherwise your trajectory file will become huge and the simulation very slow.
The atoms will move around wildly due to the high temperatures, so in order to be able to inspect your structures properly you need to send the final structure of the MD simulation from the Movie Tool to the Builder and wrap all atoms back into the cell via 'Bulk Tools > Wrap' .

[reallusion]

Dear Julian, thank you very much for the detail infomation.

[reallusion]

Thanks for the sugestion on generating amorphous structures of activated carbon in ATK. I still have some questions as follows:

(1) The reference say most carbon atoms in porous carbon reside in graphenelike structure. But i cannot get this structure in ATK with the potentials you mentioned. The generated structure is amorphous structures without any order. How to a pose some constraints on the structures in order to o yield a structure in agreement to that of the experimental one.

(2) Can i use graphite as the initial structure , instead of  diamond cell? Maybe it will yield a graphenelike structure.

(3)  If the answer to the above is yes, how to create a cubic supercell from graphite. The buttons of "Conventional" and "Transform" are grey in VNL when i use graphite to generate cubic supercell using the tutorials  http://www.quantumwise.com/publications/tutorials/item/845-generating-amorphous-structures

First of all, such porous carbon structures require considerably large system sizes, which is why you very likely need to use classical potentials to model the system.
You could follow the procedure outlined in the paper
http://homepages.rpiscrews.us/~shiy2/publications/jcp.porous.pdf

Although we do not have the exact potential of this paper implemented, you could use one of our Tersoff or Brenner potentials for Carbon.
Then, you essentially need to create a random arrangement of carbon atoms at the desired density.
You can do this e.g. by taking a diamond cell, repeating it to obtain the desired number of atoms, then increasing the lattice constants to obtain the desired atomic density (while keeping the fractional coordinates constant). As a new plugin in ATK-2015, you can also use the new Amorphous prebuilder plugin after downloading the SCAITools AddOn.

This initial structure has to be annealed at very high temperatures (in the paper they use 21 000 K) in a NVT thermostat e.g. Langevin or NVT Nose Hoover Chain to randomize the atomic arrangement. Such high temperatures require a very small time step e.g. 0.1 fs or smaller.
Then you can basically follow the instructions in the paper, by cooling the system to lower temperatures. In ATK-2015 this can easily be achieved in the NVT Nose Hoover Chain thermostat by choosing an initial temperature (e.g. 21 000 K and a final temperature, e.g. 1000 K). In ATK-2014 you need to run a series of constant temperature simulations, e.g. with the Langevin thermostat, as described in the amorphous systems tutorial.
Choose a sufficiently large number of steps, ideally resulting in several nanoseconds simulation time, to achieve a relatively slow quenching. Remember also to set the log interval parameter to something like 1000 or even larger, otherwise your trajectory file will become huge and the simulation very slow.
The atoms will move around wildly due to the high temperatures, so in order to be able to inspect your structures properly you need to send the final structure of the MD simulation from the Movie Tool to the Builder and wrap all atoms back into the cell via 'Bulk Tools > Wrap' .

[zh]

(1).  Depending on the introduced constraints on the structures, you have to do a bit coding by yourself.

(2). Of course, you can do it in that way.

(3).  The transformation from a hexagonal cell to a cubic cell could not be achieved arbitrarily.  The simple way: you first transform the hexagonal cell to an orthorhombic one, and then scale (i.e., adjust) the lattice constants to a cubic cell according to the desired atomic density of amorphous structure.

[reallusion]

Dear Support team, thank you for the reply. However, i am actually pretty new at this.

(1) Would you please give me a simple example on how to introduce constraints on the structures.
(2)  How to transform the hexagonal cell to an orthorhombic one in VNL, in case of graphite.

Many thanks

(1).  Depending on the introduced constraints on the structures, you have to do a bit coding by yourself.

(2). Of course, you can do it in that way.

(3).  The transformation from a hexagonal cell to a cubic cell could not be achieved arbitrarily.  The simple way: you first transform the hexagonal cell to an orthorhombic one, and then scale (i.e., adjust) the lattice constants to a cubic cell according to the desired atomic density of amorphous structure.

[Jess Wellendorff]

In answer to 2), please have a look at the tutorial on converting lattice types: http://docs.quantumwise.com/tutorials/rhombo_hex_trigonal.html

[Julian Schneider]

(1) Would you please give me a simple example on how to introduce constraints on the structures.

Currently, you can only use fixed-atoms constraints, i.e. fix an atom to its initial position, or rigid-body constraints, i.e. treat a groups of atoms as rigid body by allowing only translational motion of its center-of-mass. This can be done in the constraints editor in the MolecularDynamics widget. You may also have a look at the basic MD tutorial.

Unfortunately, it is currently not possible to use or implement other types of constraints, such as fixed angles, etc. Support for more general constraints will probably be available from ATK-2016.

[reallusion]

Thanks to Julian and Jess.