QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: shivani78 on January 27, 2012, 11:04

Title: analysis of stress matrix.
Post by: shivani78 on January 27, 2012, 11:04

dear Sir,
i got this stress matrix while calculating the stress of Carbon nanotube(4,0).Can u please tell me how to analyse it?

+------------------------------------------------------------------------------+
| Stress Report                                                                |
+------------------------------------------------------------------------------+
| [[  1.03412729e-04  -1.17407972e-10  -4.04151233e-16]                        |
|  [ -1.17407962e-10   1.03403618e-04   3.19191328e-16]                        |
|  [ -3.99930790e-16   3.17743407e-16   2.82561654e-02]] eV/Ang**3             |
+------------------------------------------------------------------------------+

Title: Re: analysis of stress matrix.
Post by: zh on January 27, 2012, 12:46

I guess that a supercell is employed to simulate CNT(4,0) in your calculation. The output lists the stress tensor on the supercell.  The periodicity of CNT(4,0) in your calculations is along the z direction, so  the  stress on the cross area of the CNT(4,0) is 2.82561654e-02 eV/Ang**3.

Title: Re: analysis of stress matrix.
Post by: shivani78 on January 27, 2012, 14:21

sir,
i have changed the x coordinates of two bonded carbon atom for applying stress.so can u tell me is this the right way to apply stress or there is some other way to apply stress? ???

Title: Re: analysis of stress matrix.
Post by: zh on January 28, 2012, 16:50

The adjustment of x coordinates of two bonded carbon atom is not the way to achieve the applying stress. You can adjust the length of supercell along the periodicity of CNT to apply the stress.

Title: Re: analysis of stress matrix.
Post by: shivani78 on February 1, 2012, 08:03

Dear Sir,
Can u tell me is it necessary to optimize a cnt after changing its unit cell dimension?i am trying to apply stress by changing the dimension.
And is it possible to apply stress by changing the bond length of all the carbons in a carbon nanotube?

Title: Re: analysis of stress matrix.
Post by: zh on February 1, 2012, 09:54

Of course, it is necessary to optimize the atomic coordinates after changing the unit cell dimension.

The change in the bond lengths among atoms just affects the forces on atoms. It does not affect the stress on cell.  You had better understand the concept of the stress from the textbook.

Title: Re: analysis of stress matrix.
Post by: shivani78 on February 1, 2012, 13:35

sir,
as u said that optimization is required and if i am optimizing my geometrical figure it gets maximum stress value by default as 0.05 then it will optimised to the given value then does it make sense to apply stress before optimizing?

Title: Re: analysis of stress matrix.
Post by: zh on February 2, 2012, 01:22

You misunderstood my suggestion. I said that the optimization of only the atomic coordinates is necessary. When you do the geometry optimization for given a unit cell dimension, the unit cell should be fixed and only the atomic coordinates are allowed to be relaxed.

Title: Re: analysis of stress matrix.
Post by: shivani78 on February 2, 2012, 06:52

sir,
should i drag my file of optimized structure of cnt (4,0) to the builder and then change its unit cell dimension and calculate the stress?will it work?
This will optimize  my crude nanotube atoms and will make it available for further use.

Title: Re: analysis of stress matrix.
Post by: zh on February 2, 2012, 11:23

The procedure may be listed as below:
1) Use the CNT builder in VNL to build the atomic structure of (4,0) CNT with the default C-C bond length.
2) Optimize the atomic coordinates with fixing the lattice constants of the supercell for (4,0) CNT, and then calculate the stress;
3) Change the lattice constant of the supercell along the periodicity of (4,0) CNT; 
4) optimize the atomic coordinates of (4,0) CNT with the adjusted lattice constant, and then calculate the stress;
5) repeat the 3) and 4) steps for another new value of lattice constant.
.....

The adjustment of lattice constants of the supercell for (4,0) CNT can be done by the Builder of VNL or by manually changing the values of lattice constants in the script file.

Title: Re: analysis of stress matrix.
Post by: shivani78 on February 2, 2012, 11:41

sir,
may i ask what do u actually mean by optimizing atomic coordinates?we get only a option of geometry optimization.

Title: Re: analysis of stress matrix.
Post by: zh on February 3, 2012, 02:39

The degree of freedom to be optimized in "Geometry optimization" includes the atomic coordinates,  unit cell shape and volume (or the lattice constants of unit cell,  or stress on cell).
In "Optimization of atomic coordinates", the degree of freedom to be optimized includes only the atomic coordinates.

Title: Re: analysis of stress matrix.
Post by: shivani78 on February 3, 2012, 07:23

dear sir,
i am using k points as 12*2*2 and mesh cutoff 280ev and i am not taking repititions in my cnt.am i doing correct?
Sir,wat should i do to optimize atomic coordinates in ATK?

Title: Re: analysis of stress matrix.
Post by: zh on February 6, 2012, 02:47

The optimization of atomic coordinates can be done by calling "OptimizeGeometry()" with "disable_stress=True", e.g.,

Code

new_configuration = OptimizeGeometry(
        bulk_configuration,
        maximum_forces=0.05*eV/Ang,
        disable_stress=True,
        trajectory_filename=None,
        )

If the CNT in your supercell is built along the x direction, your setup for k-point (12*2*2) may be OK. But if it is not, e.g., the periodic length of CNT along the z direction, you should change the setup for k-point into   2*2*12.