[DFTB-Plus-User] DFTB+NEGF device geometry definition for non-contiguous atoms

[anz118466 at cse.iitd.ac.in]

Dear Gabriele,

Thank you for the input. It makes sense now. The convergence does take a
lot of cycles. I will simulate with (FirstLayerAtoms =  1) for now. Hope
it works...

I shall also try to implement your scheme on the side. Waiting for your
scripts!

Kind Regards,
Anusha



> On 04/29/2014 07:18 PM, anz118466 at cse.iitd.ac.in wrote:
>> Dear Gabriele,
>>
>> Thank you for such a lucid explanation of PL definition :) It did help
>> clear few doubts of mine.
>>
>> Actually, I did simulate with (FirstLayerAtoms =  1) for my system. The
>> slow convergence, however, made me wonder if I am doing it right. My
>> system has a protein molecule on a GNR and is easily 1000+ atoms big. I
>> think partitioning it would speed up the convergence.
>
> Dear Anusha,
>
> if 'slow convergence' means that you need a lot of SCC cycles,
> partitioning will not change it (the number of SCC steps does not depend
> on partitioning) but a single SCC step will be faster.
>
> How effective the partitioning will be, it depends on the geometry of
> your system. A big elongated molecule on top of the ribbon is not a very
> fortunate case.
>
>>
>> I did not completely understand the script's function to identify blocks
>> for partitioning the device. Kindly elaborate.
>
> Basically you have to parse the coordinates of all the device atoms and
> reorder them in the transport direction in a new .gen file.
> Then you can define a cutoff distance such that the Nth PL index is
> given by the first atomic index such that its coordinate falls within
> (N-1)*cutoff and N*cutoff. I have some scripts I'd like to make public
> but you have to wait a bit.
>
> Gabriele
>
>
>>
>> Best Regards,
>> Anusha
>>
>>
>>
>>> Dear Anusha,
>>>
>>> the construction rule to define a valid PL is the following: there
>>> should be non-zero interactions only between adjacent PLs. Therefore
>>> some kind of 1D partitioning is usually possible for any transport
>>> geometry. Note that the PLs in the device region do not need to have
>>> the
>>> same size.
>>>
>>> The definition of PLs in the contacts is crucial and needs to be
>>> ensured. This point is usually easy because the contacts should be
>>> given
>>> by a semi-infinite bulk structure. Partitioning in the device region is
>>> not mandatory, but it is useful to take advantage of the iterative
>>> scheme, speed up the calculation and decrease the memory usage. In line
>>> of principle, you can define a single PL containing all the device
>>> atoms
>>> (FirstLayerAtoms =  1), and if the system is small enough the code
>>> would
>>> still work. Alternatively you can write a script which orders all the
>>> device atoms along the transport direction and identify the atom
>>> indexes
>>> collecting blocks of atoms within blocks spaced by a reasonable cutoff
>>> range (10 atomic units, for example).
>>>
>>> Note that 'contiguous' does not mean necessarily bonded, or in-plane.
>>> For example, if you have an homogeneous ribbon with an adatom on the
>>> top
>>> and the atoms in the gen file are ordered along the transport
>>> direction,
>>> you will have that the corresponding PL will contain one more atom
>>> (i.e.
>>> the adatom itself)
>>>
>>> Hope this helped, otherwise I can maybe send you an example later.
>>>
>>> Best,
>>> Gabriele
>>>
>>>
>>> On 04/29/2014 12:10 PM, anz118466 at cse.iitd.ac.in wrote:
>>>> Dear all,
>>>>
>>>> I am a new user of DFTB+ and am looking at plotting the transmission
>>>> function of a graphene nanoribbon in the presence/absence of a
>>>> biomolecule
>>>> using the DFTB+NEGF package. I have a confusion related to basic
>>>> geometry
>>>> definition of the device region, which I am sharing here.
>>>>
>>>> >From the example given in
>>>> (http://www.dftb-plus.info/fileadmin/DFTB-Plus/public/recipes-negf/html/transport/specifying_geometry.html),
>>>> I understand that my system needs to be partitioned into PLs. As per
>>>> the
>>>> reference link, a PL is a contiguous group of atoms. Since my system
>>>> consists of a biomolecule placed on top of a continuous GNR sheet, I
>>>> am
>>>> unable to decide upon the PL definition for my device region. How does
>>>> one
>>>> define a multiple layer device in DFTB+?
>>>>
>>>> I believe I might have misunderstood a fundamental point here. Any
>>>> clarification would be extremely helpful.
>>>>
>>>> Thanks in advance.
>>>>
>>>> Kind regards,
>>>> Anusha Iyer
>>>> PhD student at Indian Institute of Technology, Delhi.
>>>>
>>>> _______________________________________________
>>>> DFTB-Plus-User mailing list
>>>> DFTB-Plus-User at dftb-plus.info
>>>> http://www.dftb-plus.info/mailman/listinfo/dftb-plus-user
>>>
>>> --
>>> Dr. Gabriele Penazzi
>>> BCCMS - University of Bremen
>>>
>>> http://www.bccms.uni-bremen.de/
>>> http://sites.google.com/site/gabrielepenazzi/
>>>
>>> phone: +49 (0) 421 218 62337
>>> fax: +49 (0) 421 218 62770
>>>
>>> _______________________________________________
>>> DFTB-Plus-User mailing list
>>> DFTB-Plus-User at dftb-plus.info
>>> http://www.dftb-plus.info/mailman/listinfo/dftb-plus-user
>>>
>>
>> _______________________________________________
>> DFTB-Plus-User mailing list
>> DFTB-Plus-User at dftb-plus.info
>> http://www.dftb-plus.info/mailman/listinfo/dftb-plus-user
>
>
> --
> Dr. Gabriele Penazzi
> BCCMS - University of Bremen
>
> http://www.bccms.uni-bremen.de/
> http://sites.google.com/site/gabrielepenazzi/
>
> phone: +49 (0) 421 218 62337
> fax: +49 (0) 421 218 62770
>
> _______________________________________________
> DFTB-Plus-User mailing list
> DFTB-Plus-User at dftb-plus.info
> http://www.dftb-plus.info/mailman/listinfo/dftb-plus-user
>