[QE-users] Defining C diamond structure cp.x

Tue Mar 1 04:05:39 CET 2022

Dear Juniper,

    Perhaps I could help a bit.
    From the crystallographic point of view you don't need 18 atoms to
represent the diamond structure, actually you need only 1 atom at (0,0,0),
because the symmetry operations (translations and rotations) of the Fd-3m
space group will take care of generating the whole structure (the infinite
structure) and not only the representation of a single conventional unit
cell.
    For a nice pictographic representation, yes, you need to draw 18 atoms.
Thus, when you input the diamond CIF file, with just one atom at (0,0,0),
in the Vesta software, for example, it will apply the symmetry operations
and generate a nice representation with 18 atoms. If you use more atoms
than needed the software will just superimpose them and you'll not notice.
     Regarding MD with QE you have, at least, translational symmetry, i.e.,
you have periodic boundary conditions of your unit cell, regardless what it
is. That's why you can't use the 18 positions because you are going to
superimpose many of them and the potentials will not be correctly
calculated. Therefore you have some options to generate your simulation box
by using this periodic boundary condition, as already pointed out by
others, for example, by using ibrav=0, ibrav=1 or even ibrav=2.
     The way you input the structure will not affect the calculation
accuracy if it is done properly. Of course you'll notice minor differences
because of numeric approximations but all of them must give closely the
same results. The calculation velocity will be greatly affected. If you use
a very simple primitive cell with only 2 atoms the calculation will be
faster then using a conventional unit cell with 8 atoms.
     Regarding accurate results, much more important than the way you input
your structure is the calculation method and the convergence tests you must
apply for being sure that your calculation is performing properly.
Therefore, choosing the correct potential, the correct k-mesh, energy
cutoffs and related things are far more important for having accurate
results than the way you input your structure.
    If you really want to understand how QE works I strongly suggest you to
follow the tutorials of the web page (
https://www.quantum-espresso.org/tutorials/). It will take some time but at
the end you'll realize that you have saved time. They worked fine for me.
   A final remark, since you are going to work with crystals, going more
deep in the subject with a good crystallography book is also a good idea.
It looks like a long way in the beginning but you certainly will save time
at the end.

Best regards,

Em seg., 28 de fev. de 2022 às 20:39, Juniper Savchick <curly363 at gmail.com>
escreveu:

> To all,
>
> It is my understanding that in material science the crystal lattice
> structure is an FCC structure with 4 extra atoms inside of it. My PhD
> supervisor describes it as two overlapping FCC lattice structures. To
> create a visualization of this structure you must define the positions of
> 18 atoms where most of those atoms are only partially inside the unit cell.
> Therefore we say that there are 8 Carbon atoms inside the diamond crystal
> lattice structure. However, when you perform MD calculations you must use
> whole atoms because fractional atoms are non-physical. Therefore I need to
> define the position of 18 atoms to model the entire diamond crystal lattice.
>
> I was under the impression that the ibrav parameter in cp.x was used to
> fill out the remaining atomic positions based on the two atomic positions
> we define in the input file. I would like to verify this by defining my own
> atomic positions and using the CELL_PARAMETERS section of the cp.x input
> (ibrav=0). However, when I do this cp.x gives infinite or null data output
> values for energy and pressure.
>
> I am confused as to why my "18atom.txt" input (see google drive link in
> original message) will not work properly. I have a decent understanding of
> the physics but am confused about implementation via cp.x. I have been
> using Avogadro2 to visualize each structure, generate atomic positions and
> lattice vectors, and then implement this data into my cp.x input files.
> This method works fine when I define 8 atoms but once I try to scale up to
> all 18 atoms cp.x gives the aforementioned nonsensical outputs.
>
> Does anyone have any recommendations for how I can get my "18atom.txt"
> input file to execute properly?
>
> Sincerely,
> Juniper Savchick
> Graduate Assistant
> Virginia Commonwealth University
>
>  QE inputs
> <https://drive.google.com/drive/folders/1BLY8nVULkYzAWleKz-n6nwjFaKHc_U4m>
>
>
> On Mon, Feb 28, 2022 at 2:31 PM Juniper Savchick <curly363 at gmail.com>
> wrote:
>
>> Hello,
>>
>> I am a graduate assistant at Virginia Commonwealth University and I am
>> using Quantum Espresso to perform CP molecular dynamics calculations. I
>> would like to run a simple set up of a single Carbon Diamond crystal
>> structure. My ultimate research goal is to generate equations of state for
>> warm dense matter, starting with Carbon in a diamond structure. It is my
>> understanding that there are multiple ways to input this system into
>> Quantum Espresso but I am not getting the same results with each method. I
>> want to make sure I have an intimate understanding of the cp.x executable.
>> This is why I want to calculate the same system using different inputs. (I
>> am using QE 7.0) I have tried many different input methods with the main
>> differences being the number of atoms that are explicitly defined. (2,8,&
>> 18)
>>
>> It is my understanding that if you use input parameter ibrav=2 where 2
>> atoms are explicitly defined under "ATOMIC_POSITIONS", this is the carbon
>> diamond primitive cell. Based on QE related literature, cp.x will
>> automatically fill in the remaining atoms to complete the carbon diamond
>> crystal structure. I am skeptical that cp.x is actually doing this. Most of
>> the QE related literature pertained to pw.x so perhaps each executable
>> defines the crystal structure differently? I have looked into documentation
>> for both cp.x and pw.x to no avail. I would appreciate anyone who can help
>> clarify this input for me.
>>
>> If you use input parameter ibrav=0 then you should be able to
>> explicitly define all 18 atoms that comprise the conventional carbon
>> diamond lattice structure. However, if I try to use ibrav=0 and explicitly
>> define 18 atomic positions I get infinite values as outputs. My next
>> attempt was to use ibrav=0 and explicitly define 8 atomic positions. The 8
>> atom run gave reasonable values but the pressure was orders of magnitude
>> smaller than the 2 atom run. This is concerning to me and it has become
>> clear that I am misunderstanding the cp.x input. I have scoured QE forums
>> and documentation to no avail.
>>
>> At this point it would seem the most accurate method is the 8 atom run
>> since I can use these positions to generate a supercell that has the
>> periodic carbon diamond structure, albeit with vacancies in some of the
>> corners and sides. I assume if I make the supercell large enough the errors
>> due to imperfections will be negligible. Also, no real structure is
>> perfectly periodic so we should still be able to use the 8 atom unit cell
>> to construct at usable supercell for modeling purposes. Does this
>> assumption sound reasonable?
>>
>> To summarize by questions:
>> What is the most accurate way to input the carbon diamond crystal
>> structure in cp.x?
>> Why does ibrav=0 with 18 atoms explicitly defined not work?
>> Is my understanding of how to input the primitive cell in cp.x correct?
>> Is my assumption regarding the "8 atom generated supercell" reasonable?
>>
>> I understand this is quite a bit to ask but I am at a loss for what to do
>> next. I would appreciate any help that can be offered. I have provided
>> example input files for each type of input described. (see google drive
>> link)
>>
>> Sincerely,
>> Juniper Savchick
>> Graduate Assistant
>> Virginia Commonwealth University
>> BS Physics, minor Mathematics
>> Pursuing a PhD in Mechanical and Nuclear Engineering
>>
>>  QE inputs
>> <https://drive.google.com/drive/folders/1BLY8nVULkYzAWleKz-n6nwjFaKHc_U4m>
>>
>>
>> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users

-- 
Marcelo Falcão de Oliveira
Professor Associado
Departamento de Engenharia de Materiais
EESC - USP
Av. João Dagnone, 1100,  Jd. Sta Angelina
CEP: 13563-120, São Carlos - SP - Brasil
e-mail: marcelo.falcao at usp.br
Fone: (16) 3373-9583
Fax: (16) 3373-9590
Agenda:
http://www.google.com/calendar/embed?src=marcelo.falcao%40usp.br&ctz=America/Sao_Paulo&mode=week
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20220301/56d78f57/attachment.html>