[Pw_forum] Silicon 111 surface
Giovanni Cantele
giovanni.cantele at spin.cnr.it
Thu Feb 23 12:19:23 CET 2017
Hi,
this is the reply to the message you sent me, it is better for me to continue di discussion on PW_forum, being a way
to share doubts, mistakes, problems with many other people that might help!
You can build your Si(111) surface with any tool (by hand, using VNL, any other builder), but this tool will do just what
you ask it to do, so before using the tool you ALWAYS need basic understanding. In this respect, I also suggest to look
to the extensive documentation of Quantum-ESPRESSO and to the tutorials that are freely available in the web.
Coming to your problem, Quantum-ESPRESSO, as you know, is based on plane waves, as such can only handle
systems that are periodic in all three directions. If you need to simulate something that is periodic in only two, one or no
direction in space (a surface, a 1D wire, a cluster/molecule/quantum dot respectively) you use the supercell approach.
Let’s consider the case of a surface and let’s choose x and y as directions parallel to the surface, such that the z direction is orthogonal
to it (you can make a different choice, of course).
1) In the x-y plane you have to identify the surface unit cell, spanned by two vectors a1 and a2 that describe the surface periodicity.
Beware the choice of a1 and a2: they might be deduced from the bulk crystals or be more complicated if the surface
is along special directions or if it undergoes some reconstruction.
In the case of Si(111) you should first decide if you want to choose the minimal cell, or if you want to consider complex reconstruction
(that is structures where one or more surface unit cells are gathered to build more complex structures) The minimal cell is obtained
by considering the 111 plane of Si and trying to understand the periodicity vectors within that plane. The complex reconstruction are instead
obtained once you know how many cells are gathered along the a1 and a2 directions. For example, in the case of Si(111), whose structure
had been long debated many years ago, it can undergo a 7x7 reconstruction (with a very large surface unit cell). You can find details wherever,
for example this is one of the Google search results:
http://nanowiz.tripod.com/sisteps/si111.htm <http://nanowiz.tripod.com/sisteps/si111.htm>
2) once the surface unit cell is identified, you must decide how many planes you want to consider in the direction (z-axis) orthogonal to the surface.
In this case beware the fact that as you move in along z, you might meet one or more planes that are not equivalent to the 1st one. In other words,
the second plane might not be obtained simply through a rigid shift along z of the 1st one, but there might be 2 or more planes before one equivalent
to the first one is found
3) at this point you have the minimal number of atomic coordinates that, replicated along a1 and a2 reproduce a slab, that is, a Si(111) surface composed
by a certain number of planes. How to choose a3 (this is one of the issues in your input file)? You measure the distance d along the z axis between the two furthermost planes
AND and set the length of a3 to d+V, where V=vacuum space. The latter is needed because, in the supercell approach, you system is replicated
also along a3, but you don’t want that periodic replicas interact with each other. The amount of V, let’s say 10 A (but a larger value might be needed), is chosen
just to avoid interaction and convergence with respect to it must always be checked
4) because the z direction is not a true periodicity direction, if V is large enough you expect that the band dispersion is zero along the z direction, that is,
you should obtain flat bands along that direction. As such, Brillouin zone sampling along that direction is useless, and the k-point grid you use for scf/relax
calculations is in the form N x M x 1
Whether you performs steps 1-4 by hand or using some tool such as VNL is up to you. I suggest to start with simple surfaces and simple crystals, to start with.
Giovanni
PS after writing this message, I have tried the Google search
build surface supercell Quantum-ESPRESSO
The 1st result is this link (I’m sure that many many others of interest for you show up!)
http://www.democritos.it/events/pw-tutorial/lectures/pw-tutorial.pdf
If I were you, I would start with the exercises described therein.
> Actually sir I am trying to create a silicon 111 surface adsorbed by an Indium atom.To do this what should I be do first? The input file which I am using is create by the VNL GUI. Is there any other way to create a quantum espresso input file.
> Thanx
>
>
>> On 23 Feb 2017, at 10:18, Giovanni Cantele <giovanni.cantele at spin.cnr.it> wrote:
>>
>> 1) the atomic positions are likely wrong, I cannot understand what is the system you want to calculate. For example, the Si-In distance is 1.1 A, that looks
>> very small. Try to look at your structure, opening the input file with XCrysDen. It is worth, before performing any further calculation, to check whether it is EXACTLY what you want.
>>
>>
>> 2) in the case you would like to simulate the Si(111) SURFACE in the presence of In, your input does not correspond to a surface calculation. There is not vacuum space
>> introduced between periodic replicas of the same slab, and there is k-point sampling in the direction orthogonal to the surface. If instead you are studying something like
>> a Si heterostructure (e.g. a superlattice), please disregard this point
>>
>> 3) you are using ultra soft pseudo potential, the ecutwfc is likely too high (this however does not give convergence issues, but can make your calculation much slower
>> than needed!). On the other hand, you are not setting the ecutrho variable, that by default is 4*ecutwfc. For ultrasoft pseudo potentials, as you might learn reading
>> the relevant literature on the subject, a higher charge density cut-off is required, usually from 6 to 12 times ecutwfc
>>
>> 4) to improve convergence, increasing degauss might help, for systems very difficult to converge (but even in this case you should be aware of what degauss is and how to use it)
>> and also decrease mixing_beta to 0.3, or 0.1 or maybe even less
>>
>>
>> Giovanni
>>
>>
>>
>>> On 23 Feb 2017, at 06:59, Ubaid Mohd <onlymubaid at gmail.com <mailto:onlymubaid at gmail.com>> wrote:
>>>
>>> Hi,
>>>
>>> I am trying to scf calculation of silicon(111)_Indium and the total force on the atom is not converging. Initially, the scf cycle itself was not converging.
>>> What is wron with my input file.
>>> Thanx,
>>> Ubaid
>>>
>>>
>>>
>>>
>>> Input File
>>>
>>>
>>> &CONTROL
>>> calculation='scf',
>>> outdir='Pwscf',
>>> prefix='Si_In',
>>> pseudo_dir='.',
>>> verbosity='low',
>>> /
>>>
>>> &SYSTEM
>>> ibrav=0,
>>> celldm(1)=7.2565749368d0,
>>> nat=7,
>>> ntyp=2,
>>> ecutwfc=90,
>>> input_dft='PBE',
>>> occupations='smearing',
>>> smearing='mv',
>>> degauss=0.005d0,
>>> /
>>>
>>> &ELECTRONS
>>> conv_thr=1d-06,
>>> mixing_beta=0.7d0,
>>> /
>>>
>>> ATOMIC_SPECIES
>>> In 114.818000d0 In.pbe-d-rrkjus.UPF
>>> Si 28.085500d0 Si.pbe-n-rrkjus_psl.0.1.UPF
>>>
>>> ATOMIC_POSITIONS {crystal}
>>> Si 0.1708493746d0 0.3787081237d0 0.2302895286d0
>>> Si 0.8375160412d0 0.7120414570d0 0.3136228619d0
>>> Si 0.8375160412d0 0.7120414570d0 0.5636228619d0
>>> Si 0.5041827079d0 1.0453747903d0 0.6469561953d0
>>> Si 0.5041827079d0 1.0453747903d0 0.8969561953d0
>>> Si 0.1708493746d0 0.3787081237d0 0.9802895286d0
>>> In 0.2708493746d0 0.4787081237d0 0.3397104714d0
>>>
>>> K_POINTS {automatic}
>>> 5 5 5 0 0 0
>>>
>>> CELL_PARAMETERS {alat}
>>> 1.000000000000d0 0.000000000000d0 0.000000000000d0
>>> -0.500000000000d0 0.866025403784d0 0.000000000000d0
>>> 0.000000000000d0 0.000000000000d0 2.449489742783d0
>>>
>>>
>>> _______________________________________________
>>> Pw_forum mailing list
>>> Pw_forum at pwscf.org <mailto:Pw_forum at pwscf.org>
>>> http://pwscf.org/mailman/listinfo/pw_forum
>>
>> --
>>
>> Giovanni Cantele, PhD
>> CNR-SPIN
>> c/o Dipartimento di Fisica
>> Universita' di Napoli "Federico II"
>> Complesso Universitario M. S. Angelo - Ed. 6
>> Via Cintia, I-80126, Napoli, Italy
>> e-mail: giovanni.cantele at spin.cnr.it <mailto:giovanni.cantele at spin.cnr.it>
>> Phone: +39 081 676910
>> Skype contact: giocan74
>>
>> ResearcherID: http://www.researcherid.com/rid/A-1951-2009 <http://www.researcherid.com/rid/A-1951-2009>
>> Web page: http://people.na.infn.it/~cantele <http://people.na.infn.it/~cantele>
>>
>> _______________________________________________
>> Pw_forum mailing list
>> Pw_forum at pwscf.org
>> http://pwscf.org/mailman/listinfo/pw_forum
--
Giovanni Cantele, PhD
CNR-SPIN
c/o Dipartimento di Fisica
Universita' di Napoli "Federico II"
Complesso Universitario M. S. Angelo - Ed. 6
Via Cintia, I-80126, Napoli, Italy
e-mail: giovanni.cantele at spin.cnr.it
Phone: +39 081 676910
Skype contact: giocan74
ResearcherID: http://www.researcherid.com/rid/A-1951-2009
Web page: http://people.na.infn.it/~cantele
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20170223/f36ab261/attachment.html>
More information about the users
mailing list