Dear Mona Asadi Namin,

> However, looking at the output, I see that after a scf iteration, the 
> occupations end up being partial and also will be almost the same for all the 
> Nb atoms, which is not expected since one of them is a substitutional.


I am not familiar with your system, so I would need to read the paper that you 
cited to understand better the problem. But I would suggest to ask the authors 
of that paper to send you the input file which they used to produce the results 
(since they also used Quantum ESPRESSO, it should be straightforward to re-run 
their calculations).


Greetings,

Iurii


--
Dr. Iurii Timrov
Postdoctoral Researcher
STI - IMX - THEOS and NCCR - MARVEL
Swiss Federal Institute of Technology Lausanne (EPFL)
CH-1015 Lausanne, Switzerland
+41 21 69 34 881
http://people.epfl.ch/265334
________________________________
From: users <[email protected]> on behalf of Mona 
Asadinamin <[email protected]>
Sent: Friday, August 7, 2020 6:31:14 PM
To: Quantum ESPRESSO users Forum
Subject: Re: [QE-users] Lithium Niobate Hubbard Calculations

Dear Timrov Lurii;

Thank you for your help.

I agree with you and I think it is very important that people provide 
sufficient information and data for the reader to be able to reproduce their 
work.

In this paper, they have substituted one of the Li atoms with a Nb. I applied 
Hubbard term on the substitutional atom and in another run, I applied Hubbard 
to other Nb atoms as well. My band structure looks almost identical.

As you advised, I did another calculation with smaller smearing (please see the 
attached input). I used ecut=40 and kpoint=222 since my unit cell with 10 atoms 
was converged with this ecut and kpoint=444.

When I plot the band structure, I do not get any bands in the gap region.


Here is what I am thinking:

Based on Nb electronic configuration: [Kr] 4d4 5s1 and considering the fact 
that Nb sits in a Li site with a 5+ charge states, I defined tot_charge=+5 to 
empty the s and d orbitals.
To make sure this will be the case, I manually set the substitutional Nb 
occupations to 0.
However, looking at the output, I see that after a scf iteration, the 
occupations end up being partial and also will be almost the same for all the 
Nb atoms, which is not expected since one of them is a substitutional.


Best regards;

Mona Asadi Namin
Graduate student
Center for simulational physics
University of Georgia
--------------------------------------
Email:[email protected]<mailto:[email protected]>u<mailto:[email protected]>
Phone: +1-215-906-23-92


________________________________
From: users <[email protected]> on behalf of Timrov 
Iurii <[email protected]>
Sent: Thursday, August 6, 2020 4:10 AM
To: Quantum ESPRESSO users Forum <[email protected]>
Subject: Re: [QE-users] Lithium Niobate Hubbard Calculations

[EXTERNAL SENDER - PROCEED CAUTIOUSLY]


Drear Mona Asadi Namin,


I had a quick look at your problem. You defined two types of Nb atoms: "Nb1" 
and "Nb". Then you specified only "Hubbard_U(1)=4", which means that the 
Hubbard U correction is applied only to "Nb1" and no U correction for "Nb". Is 
this really what you want? Why do you define two types of Nb atoms?


> Also, the number of orbitals treated with Hubbard term is printed as 3 
> whereas it should be 4!


I do not understand this point. Could you clarify please?


In your input you have:

> occupations='smearing',
> smearing='mv',
> degauss=0.001,


The broadening of 0.001 Ry is extremely small, and your k-mesh is 2x2x2 with a 
shift for 80-atoms supercell. The default value is 0.02 Ry, so I would try 
that, and then decrease it if necessary.


> PHYSICAL REVIEW MATERIALS 1, 054406 (2017)


Your calculation is quite delicate, because you are trying to adjust manually 
the total magnetization and to force some eigenvalues of the occupation matrix 
to be 0 or 1. Maybe you can contact the corresponding author of this paper and 
ask for the input.

As we can see it is very difficult to reproduce results of some papers. This is 
why it is very useful and important to provide the source data together with 
the paper (using some public repositories, as e.g. Materials Cloud Archive: 
https://archive.materialscloud.org/)


HTH


Regards,

Iurii


--
Dr. Iurii Timrov
Postdoctoral Researcher
STI - IMX - THEOS and NCCR - MARVEL
Swiss Federal Institute of Technology Lausanne (EPFL)
CH-1015 Lausanne, Switzerland
+41 21 69 34 881
http://people.epfl.ch/265334
________________________________
From: users <[email protected]> on behalf of Mona 
Asadinamin <[email protected]>
Sent: Wednesday, August 5, 2020 8:24:47 PM
To: Quantum ESPRESSO users Forum
Subject: [QE-users] Lithium Niobate Hubbard Calculations


Dear all;


And I am trying to reproduce the results for NbLi(SLN) of this paper:


PHYSICAL REVIEW MATERIALS 1, 054406 (2017)


  Although I could calculate the band structure of the plain lithium niobate, I 
have some difficulties in my Hubbard calculations and I really appreciate it if 
you could please guide me through that.



   Nb sits in Li site with +5 charge, but since Li already has +1, I defined 
the total charge of the unit cell as +4. I am also setting the occupations of 
the substitutional Nb manually, so that I have 4 spin ups and zero spin downs. 
(I have attached my input file)

    Below, is a section of my output. Although the occupations are set to the 
manual ones, after a scf iteration, it changes to incorrect and partial 
occupations! Also, the number of orbitals treated with Hubbard term is printed 
as 3 whereas it should be 4!


Output:


Before scf:


LDA+U parameters:

U( 1)     =  4.00000000

alpha( 1) =  0.00000000

atom    8   Tr[ns(na)] (up, down, total) =   4.00000  0.00000  4.00000

   spin  1

    eigenvalues:

  0.000  1.000  1.000  1.000  1.000

    eigenvectors:

  1.000  0.000  0.000  0.000  0.000

  0.000  0.011  0.235  0.086  0.669

  0.000  0.178  0.176  0.637  0.008

  0.000  0.018  0.588  0.072  0.322

  0.000  0.793  0.001  0.206  0.001

    occupations:

  0.000 -0.000  0.000  0.000  0.000

 -0.000  1.000  0.000  0.000  0.000

  0.000  0.000  1.000 -0.000 -0.000

  0.000  0.000 -0.000  1.000  0.000

  0.000  0.000 -0.000  0.000  1.000

   spin  2

    eigenvalues:

  0.000  0.000  0.000  0.000  0.000

    eigenvectors:

  1.000  0.000  0.000  0.000  0.000

  0.000  1.000  0.000  0.000  0.000

  0.000  0.000  1.000  0.000  0.000

  0.000  0.000  0.000  1.000  0.000

  0.000  0.000  0.000  0.000  1.000

    occupations:

  0.000  0.000  0.000  0.000  0.000

  0.000  0.000  0.000  0.000  0.000

  0.000  0.000  0.000  0.000  0.000

  0.000  0.000  0.000  0.000  0.000

  0.000  0.000  0.000  0.000  0.000

atomic mag. moment =     4.000000

N of occupied +U levels =    4.000000



After scf:


LDA+U parameters:

U( 1)     =  4.00000000

alpha( 1) =  0.00000000

atom    8   Tr[ns(na)] (up, down, total) =   3.57808  1.74127  5.31935

   spin  1

    eigenvalues:

  0.503  0.503  0.802  0.885  0.885

    eigenvectors:

  0.000  0.000  1.000  0.000  0.000

  0.068  0.489  0.000  0.047  0.397

  0.489  0.068  0.000  0.397  0.047

  0.080  0.364  0.000  0.090  0.466

  0.364  0.080  0.000  0.466  0.090

    occupations:

  0.802 -0.000 -0.000 -0.000 -0.000

 -0.000  0.673 -0.000 -0.189 -0.016

 -0.000 -0.000  0.673 -0.016  0.189

 -0.000 -0.189 -0.016  0.715  0.000

 -0.000 -0.016  0.189  0.000  0.715

   spin  2

    eigenvalues:

  0.266  0.266  0.271  0.469  0.469

    eigenvectors:

  0.000  0.000  1.000  0.000  0.000

  0.047  0.467  0.000  0.055  0.432

  0.467  0.047  0.000  0.432  0.055

  0.079  0.408  0.000  0.097  0.416

  0.408  0.079  0.000  0.416  0.097

    occupations:

  0.271 -0.000 -0.000 -0.000 -0.000

 -0.000  0.365 -0.000  0.101  0.011

 -0.000 -0.000  0.365  0.011 -0.101

 -0.000  0.101  0.011  0.370  0.000

 -0.000  0.011 -0.101  0.000  0.370

atomic mag. moment =     1.836805

N of occupied +U levels =    5.319347



When I plot the band structure, the overall valence and conduction bands do not 
look bad (although not similar), and I see three dispersion less bands in the 
gap!


I am very confused. I tried larger U values but it does not help. I also tried 
scf after doing vc-relax, but I get the same results. I tried larger ecuts and 
kpoints, but still the same thing happens.

I appreciate any comments or suggestions.



Best regards;


Mona Asadi Namin

Graduate student

Center for simulational physics

University of Georgia

--------------------------------------

Email:[email protected]<mailto:[email protected]>u<mailto:[email protected]>

Phone: +1-215-906-23-92

_______________________________________________
Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso)
users mailing list [email protected]
https://lists.quantum-espresso.org/mailman/listinfo/users

Reply via email to