ABINIT Discussion Forums

Hello,
I am very new and very alone in the topic of DFT. Usually I am doing XPS measurements, but now I want to calculate the DOS of Polymers (beginning with PP and goal is PA6) using Abinit and than compare it with the XPS spectra (valence band) by convolve it with the correct Voigt-profile. I already calculated the DOS of Diamond, by following the tutorial (PAW1_4).
Maybe someone can give me hints how to start for polypropylene. The first problem which comes to my mind is how to define the cell parameter (rprim, xred or xcut, ...). For acell I would suggest periodic boundaries in one dimension and "infinite/large" distances in other (for 1D polymer).
e.g. acell "size of monomer" "large number" "large number" (1D)
e.g. acell "size of monomer" "smaller number" "large number" (2D)
thx for your help!

BR
maths

Try to provide a input file with the basic variables like
natom
ntypat
typat
znucl
acell
rprim
xred or xcart or xangst

Or if you have a xyz file to be reuse ?

You acell guesses for 1D and 2D are correct in the idea

Once you have that, you can work on other input parameter which drive the calculation (ecut/ecutdg/nstep/...)

cheers

Jordan

Thx for the hint!
I can not really find an input file.
I now try to use xyzfile for the tutorial PAW1_4:
So i created the diamond structure in VESTA and saved it as xyz file.
Now i want to replace this:
#Cell and atoms definition
acell 3*3.567 angstrom
rprim 0.0 1/2 1/2
1/2 0.0 1/2
1/2 1/2 0.0
ntypat 1 natom 2 typat 2*1
xred 0.0 0.0 0.0
1/4 1/4 1/4
znucl 6 nband 6
Abinit help explains xyzfile: Gives the name of a xyz format file, to take natom, ntypat, typat, znucl, and xangst from.
This means for me I still have to give an expression for prim, xred/xcat, cell ? (which is the most complicated thing)
Further Abinit help explains: Notes: do not quote the file name in the abinit input file, simply leave a space after xyzfile.
Therefore I tried this:
acell 3*3.567 angstrom
rprim 0.0 1/2 1/2
1/2 0.0 1/2
1/2 1/2 0.0
xyzfile
xred 0.0 0.0 0.0
1/4 1/4 1/4
nband 6
and i saved the Diamond.xyz file in the same folder as the in file and files file and.
So how do i correctly use this xyz file?
thx
lg maths

Yes you still have to define the cell parameter since the xyz file format does not provide it.
So in your input file you should have

xyzfile Diamond.xyz
rprim ..........
acell .........

nband .....
ecut.........
nstep .......
........


The xyz file contains cartesian positions which will be used as the input variable xangst. The type of each atom is also in the xyz file.

Hello,
the use of the xyzfile did not work, so I typed in the atomic position by hand:
xangst
0.000000 0.000000 0.000000
0.000000 1.783395 1.783395
1.783395 1.783395 0.000000
1.783395 0.000000 1.783395
2.675093 0.891698 2.675093
0.891698 0.891698 0.891698
0.891698 2.675093 2.675093
2.675093 2.675093 0.891698

i changed: ntypat 1 natom 8 typat 8*1
and scaled band from 6 to 24 (because I have now 4 times more atoms in the supercell)
it dint reaches tolwfr as quickly as before so I changed it to tolwfr 1.5d-7 ... the I got a similar but not identical DOS as before.
the log file gifs always the warning:

PAW SPHERES ARE OVERLAPPING !
Distance between atoms 2 and 6 is : 2.91862
PAW radius of the sphere around atom 2 is: 1.50000
PAW radius of the sphere around atom 6 is: 1.50000
This leads to a (voluminal) overlap ratio of 0.11 %

COMPENSATION DENSITIES ARE OVERLAPPING !!!!
Distance between atoms 2 and 6 is : 2.91862
Compensation radius around atom 2 is: 1.50000
Compensation radius around atom 6 is: 1.50000
This leads to a (voluminal) overlap ratio of 0.11 %
THIS IS DANGEROUS !, as PAW formalism assume non-overlapping compensation densities.

=> Note that other spheres are overlapping !

Overlap ratio seems to be acceptable (less than value
of "pawovlp" input parameter): execution will continue.
But be aware that results might be approximate,
and even inaccurate (depending on your physical system) !

so i had a look and the PAW radii is ratsph 1.5 larger than the atomic radii in with. ?

But know my important question i now want to try this also for PP.
..therefore I changed the following:

#Cell and atoms definition
acell 2.1247 5 5 angstrom
rprim 1.0 0.0 0.0
0.0 1.0 0.0
0.0 0.0 1.0
ntypat 2 natom 6 typat 2*1 4*2
xangst
0.63380 0.00000 0.00000
1.90140 0.00000 0.85500
0.63380 -0.85900 -0.63587
0.63380 0.85900 -0.63587
1.90140 -0.85900 1.49090
1.90140 0.85900 1.49090

znucl 6 1 nband 18 #? don t really know which number nband schuld have.

Next question is which ratsph for the hydrogen? (where do I get this info)
And finally i need a second 6c.lda.atompaw file... so where is it ? (I could only find a 1h.atompaw ...but it seems to be not the correct one)
thx so much for your help!
lg
maths

Hi

The problem you have with PAW is that the PAW spheres are overlapping because you geometry is not optimized.
If you perform a relaxation (read about optcell ionmov, ecutsm, dilatmx, ntimes, tolmxf toldff input parameters), forces between atoms will decreases and the distances between atoms should be different.

Note that you are simulating a molecule so you need to increase the acell parameters until convergence of the total energy or whatever you are interested in.
Using the PAW atomic data from the JTH table http://www.abinit.org/downloads/PAW2 I did a quick and dirty relaxation and get this new structure :
natom 6

ntypat 2

# Species
# C H
znucl
6 1

typat
1 1 2 2 2 2

acell 4.85395520568669e+00 1.14226836863715e+01 1.14226836863715e+01

rprim
1.00000000000000e+00 0.00000000000000e+00 0.00000000000000e+00
0.00000000000000e+00 1.00000000000000e+00 0.00000000000000e+00
0.00000000000000e+00 0.00000000000000e+00 1.00000000000000e+00

# angdeg 9.00000000000000e+01 9.00000000000000e+01 9.00000000000000e+01

# xcart
# 1.68335418685641e+00 0.00000000000000e+00 1.86899277036390e-01
# 4.11028237593453e+00 0.00000000000000e+00 1.76716924515101e+00
# 1.68201909043002e+00 -1.66611931093456e+00 -1.07404821073673e+00
# 1.68201909043002e+00 1.66611931093456e+00 -1.07404821073673e+00
# 4.10879893268121e+00 -1.66666261074154e+00 3.02700085129942e+00
# 4.10879893268121e+00 1.66666261074154e+00 3.02700085129942e+00

xred
3.46800519478272e-01 0.00000000000000e+00 1.63621161338278e-02
8.46790339375011e-01 0.00000000000000e+00 1.54707010512725e-01
3.46525466172294e-01 -1.45860583789292e-01 -9.40276593685416e-02
3.46525466172294e-01 1.45860583789292e-01 -9.40276593685416e-02
8.46484724017953e-01 -1.45908147025909e-01 2.64999096045264e-01
8.46484724017953e-01 1.45908147025909e-01 2.64999096045264e-01

Note that acell is quite larger than yours and you should try even bigger.
Since you have 2 C and 4 H, and because in the PAW atomic data from JTH there are 1 electron for H and 4 for C, you have a total of 2*4+2 = 12 electrons.
Assuming you are not polarized, you can set 2 electrons in on band so nband should be at least 6.

Cheers

Thx a lot!
1. The I chose this acell because I want to investigate Polyethylene and not a single molecule. I don't know a other way to get a periodic molecule.
The overlapping i get for the diamond structure. (I will try the relaxation for this problem! thx)

2. For the polypropylene I don t know how tho perform a relaxation because if the positions of the atoms change ... then the acell shape may be won't give the correct period for the polymer and the DOS is not correct anymore??

3. Augmentation radius: rpaw=0.9 a.u. (is it the same as ratsph?)

4. Thx for the link, but I am to stupid and again lost, so which is the correct file. I can download a folder (atompaw-4.0.0.12) full of something

lg
maths

For point 2) You can move the atoms keeping the same cell (optcell 0 and ionmov 2) or optimize both the positions and the cell parameter (optcell 1 ionmov 2)

3) ratsph is a variable of abinit for integrating the density inside a sphere around an atom. In the cas of PAW, ratsph is set to the PAW radius found in the PAW Atomic file.

4) On the page I linked you have a periodic table. You can click on all green elements to download the PAW atomic dataset in the LDA or GGA approximation.

++

Hello,
I am sill at the problem of calculating the DOS for polypropylene.
How I proceeded so far:
I defined the Cell and ran a relaxation like:
#determine of latice parameters
#optcell 0 (to not get a sigle molecule but a 1D periodic polymer)
#ionmov 2
#ntime 10
#dilatmx 1.1
My new Cell definition after the relaxation was then:
#Cell and atoms definition
acell 2.1247 5 5 angstrom
rprim 1.0 0.0 0.0
0.0 1.0 0.0
0.0 0.0 1.0
ntypat 2 natom 6 typat 2*1 4*2
xangst
7.3613206833E-01 0.0000000000E+00 -2.9126225013E-02
1.7990959478E+00 0.0000000000E+00 8.8412413435E-01
7.2116638321E-01 -8.9877079481E-01 -7.1413350848E-01
7.2116638321E-01 8.9877079481E-01 -7.1413350848E-01
1.8140196087E+00 -8.9876375336E-01 1.5691645538E+00
1.8140196087E+00 8.9876375336E-01 1.5691645538E+00

For the relaxation I choose pseudo potential, because the ran more stable.
Now I want to calculate the DOS. Therefore I tried it like in the PAW1 tutorial:
#Output of the projected DOS (with PAW contribs)
prtdos 3 pawprtdos 1
natsph 1 iatsph 1 ratsph 0.9 1.5

I obtained a DOS file but it look not like a DOS .
My questions:
1. Is the way of defining the DOS output correct for my problem.
2. Can I use pseudopotential or do I have to use PAW pot.? (If yes: Which PAW? 1h.paw and 6c_lda.paw?)
3. Does the order of the potentials in the files file matter? (If yes: Dose it mean 1.H 2.C in files file mean also 1.H 2.C in the in file? e.g. typat 2*1 4*2 and ratsph 0.9 1.5?)

thx for your time and help!
maths

Hello

You should always keep the ecutsm value from the relaxation in any additional run. It is very important to keep the pressure/max force residue the same as during the relaxation.

You should always use the same pseudo for all your calculations on 1 material. Don't switch from NC to PAW.
In NC, only prtdos is relevant, whereas in PAW the additional variable pawprtdos is relevant too (2 might be the value you want)
You can use the JTH PAW atomic data files available in the download section the abinit website.

Yes the order matters. Actually, if the order is not good abinit stops. FYI they must be in the same order as the znucl variable in the abinit input file.

Cheers

Jordan