Dear Angelo: many thanks for taking part in this discussion. You will forgive me if I correct some of your statements that I feel are slightly misleading. Silicon and diamond have *exactly* the same crystal structure. The Bravais lattice of the diamond (or silicon, for that matter) is FCC with two atoms per unit cell. I think you are misled by the fact that *if* you regard an FCC lattice as a simple cubic one, then the SC unit cell contains 4 FCC unit cells. So you can describe a diamond/silicon structure both as FCC (ibrav=2)  with 2 atoms per cell, or as SC (ibrav=1) with 8 atoms per cell. Hope this helps. Stefano Baroni

Hi Yin,

to answer to your question, Quantum Espresso recognizes the symmetry of the structure and reproduces the lattice sites where it puts the atoms of the structure. This is passed to the code through the input command about the crystal lattice in &SYSTEM:

ibrav= # , where # provides the right information about the primitive vectors of the cell, in the respect of the specific kind of crystal lattice. Changing #, it could require more information, such as the size of unit cell in all 3 dimension. In the case of the FCC structure you have to insert just 'celldm(1)= lattice constant in Angstrom/0.5291" which defines your unit cell lenght.

To distinguish between two different kind of crystal with the same symmetry and crystal lattice, is simply based to the fact that these have different unit cell, i.e., different lattice constant, and you insert this parameter through 'celldm(1)' which is different case for case.

I hope I have answered to your questions.

Cheers,

A.

Thank you, Angelo! Your answer really helps me to understand how pwscf works!

Could we discuss in a more detailed way? For example, how pwscf builds the diamond structure just using two basis atoms and "ibrav=2"? The difference between Fig1 and Fig2 that i want to highlight is the different number of atoms in tetrahedron holes. 4 atoms in Fig1, but 8 atoms in Fig2. If you consider all the atoms in Fig1 and Fig2 are silicons and celldm(1) are same, what will be the difference between input files for these two structures?

Dear Li,

that's a good question!

As far as I know, you won't be able to build with QE all 230 space groups structures using that kind of geometry input. Maybe (never tried myself) using atomic site symmetry letters for Wyckoff positions, as presented in QE help, from I cite

"When atomic positions are of type crystal_sg coordinates can be given

in the following three forms (Wyckoff positions):

C 1a

C 8g x

C 24m x y

The first form must be used when the Wyckoff letter determines uniquely

all three coordinates, the second form when the Wyckoff letter

and one coordinate are needed, the third form when one letter and two

coordinates are needed.

The forms:

C 8g x x x

C 24m x x y

are not allowed, but

C x x x

C x x y

are correct."

Altough both cubic FC and having the same cell constant and fractional coordinates, the two different crystals (ZnS and CaF2 - I'll attach the crystalographic informations) belong to different space groups.

I am not an expert in QE, used that software just for some EPR calculations on some doped crystals, a few years ago. As I remember, my conclusion was - at that time - that the sure method to construct any lattice was using the absolute coordinates of all atoms in the given cell (supercell in my case, Cu2+ doped MgO, ill attach that input file for you, too).

Hope to be usefull, best regards!

Hi Yin,

the facts are these: FCC Silicon Lattice is not equal to Diamond Silicon Lattice, though they have many common aspects in their geometry. 

The face centered lattice equals the simple cubic lattice with the addition of a lattice point in the center of each of the six faces of each cube. The face centered cubic lattice contains 4 lattice point per unit cell. The maximum packing density occurs when the atoms have a radius which equals one quarter of the diagonal of one face of the unit cell. The corresponding maximum packing density is 74 %. This is the highest possible packing density of any crystal structure as calculated using the assumption that atoms can be treated as rigid spheres.

The diamond lattice consist of a face centered cubic Bravais point lattice which contains two identical atoms per lattice point. The distance between the two atoms equals one quarter of the body diagonal of the cube. The diamond lattice contains also 4 lattice point per unit cell but contains 8 atoms per unit cell. The maximum packing density occurs when the atoms have a radius which equals one eighth of the body diagonal of the unit cell. The corresponding maximum packing density is 34 %.

The definition of a basis in the lattice, which contains a different number of atoms per site, lead to different structures.

In addition to this, the symmetry groups to which belong each of the two structures are different: Fd3m for Diamond lattice and Fm3m for the FCC lattice, with also different experimental lattice costants for the unit cell, respectively, 4.46 A and 5.74 A (I attached a file which can helps you).

In the end, therefore, they are the input celldm(1) and the choice of a different basis of the lattice, with different ATOMIC POSITIONS to suggest to QE how to organize the structure.

I don't know if I solved definitely your doubts. If not, please let me know.

Cheers,

A.

Well, I'm sure that the crystal symmetry theory is based on mathematical/geometrical "points", not on spheres or packing - but for the resulted packing of elements in our Universe, sure stands also these conditions - not sure that QE is so "powerfull" (?!) ... Well, I'm working with the licesend version of CRYSTAL09 - that piece of software ask for  lattice constants, atomic fractional positions - and for the space group number - but still having restrictions regarding the setting of origin...

So, in my opinion, a geometry input block should be carefull examined before giving a task for long term calculations...

I use VESTA free software to construct, make standard, view, expand structures...

Best regards!

Hi, Angelo and Adrian-Sorin! I greatly appreciate your both efforts on helping me to understand how pwscf recognizes a crystal structure! Your answers are clear for me! I have to learn some theories (space groups...) and techniques (VESTA...) of building a crystal structure prior to mastering pwscf.

As Adrian-Sorin suggested, we have to examine our input file before sending it to a time-consuming calculation. I wonder whether there is a program for examining the crystal structure of input file of pwscf.

Hi Yin,

of course! If you use Quantum Espresso, you can use XCrySDen to analyze your input file of pwscf. It allows to check the crystal structure, distance between atoms in the lattice, angles, it can also produce the k-path for band structure calculations, selected along a specific path among the high symmetry points.

After you installed it, you have to type:

xcrysden --pwi INPUT_NAMEFILE..in

the software will recognize the parameters of the structure to plot the same from the pwscf file provided by QE.

Pleased to have been of help.

Best regards,

A.

Dear Angelo: many thanks for taking part in this discussion. You will forgive me if I correct some of your statements that I feel are slightly misleading. Silicon and diamond have *exactly* the same crystal structure. The Bravais lattice of the diamond (or silicon, for that matter) is FCC with two atoms per unit cell. I think you are misled by the fact that *if* you regard an FCC lattice as a simple cubic one, then the SC unit cell contains 4 FCC unit cells. So you can describe a diamond/silicon structure both as FCC (ibrav=2)  with 2 atoms per cell, or as SC (ibrav=1) with 8 atoms per cell. Hope this helps. Stefano Baroni

HI,

I am beginner to this quantum espresso software. Which software I should use to generate input (see/draw) molecules?

Hi,

you should use XCrySDen with PWgui; the codes are free available here

http://www.xcrysden.org/

and here

http://www-k3.ijs.si/kokalj/pwgui/pwgui.html

Thanks, Adrian-Sorin! It took me a lot of energy to install xCrySDen on my old ubuntu(11.04). Some libraries are required. But "sudo apt-get install" can't work since my version is obsolete now.

Also take a look at cif2cell. Its really smooth to translate available cif into QE (and many other) inputs (Link provided with the comment).

BTW, to cut the log story short, it is the "basis" that differentiates two structures, when their lattice type is the same.

http://sourceforge.net/projects/cif2cell/

Thanks, Vaibhav! I will try it.

hi, this is the BN structure 

I give the command cif2cell -f case.cif -p Quantum espresso

but it does not work

Dear Sinkander,

  I typically use these options. Check if this helps.

cif2cell -f nis2-1991.cif -p quantum-espresso -o nis2-1991.in --cartesian --coordinate-tolerance=0.01 --setup-all

Please also consider the small but finite possibility that  " case.cif "  may not comply with the cif format and/or is erroneous.

I'll try to put my two cents to answer "How does the software distinguish the two lattice on Fig.1 and Fig.2" on examples. The link given leads to helpful slides by the QE-contributor P. Gianozzi.

I think, the simplest algorithm is as follows.

You should obtain exactly your crystal, without multiple atoms in one node or empty nodes.

Example: construct a diamond: let's set ibrav = 2 (FCC). Put C at (0,0,0) and C at (0.25 0.25 0.25) (see slide 19 attached file. Coordinates are lattice parameter fractions as in QE example).

Translating this two atoms starting from each node of FCC lattice (move origin to each node of fcc, lattice is infinite, not restricted to the nodes drawn on slides), one obtains exactly the diamond structure! The same holds for different kinds of atoms, slide 36 for ZnS.

Ok, let's think of diamond as possessing the ("virtually") cubic structure. In that case, more atoms should be explicitly set in ATOMIC_POSITIONS section. For each node of cubic structure (they are only F atoms in the "corners") we also should, e.g., supply C atom on the lower face, C atom on the left face, C atom on the front face, and 4 C atoms located inside the cell inner volume to make our lattice filled without gaps when applying translations, rule 2 above. Please, see slide 16 in the file to understand what atoms I speak about.

Applying the algorithm above for CaF2 (fluorite), one could choose atomic positions for fluorite CaF2 for ibrav=1 (simple cubic, slide 42), or for ibrav=2 (FCC, slide 43).

Also, as Angelo suggested, geometry inspection with

xcrysden --pwi MY_FILE.in

could save a lot time.

http://www.fisica.uniud.it/~giannozz/QE-Tutorial/tutorial_wyckoff.pdf

Old Ubuntu (system)?

Way #1) install Ubuntu to your USB stick. Insert it into your working machine, type in terminal:

mount

to find path where USB is mounted. It is usually somewhere in /media.

If your USB is not mounted automatically, read anywhere how to mount USB in Linux by hand.

Then type commands:

path1="enter_path_to_your_/_on_USB"

# mount devices

mount -o bind /dev "$path1"/dev

mount -o bind /sys "$path1"/sys

mount -o bind /dev/shm "$path1"/dev/shm

mount -o bind /proc "$path1"/proc

# internet

cp -f /etc/hosts "$path1"/etc/hosts

cp -f /etc/resolv.conf "$path1"/etc/resolv.conf

# graphics

xhost +

# make chroot

chroot $path1

--

You should log in as root to your "chroot jail".

In the chroot shell type

export DISPLAY=:0.0

Then, you could install and use xcrysden, or whatever you want.

Way #2) Use remote computer. The simplest solution is to pass small files and view them via Teamviewer.

if you have the *.cif file you can generate PWscf input file using "cif2qe.sh" tool. This tool is located in "QuantumEspresso/q-e-qe-6.3.rc1/PW/tools/"

Running this tool in terminal by "./cif2qe.sh(hit space bar)*.cif" it will generate the necessary input file along with ibrav and atom positions require for pw.x.

Dear friend,

Angshuman Gupta,

you say:

" it will generate the necessary input file".

Not necessarily! :)

http://lists.quantum-espresso.org/pipermail/users/2016-April/034892.html

Physics is fun! :)

https://www.youtube.com/watch?v=7dTnBAlmStE

https://www.youtube.com/watch?v=zd5VQa25MXw

Best regards,

A.S. Gruia

Here is an excellent tutorial on density functional theory using quantum espresso with input and output files included.

https://medium.com/modern-physics/tutorial-setting-up-and-running-a-dft-calculation-using-quantum-espresso-d7a2428495bd

Thank you all for your immense help

You can follow

https://www.youtube.com/watch?v=zd5VQa25MXw

can anyone tell me the cif file of CoYCrAl in quatum espresso?