monolayer sheet

[ramkrishna]

How can I make MoS2 mono-layer sheet for device calculation?? 

[Anders Blom]

This would follow basically the same steps as for graphene, as described in many of the tutorials. Just keep in mind the transport direction must be Z, so if your cell is oriented otherwise, you need to rotate it (easiest to just permute the axes in the input file, probably).

[Nordland]

Since you ask so nicely, I have found my old MoS2 system

Code

# -------------------------------------------------------------
# Bulk configuration
# -------------------------------------------------------------

# Set up lattice
vector_a = [3.16000144132, 0.0, 0.0]*Angstrom
vector_b = [-1.58000072066, 2.73664152418, 0.0]*Angstrom
vector_c = [0.0, 0.0, 18.4500082777]*Angstrom
lattice = UnitCell(vector_a, vector_b, vector_c)

# Define elements
elements = [Sulfur, Molybdenum, Sulfur]

# Define coordinates
fractional_coordinates = [[ 0.66666667,  0.33333333,  0.42208282],
                          [ 0.33333333,  0.66666667,  0.49995115],
                          [ 0.66666667,  0.33333333,  0.57780471]]

# Set up configuration
bulk_configuration = BulkConfiguration(
    bravais_lattice=lattice,
    elements=elements,
    fractional_coordinates=fractional_coordinates
    )

[ramkrishna]

Thank you for your kind reply. I have seen from your code that all the angles are not 90 degree ( gamma = 120 degree) because of the hexagonal structure of MoS2 but for device calculation I have seen from the tutorials that the angles of all the electrodes are 90 degree. So , is it necessary to make all the angles 90 degree for MoS2 system for two/ three probe calculation? If so, then how can I make the system for device calculation or how can I make the unit cell for device calculation?

[kstokbro]

transport is in C direction which is perpendicular, so this is not a problem for this system

[Anders Blom]

I guess you are interested in studying MoS2 where the transport is in the plane, not across the van der Waals gaps; therefore, you need an orthorhomic supercell of the hexagonal lattice. For your convenience:

Code: python

# -------------------------------------------------------------
# Bulk configuration
# -------------------------------------------------------------

# Set up lattice
vector_a = [5.47328304836, 0.0, 0.0]*Angstrom
vector_b = [0.0, 18.4500082777, 0.0]*Angstrom
vector_c = [1.43884903991e-12, 0.0, 3.16000144132]*Angstrom
lattice = UnitCell(vector_a, vector_b, vector_c)

# Define elements
elements = [Sulfur, Molybdenum, Sulfur, Sulfur, Molybdenum, Sulfur]

# Define coordinates
cartesian_coordinates = [[  1.82442769e+00,   7.78743152e+00,  -1.58000071e-08],
                         [  9.12213832e-01,   9.22410286e+00,   1.58000074e+00],
                         [  1.82442769e+00,   1.06605017e+01,  -1.58000071e-08],
                         [  4.56106922e+00,   7.78743152e+00,   1.58000070e+00],
                         [  3.64885536e+00,   9.22410286e+00,   1.58000071e-08],
                         [  4.56106922e+00,   1.06605017e+01,   1.58000070e+00]]*Angstrom

# Set up configuration
bulk_configuration = BulkConfiguration(
    bravais_lattice=lattice,
    elements=elements,
    cartesian_coordinates=cartesian_coordinates
    )

The perpendicular cell (vacuum) is quite large, you can probably reduce it quite a bit. It may be that you need to consider the orientation a bit, esp. if you want to make the structure finite across, i.e. a ribbon. Cf. graphene where you have metallic/semiconducting behavior depending on if the edge is zigzag or armchair; similar (but not the same) effects would occur here also, that is, if the transport occurs in the "long" or "short" directions of this cell. But that transformation is quite trivial, you would just permute the axes (and coordinates, of course). So, you have some interesting work ahead of you!

[ramkrishna]

Thank you very much Mr. Blom, that is the thing what I wanted to know .

Regards
Ramkrishna

[ramkrishna]

Dear sir,
        I am trying to investigate the band structure of Armchair and Zigzag MoS2 monolayer nanoribbon using Extended Huckel method (considering 3p rule for armchair ribbon like graphene, p is int.). I have constructed these structures by using the the above supercell script. Please find the attachments for the files and the figures. Now, it is known that zigzag structures are always metallic and armchairs are semiconducting in nature but from the band structure calculation I have not got these characteristics. In case of zigzag ribbon, the band structure shows like a semiconducting nature but it should not be. Again in case of armchair, the band structure shows that the band gap is direct but not exactly at gamma whereas in literatures it is at gamma (please check this links http://epjb.edpsciences.org/index.php?option=com_article&access=standard&Itemid=129&url=/articles/epjb/abs/2012/01/b110456/b110456.html , http://pubs.acs.org/doi/abs/10.1021/ja805545x). The band gap (which is also less than the DFT value which is mentioned in the literatures) variation is also not correct with integer p. I have tried with DFT-LDA also but find same  problem. I can't understand whether this problem is because of the ribbon structure or something else. Please help me. I am waiting for your kind reply.   

Regards
Ramkrishna

[Anders Blom]

First of all I note that the calculations in at least one of the papers you quote is spin-polarized, while yours is not. That can certainly make a big difference.

Second, you are doing the computation on a supercell, the structure is repeated more than once in C. This will cause zone folding and makes it harder to interpret the results - not to mention the fact that the calculation takes much longer.

[ramkrishna]

dear Sir,
       I am trying to do the calculation by inclusion of spin-polarization now (using the help of this tutorial http://quantumwise.com/documents/tutorials/latest/GrapheneBloch/index.html/chap.spin.html) for only one repetition along C direction of MoS2 ribbon. According to this tutorial, I have to specify the initial spins by using user spin or random spin but I don't know the proper way to specify these spins for MoS2 ribbon.

If I want to use user spin for MoS2 ribbon, then will it be same like graphene? [I mean, "Set the Initial state type to "User spin", and then under Spin, set the default spin of both carbon and hydrogen to 0 --> the two upper carbon atoms (number 0,7 in the list) and set their spin values in the table to 1.0. Repeat this procedure for the two lower carbon atoms (number 1,2 in the list) setting their initial values to -1.0. The hydrogen atoms and the four middle carbon atoms can be left with spin values equal to 0.0"]

Otherwise, if I want to specify the random spin, then if I give the default spin (or spin relative) 0 for both Mo and S, then all the elements have taken 0 spin (if 0, 1 then all are 0 and 1), so what will be the steps to specify the spins for this case?

One more thing I want to know, in the paper http://pubs.acs.org/doi/abs/10.1021/ja805545x it is mentioned that spin-polarized total energies are less favorable than the spin-unpolarized one for armchair MoS2 nanoribbons. So, I think, my spin-unpolarized calculation should give more accurate band structure (as well as band gap value) but it is not happening for my system. Please find the attachment for the file. I am not able to be sure for this calculation. Please help me.

Regards
Ramkrishna

[Anders Blom]

If you use VNL, and choose Random Spin, then the values you set are not the actual initial spins, but as the column title of table says the "Range" of spin values, i.e. if you set the range to 0.5, the initial spin will be set to a random number in the interval (-0.5,0.5). Of course, if you set the range to 0, then the initial spin will be set to zero.

About your second question, I don't have hands-on experience with this material myself to know which paper/calculation is more accurate, so had better try many things and be careful. Your parameters are fine, i.e. k-point sampling and mesh cut-off.

[ramkrishna]

Dear Sir,
        Can I use Huckel method for the  spin-polarized calculation? or is this possible only by using DFT-LSDA (or DFT-SGGA) method?

Regards
Ramkrishna

[Anders Blom]

As of 11.8 you can do spin-polarized calculations with all ATK-SE methods.

[ramkrishna]

Dear Sir,
       You have mentioned before that that if I will give repetition more than once along C then due to the zone folding it will be difficult to interpret the band structure, so it will be better if I do for the one repetition along C, but for the case of having finite width and length of a nanoribbon or nanosheet how can I then be sure about the calculation?
Regards
Ramkrishna

[Anders Blom]

The finite width is not a problem, that's the direction where you have vacuum, so you just make it as wide as you want.

A nanoribbon doesn't typically have a finite length, at least on the length scale where you do the simulations (a few nm). But if it is, then you just make it as long as you want - but then not periodic in this direction, but with vacuum added. In this case it is however a molecule, and can be computed as such - and you don't have a band structure, a finite system only has discrete energy levels.

If you want to strain the structure, you strain the cell, normally keeping the fractional coordinates at first, and then possibly relaxing them (but the symmetry rarely changes much with strain). There is no reason to make a repetition to add strain.

The only situation in which it makes sense to have the structure repeated (besides transport simulations) is if you want to add defects, impurities, etc.