Calcite and Water

[Sunnyvale]

Hello,
    Sorry to post a new thread, but I think the older thread was locked. I am trying to find the adsorption energy for calcite and water with some substitutions. I firstly want to optimize the water with calcite surface. I have tried to hold the water rigid and calcite fixed, and vice versa, and tried to hold them both rigid but nothing is happening.  If you can take a look at the attached, I'd appreciate it. Thanks!

[zh]

For the geometry optimization of the adsorption of a molecule on a substrate surface, it is usually first to optimize the substrate. In this step, depending on the slab chosen for the substrate, some atoms are selected to be fixed,  for example, the atoms in the center layers of slab.
The second step, it is to optimize the adsorbed molecule on the optimized substrate. In this case, the atoms far away from the adsorbed molecule could be fixed during the optimization.

The optimization with holding both the water and calcite fixed is not a good way. In this case, it is of course that nothing will happen for the optimization.

As for your constructed structure in the '2calcite and water.py', the slab for calcite surface may be too thin.

[Sunnyvale]

Thanks a lot. I have a cif for calcite, and I cleaved it along 104.  I set the top vacuum at 10 A, the thickness at 15, and bottom vacuum 0. I selected some atoms in the center, held them fixed, and optimized the geometry. I did not get any NC file as an output, and when I opened the py in the Builder, it looked the same as before optimized.

If you could take a look at the attached, I'd appreciate it.

[Jess Wellendorff]

Your script does not define a calculator, that's why no NC file with output is generated. Also, I recommend you constrain all atoms in a few of the bottom slab layers. See attached script.

[Sunnyvale]

Thanks a lot! That's a big help.

[Jin You Lu]

Dear Jess Wellendorff

a) I followed your instruction to study the adsorption energy of H2O on calcite.
Firstly, one water molecule is added to the relaxed calcite surface (bottom two layers fixed).
Then, a MD simulation with the NVT ensemble was done from 300k to 1k. (total time 200ps; time step 1ps). I consider the system achieve equilibrium.
The adsorption energy i got is -0.95eV similar to the result (-1 eV )published in Journal of Physical Chemistry C 118(6):3078-3087  (2014).
calcite+H2O= -40556.574 eV
calcite= -40088.730 eV
H2O=-466.867 eV
dE=-0.95eV.
b) However, if i further do the LBFGS geometry optimization for this equilibrium system.
i got calcite+H2O = -40557.62 eV, which means adsorption energy of -~2eV. I know this step is not suggested in the previous  discussion.
http://quantumwise.com/forum/index.php?topic=4148.msg19013#msg19013
But this way is similar to the tutorial of CO/Pd,
Could you help to advise why i get a different equilibrium state with a lower adsorption energy around -2.0 eV? ( which means more stable ? or something else).

c) In the beginning, i thought this is due to i didn't use proper boundary condition to remove dipole effect. So i use FFT2D with right-sided Dirichlet boundary and left-sided Neumann boundary to calculate the system again.

First, I calculate 4layers calcite again by using  LBFGS geometry optimization and fixing the bottom 2 layers. Same as previous case i did, but with a new boundary condition.
When i got surface relaxed calcite, i added one water  molecule to the surface of calcite.
Then, i directly use LBFGS geometry optimization to simulate the system and got a similar result of -2.4 eV.

The details of total energy for the system is as follows
calcite+H2O= -40551.210 eV
calcite= -40081.950 eV
H2O=-466.867 eV
dE=-2.393eV.

I attached script of case a and c for your reference.
log file it too big to attach it.
Thanks in advance.
 
 

[Ulrik G. Vej-Hansen]

Regarding a), your system is probably not in equilibrium when you have just heated it - you need to run it at constant temperature for a while before you can expect that.

Regarding b), you need to decide if you want your property at a certain temperature, or from the lowest energy configuration. If you optimize the geometry, you will remove the effect of having a finite temperature.

We will get back to you on the other points.

[Ulrik G. Vej-Hansen]

I have looked at your problem in more detail, and I need a little more information about what exactly you are trying to do. Do you want the adsorption energy of water on a calcite surface or are you looking for some other properties?

[Jin You Lu]

Thanks for your reply!
Yes, I am trying to get the absorption energy of water on calcite properly.
However, I got the value around of -2 eV different with -1 eV shown in the literature. (J. Phys. Chem. C 115, 10044 (2011) and J. Phys. Chem. C 118, 3078 (2014).

[Jess Wellendorff]

A few comments to your DFT calculations:
1) Your slab is very close to the bottom of the unit cell. You should probably move it upwards, or simply center it along C.
2) Your k-point sampling could probably be improved. The lattice vector lengths are very different, I suggest a (6,1,1) k-point grid.
3) Important: You are using FHI pseudopotentials with DZP basis set. These are default settings, but may not be sufficient for adsorption energies that are comparable to literature (plane wave?) values. Try the SG15 pseudos instead (with 100 Ha density cut-off energy), may make a significant difference.
4) Please also consider using the counterpoise correction to eliminate the basis-set superposition error, see http://docs.quantumwise.com/tutorials/geometry_optimization/geometry_optimization.html#counterpoise-correction.

[Jin You Lu]

Dear Jess Wellendorff

Thanks for the hightlights in the pseudopotentials use.
I follow procedure 1->2->3 to get adsorption energy (-1.14 eV) of one water molecule on calcite (10.4) surface. After considering BSSE, i get -1.04 eV (very close to -1 eV, shown in JPC paper). But is it necessary to do geometry optimization again for BSSE correction? (It takes time , the energy change during geometry optimization is around -0.03 eV )

for your comment 3, the author use quantumexpresso with PBE and ultrasoft  pseudopotentials to simulate this case. Quantumexpresso use Plane-Wave Self-Consistent Field. I think it is plane wave. Could you share some information on what the difference between  LCAO in ATK and plane wave method is ? Thanks a lot for your comments.

[Petr Khomyakov]

Could you share some information on what the difference between  LCAO in ATK and plane wave method is ?

The main difference between the two methods is that the LCAO and plane wave methods adopt two different basis sets: localized atomic orbitals (LCAO) and plane waves (PWs), respectively, to solve the same Kohn-Sham equations of density functional theory (DFT), which is implemented in the ATK-DFT for the LCAO basis set. However, we are currently working on the development of the ATK-DFT plane-wave based calculator as well. You may consult textbooks on Solid State Physics for more details on these two basic methods used in atomistic modeling of materials and devices.

[Petr Khomyakov]

But is it necessary to do geometry optimization again for BSSE correction?

I guess so.

[Jin You Lu]

Dear Petr Khomyakov
Thanks a lot