Questions regarding H2 adsorption onto nanolayer

[ryan_brady]

Hi all,

I'm quite new to the world of computational chem. I'm working on an independent project as part of my final undergrad unit. I'm studying the adsorption of H2 onto a BeN nanosurface, but have run into some issues with interpreting some of my initial findings, or even whether I'm running the right computations. Unfortunately, my supervisor is on extended leave and it not able to answer my emails at the moment, so I'm finding answers hard to come by.

I'm using VASP and ran a DFT relaxation just fine on the underlying structure. Then I ran some further DFT relaxation tests with H2 at different positions above the sheet. They all worked fine and I was able to determine the apparently most favourable position. However, I am running into issues now with actually trying to investigate the adsorption of multiple H2 molecules. I tried to follow this tutorial on the VASP website for H2O onto TiO2. It seemed that once I had the relaxed surface geometry and a place to put the H2 molecule(s), then you run an MD computation. I modified my INCAR file to match their additions:

IBRION = 0
POTIM = 1.
MDALGO = 1
PROB = 0.9
TEBEG = 10; TEEND = 10

However, in their INCAR, ISMEAR is meantioned twice. As ISMEAR=2 and ISMEAR=0. Is this needed? I would imagine that VASP would ignore one or the other. Then I wrote an ICONST file to maintain the bond length of the H2 molecules.

However, the results were...odd.

With only 1 H2 molecule the H2 starts at ~2.5A and then retreats to ~4.5A.
PRE:

1H2_pre.png

POST:

1H2_post.png

When I have 2 H2 molecules at ~2.2A in different (what I believed to be favourable from earlier DFT calculations) spots, one molecule retreats to ~3A and the other goes through the layer to the other side at ~2.6A.

2H2_post.png

Additionally, the geometry of the BeN2 layer changes for both computations.

How am I supposed to interpret this? Does this mean that the surface is not good for adsorption of H2, or am I doing the wrong type of test? I checked the OUTCAR and it doesn't look like the computations hit the minimum. Alternatively, are there any other good tutorials on how to run adsorption tests for multiple molecules?

[jonathan_lahnsteiner2]

I moved your post to users for users where you can discuss your issues with other users.

About your comment

When I have 2 H2 molecules at ~2.2A in different (what I believed to be favourable from earlier DFT calculations) spots, one molecule retreats to ~3A and the other goes through the layer to the other side at ~2.6A.

I don't think the molecule is going trough the layer, but it switches to the other side of the simulation box because of
the periodic boundary conditions. The reason for this might be the repulsion between the 2 H2 molecules. Like this, they might get their own absorption sites without being sterically hindered by the other molecule.


All the best Jonathan

[ryan_brady]

I don't think the molecule is going trough the layer, but it switches to the other side of the simulation box because of
the periodic boundary conditions. The reason for this might be the repulsion between the 2 H2 molecules. Like this, they might get their own absorption sites without being sterically hindered by the other molecule.

Thank you for replying Jonathan. That is a good thought - I had been so focussed on the a/b axes and completely forgot about c.

[alex]

Hello ryan_brady,

Jonathan is probably right about H2 switching sides due to large forces in the beginning.

One other point: if it comes to this kind of interaction, you will have to look into dispersion interactions (van der Waals) on top of pure DFT for starters (or even better methods). Considering the circumstances of your work I'd give the vdW-correction as built-in in VASP a go.
Without that, the potential energy surface is very(!) shallow and your optimization will not find a minimum unless you go extremely tight in criteria. However, disappointingly you'll find your H2 too far away for a chemist's taste (but at least you got your setup right :-) ).

Switch on vdW correction and have a look.

HtH

alex