Were the low-index rutile surfaces fully relaxed first? If yes, were the surfaces reconstructed? It was mentioned above "transition metal clusters" but I don't know if they are 3d, 4d ... transition metals? Because as we go from the top left to the bottom right corner of the d series electronegativities generally increase.

Dear Hashem,

Thank you so much for replying. The transition metal clusters are made up of iron, cobalt, and ruthenium.

The systems were geometry optimized before running for Bader charge analysis. Is there any method to do the surface reconstructions or just visual inspection before and after geometry optimization ?

Dear Suraj,

You're welcome! Before getting to the surface properties, do you get the band structure of bulk Rutile accurately with the DFT method you are using? The band alignment of rutile and the transition metals might be important in order to study the charge transfer...

About surface reconstruction, you want to make sure that there is no dangling bonds at the surface. Dangling bonds can make different charge transfer states. A former colleague worked on silicon surfaces had to manually create Si (001)-2×1 surface. Of course Si(111) has a much more complex reconstruction.

Dear Dr. Hashemi,

In order to validate my DFT method on bulk rutile, i just calculated the lattice parameters and compared it with the experimental values. I did not do the band structure calculations.

While reading articles, it is mentioned that in most of the cases, the charge are transferred from the cluster to the oxide support, thus making cluster positively charged.

I see surface reconstruction in surface, leaving no dangling bonds at the surface and oxygen atoms at the surface changing it's position.

I am also studying CO adsorption on these clusters. My calculation shows that there is good CO adsorption on all the clusters despite of the surface-planes and charge accumulation on the clusters.

Dear Suraj,

I suspect the atoms of the transition metal clusters will form chemical bonds with O sites of rutile surface. So band alignment might be important. Also, you may end up with wrong adsorption sites if the band gap of rutile is not accurately calculated.

Charge transfer from the cluster to the rutile surface kind of make sense(simply because of the effective electronegativity difference between the involved species).

It's great that you do get the surface reconstruction!

If the DFT software you are using is vasp, you may find the charge density difference approach (as explained in the attached document) useful.

Dear Dr. Hashemi,

Thank you so much for replying. Here is the charge density difference pic from one of my system.

I was also reading an article which mentions this " It is believed that the TiO2 support not only helps to disperse the Au atoms but also helps to activate Au particles due to the SMSI because electrons may be transferred from the cation centers of the support (e.g., Ti3+ for TiO2) to the metal particles.(17-19) In the case of CO oxidation, the TiO2 support enhances electron transfer from Au to the antibonding states of O2, and thus, the reaction barriers for O2 dissociation and CO oxidation are reduced dramatically at the Au/TiO2 interface. "

In this scenario, the cluster is made up of gold atoms.

Dear Suraj,

Nice figure!

The nature of charge transfer at the interfaces can be further illustrated by plotting the plane-averaged charge density difference along the perpendicular direction to the interface across the TM-rutile system by the charge accumulation

and depletion densities. This can provide a quantitative picture of the charge redistribution at the interface. The plane-averaged charge density difference can be written as ∆σ=σ_slab+TM −σ_slab −σ_TM . Here, σ slab+TM is the plane-averaged density of the combined rutile slab-TM system, and σ_slab and σ_TM are the plane-averaged densities of the rutile slab and TM cluster, respectively, which will calculated by freezing the atomic positions of the respective components in the combined system. Evaluation of the integrals of ∆σ (accumulations) will give you a quantitative suggestion about the charge transfer.

Dear Dr. Hashemi,

If you are suggesting me to calculate the charge densities of the respective systems by freezing all the atomic coordinates of the respective components, then the picture mentioned above resemble the same thing.

Because the CHGCAR files generated to calculate the charge density difference for the whole system was done with single point energy calculation ( i believe which is same as freezing atomic coordinates ).

These are the steps:-

1) Geometry optimization but CHGCAR was not generated.

2) Single point energy calculation to generate CHGCAR to calculate bader charges.

Dear Suraj,

I am sure your calculation is accurately done. The plane-averaged charge density (∆σ) is a two dimensional function and the evaluation of the integrals of ∆σ will give you a quantitative measure (rather than qualitative figure above) of the charge transfer. That was my main point.

Dear Dr. Hashemi,

Thank you so much for replying. I am sorry, i did not see the word "quantitative".

So in order to calculate the plane average charge density,

1) I need to fix atoms of all respective relaxed structures i.e. slab+TM , slab, and TM

2) Simulate it to generate the CHGCAR files

3) Use ∆σ=σ_slab+TM −σ_slab −σ_TM

Is that what you mean ?

Dear Suraj,

A quick search resulted the attached papers. Please look for "plane-averaged charge density" in them.