 Hello, first of all, thank you very much for the data set you provided. I don’t understand one thing. The track file of each adsorbent provides the system. txt file, and the reference_energy in it is inconsistent with the minimum energy in the track file. Should I refer to the reference_energy provided in the system. txt file?

Hi -

For each extxyz file there exists a corresponding text file with reference energies for each frame. The energy in the extxyz files are not supposed to match with what’s in the text file. The reference energy in the text file was computed offline via: relaxed_slab_energy + adsorbate_gas_phase_energy and is meant to be subtracted from the raw energies to arrive at the adsorption energy.

Let me know if this doesn’t address your concern and maybe you can provide more details or a specific example as to what you’re referring to.

Thank you very much for your reply, which has solved my doubts. I need to trouble you again .I have found the carbon atom energy of C.N.H.O in SI of the paper. May I ask, if the adsorbent is a water molecule, is the calculated energy the sum of the energy of two hydrogen atoms and one oxygen atom?

Correct - H2O would be E_gas_H2O = 2*H + O = 2*(-3.477) + (-7.204) = -14.158eV.

• Thank you very much for your help and wish you every success in your work.
1 Like

Excuse me, I want to confirm that the energy of a single water molecule is calculated as E_gas_H2O = 2H + O = 2(-3.477) + (-7.204) = -14.158eV? Or can we use DFT to calculate the energy of water molecules? Because I found that there is an error between the water molecule energy calculated by DFT and the direct calculation of atomic energy.

Correct, the gas-phase energy we use for H2O is -14.158eV. We arrive at this by doing DFT calculations of the following small molecules: H2O, CO, H2, N2. We can do a linear combination of these to arrive at H, C, O, N energies i.e.

E_H = E_H2/2
E_O = E_H2O - 2*E_H
E_C = E_CO - E_O
E_N = E_N2/2

With these energies we can compute the gas phase energy for a given adsorbate in OC20 i.e
CH4 = E_C + 4*E_H. Below is the full derivation, where (*) corresponds to the surface
García-Muelas, R., López, N.

In the case of H2O, because it was used in our reference calculation, 2*E_H + E_O will match the DFT calculation for H2O that we used. Any discrepancy you’re seeing is likely a result in DFT settings. In general, however, the referencing used doesn’t matter so long you’re consistent throughout the dataset. When adsorption energies are computed/reported, you must pay attention to the gas references used since these will impact the results. If we wanted to compare results from one reference scheme to another, one needs to simply unreference their data and reference to the molecules of the other dataset for an accurate comparison.

More information can be found in Supplementary Note 5 in this paper.

Hope this helps!