Vertical Ionization Potential (VIE) using GW formalism

[Ivan Gladich]

 Dear CP2K community,

I would like to calculate the Vertical Ionization Potential (VIE) of several species, both in the gas phase and in solution. As an initial benchmark, I attempted to compute the VIE of OH⁻ in the gas phase. Its experimental value has been reported as 42.2 kcal/mol (1.83 eV), and similar values have been obtained from SCS-MP2 calculations. (https://pubs.acs.org/doi/10.1021/acs.jpclett.2c01721)

I would like to repeat this calculation at the GW level because, to my understanding (please correct me if I am mistaken), the relation −E_{HOMO} ~ VIE is expected to be particularly accurate at GW level. I have seen a study study performing such calculations at the G₀W₀ level for OH-, although not using CP2K. (https://doi.org/10.1063/1.4887259)

I am attaching my input file. I took inspiration from the CP2K GW tutorial (https://www.cp2k.org/exercises:2017_uzh_cp2k-tutorial:gw). The OH⁻ structure was previously optimized at the PBE-D3/TZV2P-MOLOPT level, and also at the DC-R2SCAN/TZV2P level.

I tested three GW flavours (evGW, G₀W₀, and evGW₀). The closest agreement was obtained with evGW, which gives −E_{HOMO} ≈ VIE ≈ 0.8  eV, still significantly (1 eV smaller) from the reference gas-phase value (1.83 eV). 

Am I missing something in my setup? Any suggestions or recommendations (even of alternative approaches) are welcome.

Thank you very much for your time and help. 

Best, Ivan.