Issue with Phonon Thermal Conductivity Calculation of Monolayer Graphene using Elphbolt

[Zouzw76]

Dear Nakib,
I hope this message finds you well.
I have successfully reproduced the results on electron drag effect on thermal conductivity of the monolayer MoS2 from Nao Lett. 2024,24,8143-8150 .
However, I encountered an issue when calculating the phonon thermal conductivity of monolayer graphene using elphbolt to solve the phonon BTE only. The results showed oscillatory divergence.
For comparison, I calculated the thermal conductivity of monolayer graphene using ShengBTE with the same espresso.ifc2 and FORCE_CONSTANTS_3RD files as input. These calculations produced reasonable values—3031 W/(m·K) for the iterative solution and 413.4 W/(m·K) for RTA.
The input.nml file for elphbolt is as follows:
input.txt
The output file from this calculation is:
pBTE.txt
The input file for shengBTE is:
CONTROL.txt
Could you please suggest how I might resolve the issue I encountered with Elphbolt? Any advice or suggestions would be greatly appreciated.
Thank you for your time and help.
Best regards,
Ziwen

[nakib]

Dear Ziwen,

Thank you for your email. Graphene is a challenging material. From my understanding, it is debatable whether graphene's thermal conductivity is finite through a full solution of the 3-phonon limited BTE. The issue is related to how much does the low energy ZA phonons contribute to the thermal conductivity. As for the difference between ShengBTE and elphbolt, this difference at the RTA could be due to the different treatment of the delta function -- Sheng uses adaptive Gaussian and elphbolt uses linear triangular. You can start by comparing the scattering rates between the two codes. It is likely that there is only a significant difference for the low energy ZA phonons. Another point is whether the thermal conductivity is converged with respect to the q-mesh. There is some evidence that suggests that it will never converge. Another thing to check is whether your Sheng scattering rates/thermal conductivities are converged with respect to both the q-mesh and the scalebroad factor.

Best regards,
Nakib

[balabala12s]

Hi @Zouzw76 ,

Sorry to bother you. I am trying to reproduce the monolayer MoS₂ results in Nano Lett. 2024, 24, 8143–8150 using ElphBolt.

I found that my calculated phonon thermal conductivity is somewhat larger than the value reported in the paper. I suspect that the difference may come from the third-order force constants.

May I ask how you set the electronic k-point mesh for the displaced supercell SCF calculations when generating the third-order force constants? For example, after using a 6 × 6 × 1 supercell, did you use Gamma-only, 2 × 2 × 1, 3 × 3 × 1, or another k-point mesh?

Due to limited computational resources, I only performed Gamma-point SCF calculations for the different displaced configurations of the 6 × 6 × 1 supercell. I am wondering whether this may lead to an overestimated thermal conductivity.

If possible, would you be willing to share the third-order force-constant file for monolayer MoS₂, such as FORCE_CONSTANTS_3RD?

Thank you very much for your time and help.

Best,
Lee