QuantumATK Forum
QuantumATK => General Questions and Answers => Topic started by: kaihuang on October 10, 2025, 22:44
-
Dear QuantumATK team,
My research strongly relies on ATK for calculating transport properties, and the LCAO framework in ATK has been extremely helpful and valuable for my studies.
Recently, I calculated the transmission spectrum of monolayer graphene. Since graphene is a well-studied material with a simple structure, it is often used as a benchmark system. However, I encountered a strange problem.
(https://imgur.com/a/34BOTVU)
(Web image here. If it's not shown, find in the attachment.)
As shown in the figures:
- Fig.(a) presents the graphene band structure calculated using VASP along a chosen direction c. b direction has vacuum layer. The lattice has been orthogonalized, showing the well-known two Dirac cones, with the Fermi energy set to zero. ATK band-structure calculation matches this result.
- Fig.(b) and (c) show the bulk (unit cell) non-bias transmission spectrum calculated using ATK 2024. In (b), the x-axis is k-parallel and the y-axis is energy; in (c), the x-axis is energy and the y-axis is transmission. Overall, (b) looks very similar to the band structure in (a) which right, but there is one abnormal transmission channel (green line). This green channel causes the transmission at the Fermi level in (c) to be 1 instead of 0, which is clearly incorrect. The calculation was done using DFT with Dojo pseudopotentials.
To identify the issue, I tried several approaches: Using ATK 2022 instead of 2024; Rewriting the structure with cleaner fractional coordinates (Fig.(d)); Increasing the k-grid to up to 18×1×96 (Fig.(e)). None of these changes fixed the problem.
I also tried using SG15 pseudopotentials (Fig.(f)), it’s even worse that two fake conduction channels appeared (green lines).
I also tried building a bulk-like device by repeating the unit cell several times (Fig.(g)). In that case, the fake channel disappeared, but the transmission became non-integer, and the non-zero region below the Fermi level appeared as discrete lines rather than continuous bands.
Finally, I tested the Slater–Koster (method following the official tutorial (https://docs.quantumatk.com/tutorials/transmission_gr_mos2/transmission_gr_mos2.html). The result was correct and free of fake channels. However, since my main workflow is based on DFT, the S-K method is not suitable for my project. The fact that the S-K method works correctly suggests that my structure setup is fine and the problem might come from the DFT implementation in ATK.
I have attached the calculation files for both the bulk (unit cell) case (corresponds to Fig.(e)) and the bulk-like device case (corresponds to Fig.(g)).
I sincerely hope to receive your advice.
Thank you very much!
Kai
-
Hi kaihuang,
Can you check with these quick fixes for the file D0_BulkDevice.py
1. Increase electrode extension.
2. Use k_point_sampling = MonkhorstPackGrid(na=12, nc=300).
3. Use medium basis set.
-
You may need to go up to 100-300 k-points in the periodic direction to resolve artefacts related to the very specific band structure of graphene.