About the transmission spectrum of carbon nanotube

[chp]

Hi:

    I performed some test calculations on pristine (4,4) carbon nanotube. It is recognized that the (4,4) tube is metallic, but in the obtained transmission spectrum there is a deep trench at the Fermi level (please see the attach picture), which is different from other theoretical results. Moreover, the transmission coefficients (0.00815129) is not consistent with the transmission eigenvalues (1.00005082 , 0.99371959) at the energy of 0.0*electronVolt. I doubt if I set any inappropriately parameters (please see the attached file).
 
    Thanks in advance.

[zh]

This problem may be due to the numerical accuracy.  Please try to tune the value of "green_function_infinitesimal" in "calculateTransmissionSpectrum()" and it may be helpful to solve the problem.

[Anders Blom]

Your electrode is rather shallow (short in Z). It's not dramatically short for DFT (it would be for Huckel), but there are some small matrix elements that get truncated. So, you may want to increase it (double it). The effect of this is probably small once you introduce scattering, but it comes out in the perfect system, because the truncated matrix elements makes the system not 100% perfect, i.e. you introduce artificial scattering.

You may see the discussion at http://quantumwise.com/publications/tutorials/mini-tutorials/99 for details (they apply to 2008.10 as well).

[chp]

Thank you very much !