[Pw_forum] transmission calculation

[Gabriele Sclauzero]

Dear Manoj

Manoj Srivastava wrote:
> Dear Gabriele,
>  Thanks for your response. 
> On Thu, 4 Jun 2009, Gabriele Sclauzero wrote:
> 
>> Quoting Manoj Srivastava <manoj at phys.ufl.edu>:
>>
>>> Dear PWSCF users and developers,
>>>  I have been trying to do the conductance calculation for the twin
>>> boundary of Cu. I compared my results with previously done calculations
>>> and for most K points the transmission coefficients match up very well.
>>> But there are some k points that I get transmission coefficients more than
>>> 1! This is not reasonable, as the maximum value of transmission
>>> coefficient could be 1.I am getting 1.000345, and similar for other k
>>> points.
>> Which k-points? How many channels are there? Please send me the output  
>> file or post the relevant part of it.
> 
>  There are quiet a few k points for which transmission is more than 1. 

I tried to reproduce your calculation and indeed I find transmission coefficients greater 
than 1.
However I think that the problem is in your setup rather than in the program itself.
In order for the scf potential to match smoothly across the border between left lead and 
scattering region, you should include in the leftmost part of the scattering region (SR) a 
certain number of replicas of the left lead. Being your lead quite long, maybe one replica 
is sufficient (you should however check if the complex band structure obtained from the
leftmost replica in the SR agrees sufficiently with those obtained from the left lead),
but in your setup only a part of the left lead is reproduced at the beginning of the SR 
and I think this is not enough.
The same applies to the right lead, which should be included at the rightmost part of the 
SR (there the situation of your setup seems even worst, since only the atom at the border 
matches this needing).


HTH

GS



>  For example- 
>   k(    3) = (   0.2500000   0.0000000), wk =   0.1666667
> 
>  ngper, shell number =          317         133
>  ngper, n2d =          317         317
>  Nchannels of the left tip =            1
>         k1(2pi/a)    k2(2pi/a)       E-Ef (eV)
> 
>    0.0634548   0.0000000   0.0000000
>  Nchannels of the right tip =            1
>         k1(2pi/a)    k2(2pi/a)       E-Ef (eV)
> 
>    0.1266351   0.0000000   0.0000000
>  to transmit
>  T_ij for propagating states:
>     1 -->     1   1.0000806
>                  1.00008
>  Eigenchannel decomposition:
> @    1  0.00000  1.00008
>                       1.00000
> E-Ef(ev), T(x2 spins) =    0.0000000   2.0001613
> 
> Also for another k point - 
> 
>  k(    5) = (   0.0000000  -0.2500000), wk =   0.1666667
> 
>  ngper, shell number =          317         133
>  ngper, n2d =          317         317
>  Nchannels of the left tip =            1
>         k1(2pi/a)    k2(2pi/a)       E-Ef (eV)
> 
>    0.0634548   0.0000000   0.0000000
>  Nchannels of the right tip =            1
>         k1(2pi/a)    k2(2pi/a)       E-Ef (eV)
> 
>    0.1266351   0.0000000   0.0000000
>  to transmit
>  T_ij for propagating states:
>     1 -->     1   1.0000806
>                  1.00008
>  Eigenchannel decomposition:
> @    1  0.00000  1.00008
>                       1.00000
> E-Ef(ev), T(x2 spins) =    0.0000000   2.0001613 
> 
-- 


o ------------------------------------------------ o
| Gabriele Sclauzero, PhD Student                  |
| c/o:   SISSA & CNR-INFM Democritos,              |
|        via Beirut 2-4, 34014 Trieste (Italy)     |
| email: sclauzer at sissa.it                         |
| phone: +39 040 3787 511                          |
| skype: gurlonotturno                             |
o ------------------------------------------------ o