QuantumATK Forum
QuantumATK => General Questions and Answers => Topic started by: run on June 22, 2011, 08:51
-
Recently, I calculate a spin-polarized system. When the bias is 1.0 V, the spin up and spin down current is 5.0551131779e-12 and 5.41429995254e-08, respectively. However, the simultaneously
calculated transmission coefficient for spin up and spin down is zero whthin bias window. Why?
# Energy (eV) Spin-Up Coefficient Spin-Down Coefficient
-1.00 0.0001 0.0007
-0.98 0.0001 0.0002
-0.96 0.0000 0.0001
-0.94 0.0000 0.0001
-0.92 0.0000 0.0000
-0.90 0.0000 0.0000
-0.88 0.0000 0.0000
-0.86 0.0000 0.0000
-0.84 0.0000 0.0000
-0.82 0.0000 0.0000
-0.80 0.0000 0.0000
-0.78 0.0000 0.0000
-0.76 0.0000 0.0000
-0.74 0.0000 0.0000
-0.72 0.0000 0.0000
-0.70 0.0000 0.0000
-0.68 0.0000 0.0000
-0.66 0.0000 0.0000
-0.64 0.0000 0.0000
-0.62 0.0000 0.0000
-0.60 0.0000 0.0000
-0.58 0.0000 0.0000
-0.56 0.0000 0.0000
-0.54 0.0000 0.0000
-0.52 0.0000 0.0000
-0.50 0.0000 0.0000
-0.48 0.0000 0.0000
-0.46 0.0000 0.0000
-0.44 0.0000 0.0000
-0.42 0.0000 0.0000
-0.40 0.0000 0.0000
-0.38 0.0426 0.0000
-0.36 0.1117 0.0000
-0.34 0.1728 0.0000
-0.32 0.2361 0.0000
-0.30 0.3033 0.0000
-0.28 0.3752 0.0000
-0.26 0.4511 0.0000
-0.24 0.5724 0.0000
-0.22 0.6425 0.0000
-0.20 0.0010 0.0000
-0.18 0.0000 0.0000
-0.16 0.0000 0.0000
-0.14 0.0000 0.0000
-0.12 0.0000 0.0000
-0.10 0.0000 0.0000
-0.08 0.0000 0.0000
-0.06 0.0000 0.0000
-0.04 0.0000 0.0000
-0.02 0.0000 0.0000
0.00 0.0000 0.0000
0.02 0.0000 0.0000
0.04 0.0000 0.0000
0.06 0.0000 0.0000
0.08 0.0000 0.0000
0.10 0.0000 0.0000
0.12 0.0000 0.7371
0.14 0.0000 0.5006
0.16 0.0000 0.3802
0.18 0.0000 0.2697
0.20 0.0000 0.1390
0.22 0.0000 0.0000
0.24 0.0000 0.0000
0.26 0.0000 0.0000
0.28 0.0000 0.0000
0.30 0.0000 0.0000
0.32 0.0000 0.0000
0.34 0.0000 0.0000
0.36 0.0000 0.0000
0.38 0.0000 0.0000
0.40 0.0000 0.0000
0.42 0.0000 0.0000
0.44 0.0000 0.0000
0.46 0.0000 0.0000
0.48 0.0000 0.0000
0.50 0.0000 0.0000
0.52 0.0000 0.0000
0.54 0.0000 0.0000
0.56 0.0000 0.0000
0.58 0.0000 0.0000
0.60 0.0000 0.0000
0.62 0.0000 0.0000
0.64 0.0000 0.0000
0.66 0.0000 0.0000
0.68 0.0000 0.0000
0.70 0.0000 0.0000
0.72 0.0000 0.0000
0.74 0.0000 0.0000
0.76 0.0000 0.0000
0.78 0.0000 0.0000
0.80 0.0001 0.0000
0.82 0.0001 0.0000
0.84 0.0001 0.0001
0.86 0.0003 0.0001
0.88 0.6576 0.0001
0.90 0.8300 0.0002
0.92 0.8966 0.0047
0.94 0.9314 0.8496
0.96 0.9521 0.9304
0.98 0.9653 0.9609
1.00 0.9742 0.9756
-
In previos calculation, we also meet similar question. For example, when I calculate a two probe system, the left and right electrtodes are same semiconductor electrodes. The band gap is 0.1 eV. Theoretically, the threshold voltage should be 0.2 V. However, when the bias is set to 0.01 V, the current still can reach 1.19036166221e-07 A. It seems that I-V current cannot represent the semiconductor performance of the system. In calculation, the basis set is DZP.
-
When the bias is 1.0 V
is this a typo and is the actual bias 0.1V ?
the simultaneously calculated transmission coefficient for spin up and spin down is zero whthin bias window.
More precisely the transmission coefficients aren't zero but very small finite value numerically.
The output just represents that the transmission coefficients within the bias window are smaller than 5.e-5.
The band gap is 0.1 eV. Theoretically, the threshold voltage should be 0.2 V.
I think the threshold voltage should be 0.1 V in this case.
However, when the bias is set to 0.01 V, the current still can reach 1.19036166221e-07 A.
I'm afraid I can't offer the appropriate answer without the script you used.
But it would be derived from inaccuracy of self energies or that the fermi level does not converge enough.
-
I am sorry that it is a typo. The bias is 0.1 V. If I will publish this result in Scientific journal,what can I do with this situation?
For the second question,the threshold voltage should be 0.1 V when the band gap is 0.1 V. In my modle, the system is quasi-one-dimensional structure. Therefore, in ATK 2008.10, I only change the K point into 1, 1, 100 in the two probe system. Other parameters are the default. Should I change some parameters? However, when the band gap is much lager than 0.1 eV, similar situation can not be found. The I-V curve can reprensent the semiconductor performance of the system.
-
what can I do with this situation?
You can obtain numerically precise values with the method illustrated in the manual:
http://www.quantumwise.com/documents/manuals/ATK-2008.10/ref.calculatetransmissionspectrum.html
Should I change some parameters?
I think k-point sampling is enough.
Firstly, you should check the converged mulliken population in order to confirm that the SCF converged at the appropriate state.
And if the mulliken population look strange, it may be due to too short electrodes or too short central region.
In addition, of course you should check again whether or not the bias is 0.01V.
-
Just to clarify, the reason you see zero transmission in the log file is because it's printed with (too) few decimals. The number is non-zero, but smaller than 1e-5. So the values you are getting for the current are correct, but to get the transmission spectrum with all decimals, for plotting, follow the link already posted by nori above.