I tried the examples of Al-Eliashberg and Nb-superconductor, and the calculated e-ph coupling constant lambda is too high and thus I got a pretty high trasition temperature (Nb is 700.2666061 Kelvin and Al to be 69.37774985). It is not consistent with the results offered in these examples.
The phonon dispersion results are not quite the same as provided, either.
I did not change any input blocks when running the code, should some changes be made there? And are there any other details I should pay attention to?
Hope somebody could help me.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Actually, I am also having exactly same problem; superconducting Tc for Nb and Al are overestimated to be about 700 and 70K, respectively, even though I used the input file provided in the example. Changing other parameters (ngridq/k, rgkmax, angular momentum cutoff, and so on, for example) didn't fix this behavior.
Interestingly, the shape of the phonon DOS and alpha^2 F(omega) are well-reproduced for Al. Logarithmic average frequency omega_{ln} also seems to be reasonable: about 27meV, which is also consistent with the value calculated in the work of Savrasov(http://journals.aps.org/prb/abstract/10.1103/PhysRevB.54.16487). Instead, the value of lambda (el-ph coupling constant provided in the MCMILLAN.OUT file) is 2.81, which is much larger than the value provided in the above reference(lambda=0.44). If I choose lambda to be 0.44 and estimate Tc, then I get a reasonable value of Tc~7K. If the value of alpha^2 F(omega) are reduced to give the reasonable lambda, then one may get the correct Tc in Al.
Any comments are appreciated!
Best regards,
Heung-Sik
Last edit: Heung-Sik Kim 2014-05-20
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Did you check the phonon dispersion curve? These 2 are not exactly the same.
Anyway in my opinion, I think the calculation of electron-phonon coupling constant calculation is wrong. I checked the LAMBDAQ.OUT, the results show some unusual results -- they tend to be too large. And this would lead to the finally wrong results.
Does anyone know how to solve it?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Actually there is still a bug in the electron-phonon coupling routine.
I'll get this fixed ASAP. I'm also currently adding e-p coupling to the linear-response phonon calculations. This will make the calculations much faster.
Regards,
Kay.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thanks for your reply, and also for the new version!
I tried the current 2.3.22 version, and the phonon dispersion and the critical temperature seem much more reasonable now(I get about Tc=8K for Nb). I will come back after I finish test calculations for other materials.
And, if it does not bother you too much, could you let us know a little bit more about the bug in the el-ph routine, so that one may be cautious in trying test calculations? (Instead you may finish and update the fixed code, of course.)
Best,
Heung-Sik
Last edit: Heung-Sik Kim 2014-05-29
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Upon your result, I noticed the real Tc for Nb is 9.2K, and the reference number given by the examples is about 12K, so will this method get a generally higher or lower critical temperature?
Best regards,
Xiaoyu Liu
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I was just about to check the difference between the older versions(1.4.5) and the new ones as I found this bug to be really strange. Glad to hear the news, and I will save the labor. I will check it out ASAP. I am so happy to hear it.
Best,
Xiaoyu Liu
Last edit: wuhuagumu 2014-06-01
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I tried the examples of Al-Eliashberg and Nb-superconductor, and the calculated e-ph coupling constant lambda is too high and thus I got a pretty high trasition temperature (Nb is 700.2666061 Kelvin and Al to be 69.37774985). It is not consistent with the results offered in these examples.
The phonon dispersion results are not quite the same as provided, either.
I did not change any input blocks when running the code, should some changes be made there? And are there any other details I should pay attention to?
Hope somebody could help me.
Dear Elk users and developers,
Actually, I am also having exactly same problem; superconducting Tc for Nb and Al are overestimated to be about 700 and 70K, respectively, even though I used the input file provided in the example. Changing other parameters (ngridq/k, rgkmax, angular momentum cutoff, and so on, for example) didn't fix this behavior.
Interestingly, the shape of the phonon DOS and alpha^2 F(omega) are well-reproduced for Al. Logarithmic average frequency omega_{ln} also seems to be reasonable: about 27meV, which is also consistent with the value calculated in the work of Savrasov(http://journals.aps.org/prb/abstract/10.1103/PhysRevB.54.16487). Instead, the value of lambda (el-ph coupling constant provided in the MCMILLAN.OUT file) is 2.81, which is much larger than the value provided in the above reference(lambda=0.44). If I choose lambda to be 0.44 and estimate Tc, then I get a reasonable value of Tc~7K. If the value of alpha^2 F(omega) are reduced to give the reasonable lambda, then one may get the correct Tc in Al.
Any comments are appreciated!
Best regards,
Heung-Sik
Last edit: Heung-Sik Kim 2014-05-20
Did you check the phonon dispersion curve? These 2 are not exactly the same.
Anyway in my opinion, I think the calculation of electron-phonon coupling constant calculation is wrong. I checked the LAMBDAQ.OUT, the results show some unusual results -- they tend to be too large. And this would lead to the finally wrong results.
Does anyone know how to solve it?
Please try the latest version which includes a patch for electron-phonon coupling.
Best regards,
Lars
Hi All,
Actually there is still a bug in the electron-phonon coupling routine.
I'll get this fixed ASAP. I'm also currently adding e-p coupling to the linear-response phonon calculations. This will make the calculations much faster.
Regards,
Kay.
Dear John,
Thanks for your reply, and also for the new version!
I tried the current 2.3.22 version, and the phonon dispersion and the critical temperature seem much more reasonable now(I get about Tc=8K for Nb). I will come back after I finish test calculations for other materials.
And, if it does not bother you too much, could you let us know a little bit more about the bug in the el-ph routine, so that one may be cautious in trying test calculations? (Instead you may finish and update the fixed code, of course.)
Best,
Heung-Sik
Last edit: Heung-Sik Kim 2014-05-29
Hi,
Upon your result, I noticed the real Tc for Nb is 9.2K, and the reference number given by the examples is about 12K, so will this method get a generally higher or lower critical temperature?
Best regards,
Xiaoyu Liu
Hi, John,
I was just about to check the difference between the older versions(1.4.5) and the new ones as I found this bug to be really strange. Glad to hear the news, and I will save the labor. I will check it out ASAP. I am so happy to hear it.
Best,
Xiaoyu Liu
Last edit: wuhuagumu 2014-06-01