Phonon linewidth of B-doped diamond
Posted: Thu Jan 25, 2018 4:49 am
Dear All,
I am new to EPW and has done the first tutorial about B-doped diamond on the EPW official website (http://epw.org.uk/Documentation/B-dopedDiamond). After browsing the internet and deep discussion with my colleagues, I still have some questions related to the results of B-doped diamond and usage of EPW and I summarize them here:
1. I calculated the phonon dynamical matrix, frequencies and the variations of the self-consistent potential with the provided inputs. When executing ph.x, I found there were some notices like "possibly too few bands at point 1/2/3/4 ..." Does this affect the result? This also occurred when executing epw.x. I have drawn my result for phonon dispersion of B-doped diamond as shown below. There is a difference compared to the figure. 2 in that example and I circled it in red out for your check.
2. Why should we generate the kpoints manually for nscf calculation in order to set all positive kponts and for which purpose? And why we do not need to set all positive q-points for phonon DFPT calculation because they are also the inputs for EPW?
3. In the tutorial of B-doped diamond, we use pp.py to obtain the irreducible qpoints for EPW. So does EPW also only read the wavefunction and electron eigenvalues in the irreducible BZ?
4. My computed phonon linewidths for B-doped diamond are different from the file linewith.phself_50 in that example reference. Why does this happen because I only run the input provided by the tutorial with any modification?
5. In the EPW inputs descriptions, there are the cards labled 'wmax', 'wmin' and 'nw'. I found here (http://epw.org.uk/uploads/Ford2/module/control_epw.html) that tells these cards are only used when the strict selection rule is applied. However, I notice that the card 'epstrict' has been removed after EPW v4.1. So does these cards 'wmax', 'wmin' and 'nw' still work? Last, what is the function of 'nw' because I cannot understand the description of it " Number of bins for frequency scan in \delta( e_k - e_k+q - w)"?
6. What is the content of output file crystal.fml and demedata.fmt? Also, could anyone tell me the difference bwtween file decay.epmat_wanep and decay.epwane because I think decay.epmat_wanep has listed all the maximum e-ph matrix element in MLWF representation as a function of Re and Rp? Does decay.epwane contain any information that are not in decay.epmat_wanep?
7. I want to use EPW to calculate phonon relaxation time due to e-ph interactions? After I get the linewidth.phself (in unit of mev), in the Migdal approximation, I need to multiply them with the factor of 2/hbar to get relaxation time in unit of like ps. Am I right? Is there any difference between when I directly apply Fermi's golden rule and use EPW to get relaxation time of phonon due to e-ph interaction? From the formula, I think they are equivalent.
8. From the literature, it is said that phonon and electron linewidth are very sensitive to the choice of degaussw that is implemented in enery conservation delta function. How could I say that my results are well convergent with respect to degaussw, kmesh and qmesh? So I think tetrahedron method for 3D system and triangle method for 2D system will has much better accuracy.
Thanks for your time.
Philo
Tsinghua University
I am new to EPW and has done the first tutorial about B-doped diamond on the EPW official website (http://epw.org.uk/Documentation/B-dopedDiamond). After browsing the internet and deep discussion with my colleagues, I still have some questions related to the results of B-doped diamond and usage of EPW and I summarize them here:
1. I calculated the phonon dynamical matrix, frequencies and the variations of the self-consistent potential with the provided inputs. When executing ph.x, I found there were some notices like "possibly too few bands at point 1/2/3/4 ..." Does this affect the result? This also occurred when executing epw.x. I have drawn my result for phonon dispersion of B-doped diamond as shown below. There is a difference compared to the figure. 2 in that example and I circled it in red out for your check.
2. Why should we generate the kpoints manually for nscf calculation in order to set all positive kponts and for which purpose? And why we do not need to set all positive q-points for phonon DFPT calculation because they are also the inputs for EPW?
3. In the tutorial of B-doped diamond, we use pp.py to obtain the irreducible qpoints for EPW. So does EPW also only read the wavefunction and electron eigenvalues in the irreducible BZ?
4. My computed phonon linewidths for B-doped diamond are different from the file linewith.phself_50 in that example reference. Why does this happen because I only run the input provided by the tutorial with any modification?
5. In the EPW inputs descriptions, there are the cards labled 'wmax', 'wmin' and 'nw'. I found here (http://epw.org.uk/uploads/Ford2/module/control_epw.html) that tells these cards are only used when the strict selection rule is applied. However, I notice that the card 'epstrict' has been removed after EPW v4.1. So does these cards 'wmax', 'wmin' and 'nw' still work? Last, what is the function of 'nw' because I cannot understand the description of it " Number of bins for frequency scan in \delta( e_k - e_k+q - w)"?
6. What is the content of output file crystal.fml and demedata.fmt? Also, could anyone tell me the difference bwtween file decay.epmat_wanep and decay.epwane because I think decay.epmat_wanep has listed all the maximum e-ph matrix element in MLWF representation as a function of Re and Rp? Does decay.epwane contain any information that are not in decay.epmat_wanep?
7. I want to use EPW to calculate phonon relaxation time due to e-ph interactions? After I get the linewidth.phself (in unit of mev), in the Migdal approximation, I need to multiply them with the factor of 2/hbar to get relaxation time in unit of like ps. Am I right? Is there any difference between when I directly apply Fermi's golden rule and use EPW to get relaxation time of phonon due to e-ph interaction? From the formula, I think they are equivalent.
8. From the literature, it is said that phonon and electron linewidth are very sensitive to the choice of degaussw that is implemented in enery conservation delta function. How could I say that my results are well convergent with respect to degaussw, kmesh and qmesh? So I think tetrahedron method for 3D system and triangle method for 2D system will has much better accuracy.
Thanks for your time.
Philo
Tsinghua University