[Pw_forum] Strange phonon branches appear in phonon dispersion
Marcel Mohr
marcel at physik.tu-berlin.de
Thu Dec 13 11:02:58 CET 2007
>>
>> Dear Dr. Stefano and Dr. Mohr,
^^
That is too much honour for me as I am still working on these two letters.
The input below looks ok for me, but two things:
1.) for electronic wavefunctions you choose a 4 4 40 grid. It should be no
difference to a 1 1 40 one (because of 1 dim, there are no bloch waves
perp to your wire axis)
2.) your tr2_ph=1.0d-14 is maybe too small. try to use a value around
1.0d-20
What about the convergence of the other parameters (ecutwfc)
with respect to your phonon frequencies?
Maybe you find some information in the paper verstraete PhysRevB 74 153408
(2006).
Best Marcel
>
>
> Thank you for your comments,
>
> Actually I had successfully done phonon dispersion and EPC for Al bulk
> structure with good results so from now I want to apply for the Al wires
> structures. Firstly I calculate for 2 atom cell of Al with zigzag wires, and
> I got 6 phonon modes along q-points and appear some negative frequencies, so
> I tried to improve my calculation but always appear negative acoustic
> frequencies (optical branch is no problem) and then I did with 4, 6 atom
> cell and of course I got 3-n modes but my results still no good like I
> expect. :(
> To check my set up parameters I calculate the same procedure for Al 1D
> linear chain with 1 and 2 atom cell with hopefully the phonon curves will be
> the same as I studied theoretically in Solid States books, but it was not
> :(. So should my calculation be wrong with this system. Could you give me
> advice about my input parameters? Please keep your eyes my INPUT files (in
> case of 2 atoms in zigzag chain)
>
> Al.scf.fit.in
>
> &SYSTEM
> ibrav = 6,
> celldm(1) = 39.994,
> celldm(3) = 0.199762906,
> nat = 2,
> ntyp = 1,
> ecutwfc = 25.D0 ,
> ecutrho = 300,
> occupations = 'smearing' ,
> degauss = 0.05 ,
> smearing = 'methfessel-paxton' ,
> la2F = .true.
> /
> &ELECTRONS
> conv_thr = 1.D-10 ,
> mixing_beta = 0.7D0 ,
> /
> ATOMIC_SPECIES
> Al 26.98000 Al.vbc.UPF
> ATOMIC_POSITIONS angstrom
> Al 0.000000000 1.517315647 2.138739361
> Al 0.000000000 0.116702849 0.023699809
> K_POINTS automatic
> 4 4 40 0 0 0
>
> Al.scf.in
>
> &SYSTEM
> ibrav = 6,
> celldm(1) = 39.994,
> celldm(3) = 0.199762906,
> nat = 2,
> ntyp = 1,
> ecutwfc = 25.D0 ,
> ecutrho = 300,
> occupations = 'smearing' ,
> degauss = 0.05 ,
> smearing = 'methfessel-paxton'
> /
> &ELECTRONS
> conv_thr = 1.D-10 ,
> mixing_beta = 0.7D0 ,
> /
> ATOMIC_SPECIES
> Al 26.98000 Al.vbc.UPF
> ATOMIC_POSITIONS angstrom
> Al 0.000000000 1.517315647 2.138739361
> Al 0.000000000 0.116702849 0.023699809
> K_POINTS automatic
> 2 2 20 0 0 0
>
> Al.elph.in
>
> Electron-phonon coefficients for Alwire_2
> &inputph
> tr2_ph=1.0d-14,
> prefix='al2_elph',
> fildvscf='al2d',
> filelph='matrix'
> amass(1)=26.98,
> outdir='/root/tmp/',
> fildyn='al2m.dyn',
> elph=.true.,
> trans=.true.,
> ldisp=.true.
> nq1=1, nq2=1, nq3=10,
>
> And hereafter 6 modes I got
>
> &plot nbnd= 6, nks= 27 /
> 0.000000 0.000000 0.000000
> -0.0855 -0.0450 0.0000 20.7938 196.2309 372.4942
> 0.000000 0.000000 0.100000
> -13.1432 5.6657 5.7114 20.3098 196.7060 372.0899
> 0.000000 0.000000 0.200000
> -25.6753 10.2550 10.5805 18.8468 198.1533 371.0008
> 0.000000 0.000000 0.300000
> -37.6999 14.0728 14.1858 16.2544 200.5950 369.4869
> 0.000000 0.000000 0.400000
> -48.8295 11.6034 16.4265 16.4576 203.9728 367.7201
> 0.000000 0.000000 0.500000
> -58.7192 -5.6944 16.7453 17.6129 208.1190 365.6663
> 0.000000 0.000000 0.600000
> -67.0895 -15.3952 14.9673 18.4203 212.8051 363.1793
> 0.000000 0.000000 0.700000
> -73.7443 -21.7038 11.9605 19.3942 217.8126 360.1938
> 0.000000 0.000000 0.800000
> -78.6120 -26.9948 9.8326 20.4642 222.9635 356.8134
> 0.000000 0.000000 0.900000
> -81.7857 -31.7595 10.1665 21.2176 228.1041 353.2106
> 0.000000 0.000000 1.000000
> -83.4878 -36.1149 10.5785 21.4924 233.0809 349.4591
> 0.000000 0.000000 1.100000
> -83.9637 -39.9380 6.4409 21.5912 237.7521 345.4939
> 0.000000 0.000000 1.200000
> -83.3549 -43.0700 -11.8614 21.8643 242.0400 341.2345
> 0.000000 0.000000 1.300000
> -81.7148 -45.5109 -19.5363 22.1233 245.9755 336.7214
> 0.000000 0.000000 1.400000
> -79.2450 -47.4002 -24.2602 21.7049 249.6731 332.1110
> 0.000000 0.000000 1.500000
> -76.4698 -48.7989 -26.8527 20.1426 253.2412 327.5276
> 0.000000 0.000000 1.600000
> -74.0708 -49.5494 -28.8342 17.6459 256.7020 322.9460
> 0.000000 0.000000 1.700000
> -72.4259 -49.4368 -31.6947 14.8469 259.9856 318.2672
> 0.000000 0.000000 1.800000
> -71.2114 -48.4948 -35.6608 11.6256 263.0044 313.5063
> 0.000000 0.000000 1.900000
> -69.6799 -47.0750 -39.7963 4.2536 265.7429 308.8670
> 0.000000 0.000000 2.000000
> -67.4586 -45.5010 -43.2503 -12.7026 268.2849 304.5960
> 0.000000 0.000000 2.100000
> -64.9647 -46.1360 -43.6300 -19.3793 270.7903 300.7406
> 0.000000 0.000000 2.200000
> -63.0506 -49.2247 -40.9136 -23.4729 273.4711 297.0578
> 0.000000 0.000000 2.300000
> -62.1609 -52.8376 -37.0315 -25.3558 276.5514 293.2015
> 0.000000 0.000000 2.400000
> -61.8563 -56.2311 -32.5370 -26.2996 280.1674 289.0264
> 0.000000 0.000000 2.500000
> -61.6387 -57.8617 -29.3445 -27.6304 283.8161 285.1565
> 0.000000 0.000000 2.503000
> -61.6384 -57.8633 -29.3391 -27.6341 283.8284 285.1440
>
>
> Bac,
>
> ------------------------------
> Phung Viet Bac
> Division of Mathematical and Physical Science,
> Graduate School of Natural Science and Technology,
> Kanazawa University,
> Kakuma, Kanazawa 920-1192, JAPAN
>
>
>
>
>
>
>
>
>>
>> your phonon branches are really strange...
>> with two atoms I would expect 6 modes per each q-point
>>
>> stefano
>>
>>
>>
>> 6 branches for 2 atom are what one expects (in general 3n branches and 3
>> acoustical and 3n-3 optical ones).
>>
>> However negative values indicate some error in your calculation, dependent
>> on the magnitude and system.
>> You can impose some sum-rules for the acoustic ones, to force them to be
>> 0 at the Gamma point.
>>
>> Acoustic forces are harder to calculate (smaller forces). In your case I
>> would guess up to 15 cm^1 should be ok for the acoustic ones.
>> Negative optical ones should not occur.
>>
>> Best
>> Marcel Mohr
>>
>>
>>
>
More information about the users
mailing list