[Pw_forum] Huge imaginary and positive frequencies
Cameron Foss
cjfoss at umass.edu
Sat Nov 18 00:12:38 CET 2017
Hello all,
I am trying to calculate the phonon dispersion of a monolayer MXene
structure, Ti3C2O2. My input files roughly follow the recipe laid out in
the following publication: *Phys. Chem. Chem. Phys., 2015, 17, 9997--10003*,
with the exception that I am trying to use Perturbation theory rather than
a supercell approach. I have pasted my input files below for clarity. I am
using QE-6.1. I start with a force relaxation calculation where the forces
relax to*: 0.000039 Ry/bohr or ~0.001 eV/Ang*. After the structural
relaxation, I perform a gamma-only phonon calculation to see if the
gamma-point frequencies reproduce those shown the in publication above, and
they more or less do. So I then move on to a full phonon dispersion
calculation, where the dyn1 Gamma-point converges well (with
conv_thr=1.0d-12) with reasonable frequencies (see below) but dyn2 does not
converge unless I loosen the conv_thr to ~2.0d-8. In doing so, I get very
erratic frequencies.
*%% resulting forces and pressure*
Forces acting on atoms (cartesian axes, Ry/au):
atom 1 type 1 force = 0.00000000 0.00000000 0.00002656
atom 2 type 1 force = 0.00000000 0.00000000 -0.00001506
atom 3 type 1 force = 0.00000000 0.00000000 -0.00001422
atom 4 type 2 force = 0.00000000 0.00000000 0.00001412
atom 5 type 2 force = 0.00000000 0.00000000 -0.00001214
atom 6 type 3 force = 0.00000000 0.00000000 -0.00000168
atom 7 type 3 force = 0.00000000 0.00000000 0.00000242
*Total force = 0.000039 * Total SCF correction = 0.000002
Computing stress (Cartesian axis) and pressure
total stress (Ry/bohr**3) (kbar) P=
-3.79
-0.00000053 -0.00000000 0.00000000 -0.08 -0.00 0.00
-0.00000000 -0.00000053 0.00000000 -0.00 -0.08 0.00
0.00000000 0.00000000 -0.00007620 0.00 0.00
* -11.21 !! is this alarmingly high? even though the forces look fine.*
*%% Relaxation input file that give the above forces*
&control
calculation='vc-relax',
restart_mode='from_scratch',
outdir='./out_Ti3C2',
max_seconds=144000,
prefix='ti3c2o2_phonons',
pseudo_dir='/qe-6.1/pseudo',
forc_conv_thr=5.0D-5,
etot_conv_thr=5.0D-5,
/
&system
ibrav=0, celldm(1)=5.71020728999, !celldm(3)=6.120077,
nat=7, ntyp=3,
ecutwfc=40, ecutrho=320, occupations='smearing',
smearing='methfessel-paxton',
degauss=0.007349862,
/
&electrons
electron_maxstep=250,
mixing_beta=0.7,
conv_thr=1.0d-12
/
&ions
ion_dynamics='bfgs',
/
&cell
cell_dynamics='bfgs',
cell_dofree='2Dxy',
/
ATOMIC_SPECIES
Ti 47.867 Ti.pbe-sp-van_ak.UPF
C 12.01 C.pbe-van_ak.UPF
O 15.999 O.pbe-van_ak.UPF
ATOMIC_POSITIONS {alat}
Ti 0.0000000000 0.0000000000 2.0371640580 1 1 1
Ti 0.0000000000 0.5773502692 2.8897518789 1 1 1
Ti 0.5000000000 0.288675135 1.1845717437 1 1 1
C 0.5000000000 0.288675135 2.4539683895 1 1 1
C 0.0000000000 0.5773502692 1.6203671083 1 1 1
O 0.0000000000 0.0000000000 3.1916866966 1 1 1
O 0.0000000000 0.0000000000 0.8826244595 1 1 1
K_POINTS automatic
27 27 3 0 0 0
CELL_PARAMETERS alat
0.999999999 -0.000000000 0.000000000 1 1 0
-0.500000000 0.866025404 0.000000000 1 1 0
-0.000000000 -0.000000000 6.705027151 0 0 0
*SCF pw.x input file nearly identical minus the options for the relaxation
calc. Also MXene's of this type are metallic so smearing is used. I use the
same amount of smearing as the publication above.*
*%% Phonon ph.x input*
&inputph
!recover=.true.,
tr2_ph=1.0d-12, ! desired conv_thr
! tr2_ph=2.0d-9, ! used for dyn2 mode 1 !! pw_forum can ignore this
! tr2_ph=2.0d-8, ! for the stubborn dyn2
max_seconds=324000,
prefix='ti3c2o2_phonons',
ldisp=.true.,
nq1=9, nq2=9, nq3=1,
start_q=2, last_q=2,
amass(1)=47.867,
amass(2)=12.01,
amass(3)=15.999,
outdir='./out_Ti3C2'
fildyn='ti3c2o2.dyn',
/
*%% frequencies from gamma-only phonon calculation (input file not shown)
but conv_thr was 1.0d-12*
Diagonalizing the dynamical matrix
q = ( 0.000000000 0.000000000 0.000000000 )
**************************************************************************
freq ( 1) = 0.253345 [THz] = 8.450683 [cm-1]
freq ( 2) = 1.719726 [THz] = 57.363871 [cm-1]
freq ( 3) = 1.719726 [THz] = 57.363871 [cm-1]
freq ( 4) = 3.619841 [THz] = 120.744905 [cm-1]
freq ( 5) = 3.619841 [THz] = 120.744905 [cm-1]
freq ( 6) = 5.807810 [THz] = 193.727688 [cm-1]
freq ( 7) = 5.807810 [THz] = 193.727689 [cm-1]
freq ( 8) = 5.960133 [THz] = 198.808631 [cm-1]
freq ( 9) = 9.061633 [THz] = 302.263531 [cm-1]
freq ( 10) = 9.061633 [THz] = 302.263531 [cm-1]
freq ( 11) = 9.786433 [THz] = 326.440252 [cm-1]
freq ( 12) = 9.786433 [THz] = 326.440252 [cm-1]
freq ( 13) = 10.842274 [THz] = 361.659340 [cm-1]
freq ( 14) = 14.767962 [THz] = 492.606185 [cm-1]
freq ( 15) = 14.767962 [THz] = 492.606185 [cm-1]
freq ( 16) = 16.094128 [THz] = 536.842320 [cm-1]
freq ( 17) = 16.094128 [THz] = 536.842320 [cm-1]
freq ( 18) = 17.189295 [THz] = 573.373155 [cm-1]
freq ( 19) = 17.357952 [THz] = 578.998944 [cm-1]
freq ( 20) = 20.684968 [THz] = 689.976250 [cm-1]
freq ( 21) = 22.301687 [THz] = 743.904207 [cm-1]
*%% Generated q-points for 9x9x1 MP grid*
Dynamical matrices for ( 9, 9, 1) uniform grid of q-points
( 12q-points):
N xq(1) xq(2) xq(3)
1 0.000000000 0.000000000 0.000000000
2 0.000000000 0.128304047 0.000000000
3 0.000000000 0.256608094 0.000000000
4 0.000000000 0.384912141 0.000000000
5 0.000000000 0.513216187 0.000000000
6 0.111114564 0.192456070 0.000000000
7 0.111114564 0.320760117 0.000000000
8 0.111114564 0.449064164 0.000000000
9 0.111114564 0.577368211 0.000000000
10 0.222229128 0.384912141 0.000000000
11 0.222229128 0.513216187 0.000000000
12 0.333343692 0.577368211 0.000000000
*%% frequencies at Gamma-point from full dispersion calculation for
conv_thr=1.0d-12*
Diagonalizing the dynamical matrix
q = ( 0.000000000 0.000000000 0.000000000 )
**************************************************************************
freq ( 1) = -1.943043 [THz] = -64.812924 [cm-1]
freq ( 2) = -1.943043 [THz] = -64.812924 [cm-1]
freq ( 3) = -1.146697 [THz] = -38.249699 [cm-1]
freq ( 4) = 3.920804 [THz] = 130.783951 [cm-1]
freq ( 5) = 3.920804 [THz] = 130.783951 [cm-1]
freq ( 6) = 5.875260 [THz] = 195.977567 [cm-1]
freq ( 7) = 6.291059 [THz] = 209.847157 [cm-1]
freq ( 8) = 6.291059 [THz] = 209.847157 [cm-1]
freq ( 9) = 9.352154 [THz] = 311.954277 [cm-1]
freq ( 10) = 9.352154 [THz] = 311.954277 [cm-1]
freq ( 11) = 10.025397 [THz] = 334.411240 [cm-1]
freq ( 12) = 10.025397 [THz] = 334.411240 [cm-1]
freq ( 13) = 10.807507 [THz] = 360.499614 [cm-1]
freq ( 14) = 14.986167 [THz] = 499.884708 [cm-1]
freq ( 15) = 14.986167 [THz] = 499.884708 [cm-1]
freq ( 16) = 16.034628 [THz] = 534.857602 [cm-1]
freq ( 17) = 16.034628 [THz] = 534.857603 [cm-1]
freq ( 18) = 17.202121 [THz] = 573.801000 [cm-1]
freq ( 19) = 17.410826 [THz] = 580.762646 [cm-1]
freq ( 20) = 20.676124 [THz] = 689.681265 [cm-1]
freq ( 21) = 22.349113 [THz] = 745.486162 [cm-1]
(In my past simulations of various materials where I've achieved a good
dispersion, I've noticed that small negative/imaginary frequencies near the
gamma-point for acoustic modes are ok..)
*%% frequencies at second q-point from full dispersion calculation after
loosening conv_thr to 2.0d-8 due to convergence issues*
Diagonalizing the dynamical matrix
q = ( 0.000000000 0.128304047 0.000000000 )
**************************************************************************
freq ( 1) = -58.972744 [THz] = -1967.119008 [cm-1]
freq ( 2) = -57.240460 [THz] = -1909.336227 [cm-1]
freq ( 3) = -48.504741 [THz] = -1617.944007 [cm-1]
freq ( 4) = -44.973984 [THz] = -1500.170631 [cm-1]
freq ( 5) = -24.011317 [THz] = -800.931308 [cm-1]
freq ( 6) = -20.216067 [THz] = -674.335394 [cm-1]
freq ( 7) = -14.459694 [THz] = -482.323488 [cm-1]
freq ( 8) = 5.896048 [THz] = 196.671001 [cm-1]
freq ( 9) = 10.557595 [THz] = 352.163448 [cm-1]
freq ( 10) = 12.651229 [THz] = 421.999584 [cm-1]
freq ( 11) = 15.876635 [THz] = 529.587523 [cm-1]
freq ( 12) = 19.799490 [THz] = 660.439907 [cm-1]
freq ( 13) = 21.309016 [THz] = 710.792280 [cm-1]
freq ( 14) = 24.971802 [THz] = 832.969663 [cm-1]
freq ( 15) = 25.083773 [THz] = 836.704597 [cm-1]
freq ( 16) = 25.193004 [THz] = 840.348172 [cm-1]
freq ( 17) = 32.881094 [THz] = 1096.795223 [cm-1]
freq ( 18) = 41.517549 [THz] = 1384.876376 [cm-1]
freq ( 19) = 68.062454 [THz] = 2270.319090 [cm-1]
freq ( 20) = 71.147302 [THz] = 2373.218533 [cm-1]
freq ( 21) = 72.136680 [THz] = 2406.220630 [cm-1]
I've bolded part of the forces output above that shows a large negative
pressure in the z-direction. It is unclear to me if that would cause such
behavior.
If anyone notices an error in my input files or has had some experience
with MXene structures please feel free to chime in. Any advice is greatly
appreciated.
Cheers,
Cameron
--
NanoEnergy and Thermophysics Lab
Electrical and Computer Engineering Dept.
University of Massachusetts Amherst
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20171117/32619d74/attachment.html>
More information about the users
mailing list