Negative Frequency in Phonon Dispersion of SnSe

[zohansyahfatomi]

Dear Tadano Sensei,

Good evening.
I am calculating the phonon-dispersion of SnSe.
However, I got the imaginary frequency.
Is it due to the interlayer configuration of SnSe that phonon calculation is hard?
Do you have any suggestions?
Thank you very much.

Best regards,
Zohan

&general
 PREFIX = snse222
 MODE = suggest
 NAT = 64; NKD = 2
 KD = Sn Se
/

&interaction
 NORDER = 1
/

&cutoff
*-* None
/

&cell
1.0
44.292856494303000	0.000000000000000	0.000000000000000
0.000000000000000	15.960091768091000	0.000000000000000
0.000000000000000	0.000000000000000	17.123091497091000
/

&position
1	0.061148663000000	0.124999999000000	0.050059603000000
1	0.061148663000000	0.124999999000000	0.550059603000000
1	0.061148663000000	0.624999999000000	0.050059603000000
1	0.061148663000000	0.624999999000000	0.550059603000000
1	0.561148663000000	0.124999999000000	0.050059603000000
1	0.561148663000000	0.124999999000000	0.550059603000000
1	0.561148663000000	0.624999999000000	0.050059603000000
1	0.561148663000000	0.624999999000000	0.550059603000000
1	0.438851337000000	0.374999996000000	0.449940399000000
1	0.438851337000000	0.374999996000000	0.949940399000000
1	0.438851337000000	0.874999996000000	0.449940399000000
1	0.438851337000000	0.874999996000000	0.949940399000000
1	0.938851337000000	0.374999996000000	0.449940399000000
1	0.938851337000000	0.374999996000000	0.949940399000000
1	0.938851337000000	0.874999996000000	0.449940399000000
1	0.938851337000000	0.874999996000000	0.949940399000000
1	0.190024306000000	0.374999996000000	0.300072363000000
1	0.190024306000000	0.374999996000000	0.800072363000000
1	0.190024306000000	0.874999996000000	0.300072363000000
1	0.190024306000000	0.874999996000000	0.800072363000000
1	0.690024306000000	0.374999996000000	0.300072363000000
1	0.690024306000000	0.374999996000000	0.800072363000000
1	0.690024306000000	0.874999996000000	0.300072363000000
1	0.690024306000000	0.874999996000000	0.800072363000000
1	0.309975698000000	0.124999999000000	0.199927639000000
1	0.309975698000000	0.124999999000000	0.699927639000000
1	0.309975698000000	0.624999999000000	0.199927639000000
1	0.309975698000000	0.624999999000000	0.699927639000000
1	0.809975698000000	0.124999999000000	0.199927639000000
1	0.809975698000000	0.124999999000000	0.699927639000000
1	0.809975698000000	0.624999999000000	0.199927639000000
1	0.809975698000000	0.624999999000000	0.699927639000000
2	0.427929495000000	0.124999999000000	0.241227348000000
2	0.427929495000000	0.124999999000000	0.741227348000000
2	0.427929495000000	0.624999999000000	0.241227348000000
2	0.427929495000000	0.624999999000000	0.741227348000000
2	0.927929495000000	0.124999999000000	0.241227348000000
2	0.927929495000000	0.124999999000000	0.741227348000000
2	0.927929495000000	0.624999999000000	0.241227348000000
2	0.927929495000000	0.624999999000000	0.741227348000000
2	0.072070509000000	0.374999996000000	0.258772654000000
2	0.072070509000000	0.374999996000000	0.758772654000000
2	0.072070509000000	0.874999996000000	0.258772654000000
2	0.072070509000000	0.874999996000000	0.758772654000000
2	0.572070509000000	0.374999996000000	0.258772654000000
2	0.572070509000000	0.374999996000000	0.758772654000000
2	0.572070509000000	0.874999996000000	0.258772654000000
2	0.572070509000000	0.874999996000000	0.758772654000000
2	0.321043050000000	0.374999996000000	0.491327381000000
2	0.321043050000000	0.374999996000000	0.991327381000000
2	0.321043050000000	0.874999996000000	0.491327381000000
2	0.321043050000000	0.874999996000000	0.991327381000000
2	0.821043050000000	0.374999996000000	0.491327381000000
2	0.821043050000000	0.374999996000000	0.991327381000000
2	0.821043050000000	0.874999996000000	0.491327381000000
2	0.821043050000000	0.874999996000000	0.991327381000000
2	0.178956950000000	0.124999999000000	0.008672620000000
2	0.178956950000000	0.124999999000000	0.508672620000000
2	0.178956950000000	0.624999999000000	0.008672620000000
2	0.178956950000000	0.624999999000000	0.508672620000000
2	0.678956950000000	0.124999999000000	0.008672620000000
2	0.678956950000000	0.124999999000000	0.508672620000000
2	0.678956950000000	0.624999999000000	0.008672620000000
2	0.678956950000000	0.624999999000000	0.508672620000000
/

[ttadano]

This result indicates some errors in your procedure.
I remember I obtained a similar result to https://arxiv.org/vc/arxiv/papers/1602/1602.03762v2.pdf a long time ago.

What is the primitive unit cell you used?

[zohansyahfatomi]

Dear Tadano Sensei,

Thank you for the comments.
I already try it as follows:

&general
  PREFIX = snse222
  MODE   = phonons
  FCSXML = snse222.xml

  NKD = 2; KD = Sn Se
  MASS = 118.71 78.96
/

&cell
1.0
22.1464282472	0.000000000000000	0.000000000000000
0.000000000000000	7.98004588405	0.000000000000000
0.000000000000000	0.000000000000000	8.56154574855
/

&kpoint
 1  # KPMODE = 1: line mode
G	0	0	0	X	0.5	0	0	51
X	0.5	0	0	S	0.5	0.5	0	51
S	0.5	0.5	0	Y	0	0.5	0.5	51
Y	0	0.5	0.5	G	0	0	0	51
G	0	0	0	Z	0.5	0	0	51
Z	0.5	0	0	U	0.5	0.5	0	51
U	0.5	0.5	0	R	0.5	0.5	0.5	51
R	0.5	0.5	0.5	T	0.5	0	0.5	51
T	0.5	0	0.5	Z	0.5	0	0	51
Y	0	0.5	0.5	T	0.5	0	0.5	51
U	0.5	0.5	0	X	0	0.5	0	51
S	0	0.5	0.5	R	0.5	0.5	0.5	51
/

Just for your information, for calculating IFC, I use openMX.
I use displacement of 0.00001 Angstrom
Thank you very much.

[ttadano]

The displacement magnitude is too small. Please use 0.01–0.04 Angstroms.

[zohansyahfatomi]

Dear Tadano Sensei,

I understand.
I have questions:

1. Is 64 atoms enough for phonon calculation? Because I checked the silicon case, 64 is enough.
2. Do we need to optimize the supercell (64 atoms) because I already optimized the unit cell (8 atoms)
   Thank you very much.

[ttadano]

For SnSe, I think the supercell should be larger. Please read the previous papers on phonon calculations of SnSe.

You do not need to relax atom positions using the supercell further, but I recommend that you subtract the offset forces following https://alamode.readthedocs.io/en/latest/faq.html#frequently-asked-questions-faq.

[zohansyahfatomi]

Dear Tadano Sensei,

Thank you for the comments.
I do convergence test of supercell size:
The fitting error:
8 atoms: 25%
64 atoms: 15 %
216 atoms: being calculated, but I guess it will be less than 10%.
My question is, why does a larger supercell have a less fitting error? I think that 8 atoms and 64 atoms represent the same system of SnSe.
Also, does alamode can be implemented for the mixed-space approach? if so, how to do that?
Thank you very much.

Best regards,
Zohan

8 atoms

64 atoms