[QE-users] Problem obtaining reliable cell constants from vc-relax (molecular crystals)

[Eric Chan]

Hi,

I am a chemical crystallographer attempting to better understand and test
usage of "supposedly straightforward" QE vc-relax in optimizing cell
constants.  I am using form I of paracetamol as a test case. The starting
structure is from the 20K neutron data.   For this test calculation I am
actually following guidelines which were outlined in an article on the
subject (J. Chem. Theory Comput. 2014, 10, 3423−3437), the dispersion
correction is vdW-DF2.

the experimental lattice parameters are as follows :

_cell_length_a                   12.667(4)
_cell_length_b                   9.166(3)
_cell_length_c                   7.073(3)
_cell_angle_alpha                90
_cell_angle_beta                 115.51(2)
_cell_angle_gamma                90
_cell_volume                     741.156

the output form the optimization is

_cell_length_a           12.994686
_cell_length_b           9.115636
_cell_length_c           6.461433
_cell_angle_alpha        90.000000
_cell_angle_beta         115.510024
_cell_angle_gamma        90.000000

which is strange because the angle beta is unchanged and the a and c
lengths have become inaccurate. I believe the deviation is too high.

attached is the .in file
I have  missed something, and cannot figure out what that is?
can anyone help me master the art of predicting reliable cell constants
using  QE vc-relax?

any help is greatly appreciated.

-Eric
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20181002/46dea7dc/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: form1_20K_HXACAN13_relax-vc_all.in
Type: application/octet-stream
Size: 3844 bytes
Desc: not available
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20181002/46dea7dc/attachment.obj>