[Pw_forum] vc-relax

Giuseppe Mattioli giuseppe.mattioli at ism.cnr.it
Thu Oct 19 11:56:56 CEST 2017

Dear Ibrahim
I checked the structure with xcrysden and it seems correct to me, thus  
we can rule out the usually main reason for bad scf convergence. :-)
However there are still a few thing you should try to adjust:

1) celldm(1)=5.9722904441d0, celldm(3)=15.8777369953d0,

celldm(3) is too small. you have a thick slab (~40 A) and a very thin  
vacuum space between periodically repeated images (~10 A). You have  
therefore a strong interaction along z between the images, and this is  
not only a source of instability, but also a bad practice when you  
want to investigate surface properties. You should consider *at least*  
a vacuum space between slabs as large as the slab itself. And  
decoupling the images by using one of the possible approaches (e.g.,  
assume_isolated='esm'; esm_bc='bc1') can be also useful in the case of  
a strong dipole along z.

2) mixing_beta=0.7d0

This is the default value. In difficult cases you should lower it down  
to 0.1, or even to 0.01 in *very* difficult cases.

3) smearing='mv',

A 'gaussian' smearing generally yields a more stable algorithm. You  
can optimize your system and then restart the calculation with a  
different smearing scheme, if you are sure you need it.

4) vdw_corr='grimme-d2',

This is not directly related with your convergence issue, as DFT-D2 is  
a semiempirical plug-in independent of density or wavefunctions. But  
it is well known that it overbinds metal-containing slabs. You should  
try (maybe starting with a MoS2 bilayer) to evaluate the effect of  
different schemes, such as rVV10 or c09-vdw-df, on the structural and  
electronic properties of your slab.

5) ecutwfc=60,

The value is quite high for US pseudopotentials. If you are sure you  
need such an accurate basis set you might consider the use of NC  
pseudopotentials. ONCV PPs should be a very good choice: a wfc cutoff  
a bit higher (80 Ry) but a density cutoff quite lower (320 Ry), and no  
overhead with the US augmentation charge...


Quoting Ibrahim Yahaya Muhammad <ibrahimyahayamuhammad at gmail.com>:

> Dear all,
> I am trying to perform a 'vc-relax' calculation for seven layers MoS2. But
> convergence not achieved. I increase electron_maxstep to 1000 still
> convergence not achieved.
> Please find attached, the input and output.
> Best Regards,
> Ibrahim Yahaya Muhammad

Via Salaria Km 29,300 - C.P. 10
I-00015 - Monterotondo Scalo (RM)
Mob (*preferred*) +39 373 7305625
Tel + 39 06 90672342 - Fax +39 06 90672316
E-mail: <giuseppe.mattioli at ism.cnr.it>

More information about the users mailing list