[CP2K-user] [CP2K:17166] Re: COM Drift in Droplet Thermolysis

[Marcella Iannuzzi]

Dear Daniel 

The COM velocity is a constant of motion, if this is not the case there is 
a problem with the integration of the equations of motion. 
Without more information it is difficult to guess. For instance, are you 
sure that the SCF is sufficiently converged, that the time step is adequate 
at the temperature you are simulating, that the box is large enough, that 
the simulation is correctly restarted ...? 
Why is PBC-POISSON used for a droplet? For ionic liquid the electrostatics 
might create issues, I suppose. 

Regard 
Marcella

On Monday, June 13, 2022 at 6:59:44 AM UTC+2 depew.... at gmail.com wrote:

> Hi,
>
> I'm new to CP2K and am hoping to simulate droplet evaporation/thermolysis 
> of ionic liquids. I had previously set up these droplets with classical MD 
> and plenty of surrounding vacuum (~15-20 Angstroms on each side, sized for 
> ~1 MPa of pressure for ideal gaseous products).
>
> Based on a literature review, I had set up a simulation with a revPBE-D3 
> functional and standard DZVP-MOLOPT-SR-GTH basis set. I'll post the whole 
> &DFT section below, but to summarize I used a grid cutoff of 400 Ry and SCF 
> convergence of 1e-6. I also used XC smoothing (NN10) for the grid.
>
> I heated the droplet up to 600K from an initial 300K using 0.5 ps 
> simulations with steps of 50 K). The time step was 0.5 fs. At this point, I 
> was hoping to do production runs, but I noticed that when I restarted at 
> 600 K, the initial temperature spiked to over 1800 K despite being ~600 K 
> on the previous step. I traced this down to a nonzero center of mass 
> velocity and the way that initial temperature is computed in md_vel_utils.F 
> -- the COM velocity is subtracted on the initial time step when computing 
> temperature, but it is not on subsequent steps.
>
> This post <https://groups.google.com/g/cp2k/c/VmDctR1A1Jg> suggests that 
> I'm not computing the forces accurately enough, so I went back to check 
> grid convergence for the first step. The total charge density 
> <https://www.cp2k.org/faq:cutoff> is on the order 10^-8 and I'm unable to converge 
> the cutoff <https://www.cp2k.org/howto:converging_cutoff> even at 1000 
> Ry. Because of the system size, I'm hesitant to increase the computational 
> cost much more. That post suggests either trying GAPW or simply subtracting 
> the COM.
>
> Before going any further, I wanted to consult your expertise. Is it 
> primarily the lack of energy convergence driving the COM drift? Would 
> higher temperatures (i.e. larger atomic displacements) tend to exacerbate 
> this effect? Is this something encountered in bulk-phase simulations but 
> gets washed out due to the periodic nature?
>
> I should mention that the simulations I'm doing are primarily to analyze 
> structure, reactivity, and the initial evaporation processes at the 
> interface, so it's largely a qualitative study meant to supplement other 
> work.
>
> Here is the &DFT input section:
>    &DFT
>      BASIS_SET_FILE_NAME BASIS_MOLOPT
>      POTENTIAL_FILE_NAME GTH_POTENTIALS
>      MULTIPLICITY  1
>      CHARGE  0
>      &SCF
>        SCF_GUESS ATOMIC
>        MAX_SCF 50
>        EPS_SCF 1.0E-6
>        &OT
>          PRECONDITIONER FULL_SINGLE_INVERSE
>          MINIMIZER DIIS
>        &END OT
>        &OUTER_SCF
>          MAX_SCF 10
>          EPS_SCF 1.0E-6
>        &END
>        &PRINT
>          &RESTART OFF
>          &END
>        &END
>      &END SCF
>      &QS
>        EPS_DEFAULT     1.0000000000000000E-10
>        EXTRAPOLATION  ASPC
>        METHOD  GPW
>      &END QS
>      &MGRID
>         CUTOFF 400
>         NGRIDS 5
>         REL_CUTOFF 30.0
>      &END
>      &XC
>        DENSITY_CUTOFF     1.0000000000000000E-10
>        GRADIENT_CUTOFF     1.0000000000000000E-10
>        TAU_CUTOFF     1.0000000000000000E-10
>        &XC_GRID
>          XC_SMOOTH_RHO  NN10
>          XC_DERIV  NN10_SMOOTH
>        &END XC_GRID
>        &XC_FUNCTIONAL  NO_SHORTCUT
>          &GGA_X_PBE_R  T
>          &END GGA_X_PBE_R
>          &GGA_C_PBE  T
>          &END GGA_C_PBE
>        &END XC_FUNCTIONAL
>        &VDW_POTENTIAL
>          POTENTIAL_TYPE  PAIR_POTENTIAL
>          &PAIR_POTENTIAL
>            R_CUTOFF     1.6000000000000000E+01
>            TYPE  DFTD3
>            PARAMETER_FILE_NAME dftd3.dat
>            REFERENCE_FUNCTIONAL PBE
>          &END PAIR_POTENTIAL
>        &END VDW_POTENTIAL
>      &END XC
>      &POISSON
>        PERIODIC  XYZ
>      &END POISSON
>    &END DFT
>
> Thanks,
> Daniel
>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/83f568c7-6162-4d0e-b858-6651341ff077n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20220614/3d3ea716/attachment.htm>