WARNING : Breaking symmetry due to perfect biradical

GULP
1

Hello Dear,

I have received such an error in MOF crystal with P1 symmetry along with the charge calculation error.

What might be the possible reason?

Thanks in advance.

Emre

INPUT:
opti conv conse qok nomod pres spat prop pot efg qiter gfnff
ftol 1e-004
gtol 0.001
xtol 1e-004
maxcyc 1000

title
GULP calculation from Materials Studio for UiO66_48_1
end
cell
20.700400 20.700400 20.700400 90.000000 90.000000 90.000000 1 1 1 1 1 1
COORDINATES
.
.
.

OUTPUT:
!!!
!! WARNING : Breaking symmetry due to perfect biradical
!!!

!!!
!! WARNING : Breaking symmetry due to perfect biradical
!!!

Components of energy :

Interatomic potentials = 0.00000000 eV
GFNFF force field = 3433.81999812 eV
Monopole - monopole (real) = -4902.10911932 eV
Monopole - monopole (recip)= -1697.72377047 eV
Monopole - monopole (total)= -6599.83288980 eV
GFNFF Coulomb correction = 64.20109351 eV

Total lattice energy = -3101.81179817 eV

Total lattice energy = -299277.4230 kJ/(mole unit cells)

Number of variables = 1725

Maximum number of calculations = 1000
Maximum Hessian update interval = 10
Maximum step size = 1.000000000
Maximum parameter tolerance = 0.000100000
Maximum function tolerance = 0.000100000
Maximum gradient tolerance = 0.001000000
Maximum gradient component = 0.010000000

Symmetry not applied to optimisation

Cell parameters to be optimised using strains

Newton-Raphson optimiser to be used

BFGS hessian update to be used

Hessian eigenvalue shifts to be used to ensure all values are positive

General 2-D hessian to be used

Start of bulk optimisation :

Cycle: 0 Energy: -3101.811800 Gnorm: 4.442249 CPU: 4.524
** Hessian lowest eigenvalue is negative = -2.089E+06
** Hessian calculated **
Cycle: 1 Energy: -3342.921440 Gnorm: 2124.183022 CPU: 6.658
** Hessian lowest eigenvalue is negative = -1.148E+11
** Hessian calculated **
Cycle: 2 Energy: -3453.117855 Gnorm: 4597.553258 CPU: 12.043
** Hessian lowest eigenvalue is negative = -3.656E+11
** Hessian calculated **
Cycle: 3 Energy: -3615.253320 Gnorm: 9951.516318 CPU: 17.366
** Hessian lowest eigenvalue is negative = -1.164E+12
** Hessian calculated **
Cycle: 4 Energy: -3648.998281 Gnorm: 11322.535930 CPU: 22.178
** Hessian lowest eigenvalue is negative = -1.413E+12
** Hessian calculated **

!!!
!! ERROR : iterative charge solution failed in eemd_gfnff
!!!

Program terminated by processor 0 in eemd_gfnff

2

Hi Emre,
The issue here isn’t really the final error (i.e. that the charge solution has failed), but more what is happening to the system during the optimisation. As you will have hopefully noticed, the Gnorm (i.e. the gradients) are increasing all the time in a way that is unphysical, which means that something in your structure is collapsing and the system is imploding. As you haven’t posted the full input file it’s impossible to say why, but it probably means either there is something wrong with the structure or the force field can’t handle the system as it currently is set up (see the pGFNFF paper for discussion of how topology and other factors can influence the result).
Hope that helps,
Julian

3

Hi Dear Prof. Gale,

Thank you for the informative reply. When the MOFs are considered, one must be careful regarding the crystallographic errors.
The input is attached.
Best regards
Emre

UiO66_48_1.gin (36.5 KB)

4

Hi Emre,
Thanks for providing your structure. After changing the method to be pGFN-FF (rather than UFF) as in the file) then I was able to investigate. The issue here is that for the periodic version of GFN-FF (as opposed to the original finite system form) you really need to use the Wolf sum to converge the electrostatic interactions by adding the keyword “gwolf”. With this added the optimisation is now stable and convergent, rather than divergent.
Regards,
Julian

5

Dear Prof.

I applied your suggestions and now working.

Thank you very much.

Emre