Here is my input:
%nprocshared=8
%mem=2GB
# opt freq b3lyp/6-31g(d) counterpoise=2
graphene atop optimization
0 3 0 1 0 1
Pd(Fragment=2) 4.61875610 3.25307744 3.04247012
Pd(Fragment=2) 2.01705610 3.34377744 3.22947012
Pd(Fragment=2) 3.39185610 5.56137744 3.11797012
Pd(Fragment=2) 3.48075610 4.05157744 5.15167012
C(Fragment=1) 1.97455610 0.49047744 1.10377012
C(Fragment=1) 0.74655610 1.19947744 1.10377012
C(Fragment=1) 0.74655610 2.61747744 1.10377012
C(Fragment=1) 1.97455610 3.32647744 1.10377012
C(Fragment=1) 4.43055610 0.49047744 1.10377012
C(Fragment=1) 3.20255610 1.19947744 1.10377012
C(Fragment=1) 3.20255610 2.61747744 1.10377012
C(Fragment=1) 4.43055610 3.32647744 1.10377012
C(Fragment=1) 6.88665610 0.49047744 1.10377012
C(Fragment=1) 5.65865610 1.19947744 1.10377012
C(Fragment=1) 5.65865610 2.61747744 1.10377012
C(Fragment=1) 6.88665610 3.32647744 1.10377012
C(Fragment=1) 1.97455610 4.74447744 1.10377012
C(Fragment=1) 0.74655610 5.45347744 1.10377012
C(Fragment=1) 0.74655610 6.87147744 1.10377012
C(Fragment=1) 1.97455610 7.58047744 1.10377012
C(Fragment=1) 4.43055610 4.74447744 1.10377012
C(Fragment=1) 3.20255610 5.45347744 1.10377012
C(Fragment=1) 3.20255610 6.87147744 1.10377012
C(Fragment=1) 4.43055610 7.58047744 1.10377012
C(Fragment=1) 6.88665610 4.74447744 1.10377012
C(Fragment=1) 5.65865610 5.45347744 1.10377012
C(Fragment=1) 5.65865610 6.87147744 1.10377012
C(Fragment=1) 6.88665610 7.58047744 1.10377012
Tv 7.14010000 0.00000000 0.00000000
Tv 0.00000000 8.09000000 0.00000000
Tv 0.00000000 0.00000000 5.04790000
How can I fragmentize the translational vectors?
Thanks
In addition, if you are modelling Pd4 cluster on grephene, your third translation vector is too short (top Pd atom touch periodic image of graphene in third (z) dimension). If it is really two-dimensional problem (Pd4 cluster on graphene), you shoul take advantage of implementation of two-dimensional (2D) periodic boundary conditions (PBC) implemented in Gaussian09. You switch to this 2D PBC just by deleting the third translation vector. There will be big reduction on cpu time.
You could start optimization with GGA DFT (e.g. PBE). I am also afraid of freq in Gaussian09 PBC. Probably, this is not implemented for PBC. Therefore you could do firstly e.g.
# opt pbepbe/6-31g(d)
and then for resulting geometry (if you need hybrid DFT) continue with
# opt hseh1pbe/6-31g(d)
(you can also consider pseudopotential for Pd)
Hi,
It may be that automatic counterpoise correction cannot be done using periodic boundary conditions. You can try to do the correction manually after optimization by using ghost atoms (Pd-Bq, C-Bq). The following page can be helpful in understanding what to calculate:
http://scicomp.stackexchange.com/questions/3/what-is-counterpoise-correction
Your system is also very large for using three dimensions. You might want to delete the last one or two translational vectors first to see how much your computer can handle.
I hope this helps.
Unfortunately, counterpoise correction is not implemented in periodic boundary conditions (PBC) in Gaussian. You can do it manually as written above. Or, e.g. in Crystal software(this also use gaussian basis sets) this is implemented. By the way, PBC & B3LYP will be very time consuming (you can consult notes for developement version of gaussian http://scuseria.rice.edu/gau/gdv_pbc/pbc_guide.htm, Chapter 1, "What not to do"). If you like to use hybrid DFT & PBC, it is reasonable to use short-range hybrid functional HSE06 (keyword HSEh1PBE).
In addition, if you are modelling Pd4 cluster on grephene, your third translation vector is too short (top Pd atom touch periodic image of graphene in third (z) dimension). If it is really two-dimensional problem (Pd4 cluster on graphene), you shoul take advantage of implementation of two-dimensional (2D) periodic boundary conditions (PBC) implemented in Gaussian09. You switch to this 2D PBC just by deleting the third translation vector. There will be big reduction on cpu time.
You could start optimization with GGA DFT (e.g. PBE). I am also afraid of freq in Gaussian09 PBC. Probably, this is not implemented for PBC. Therefore you could do firstly e.g.
# opt pbepbe/6-31g(d)
and then for resulting geometry (if you need hybrid DFT) continue with
# opt hseh1pbe/6-31g(d)
(you can also consider pseudopotential for Pd)
There are weak vdW forces between meatl cluster and graphene, therefore pure or hybrid DFT is generally not not enough, see e.g.
https://www.researchgate.net/publication/229076483_Interaction_of_Graphene_and_Arenes_with_Noble_Metals?ev=prf_pub
Article Interaction of Graphene and Arenes with Noble Metals
- Badges
- Science topic
- Similar topics
- Chemistry
- Theoretical Chemistry