If you use Gromacs and running energy minimization, this means you were able to generate tpr file. Grompp will gives error, If any unrecognized atom is found. So, how come!!!

If you want to add a new atom type to your FF the safest way to do is to create an itp file and define new atom type (require mass  sigma LJ and eLJ and the charge) in the file like below (the file name should be included in the your topology file).  This prevent you to destroy your entire FF. However, you should always be careful about any parameter that you put in. Most of the parameter require QM calculation. pairtypes are more difficult. 

[ atomtypes ]

    CEL3     6    12.0110     -0.150     A    3.72395664183e-01    2.845120e-01

[ bondtypes ]

[ pairtypes ]

[ angletypes ]

The better question is where are these atom types coming from? I presume you've built some topology for silica somehow (unlikely within GROMACS), which requires some parameters that are not available. Is this topology compatible with GROMOS96 54A7? Where did you get it? If the topology requires some new atom types, the source of that topology should provide suitable parameters for those new types. If not, then you should be suspicious of whether or not that topology is actually compatible with the force field.

can you copy and paste a part of your .gro, like your DMSO atoms?

If you use Gromacs and running energy minimization, this means you were able to generate tpr file. Grompp will gives error, If any unrecognized atom is found. So, how come!!!

If you want to add a new atom type to your FF the safest way to do is to create an itp file and define new atom type (require mass  sigma LJ and eLJ and the charge) in the file like below (the file name should be included in the your topology file).  This prevent you to destroy your entire FF. However, you should always be careful about any parameter that you put in. Most of the parameter require QM calculation. pairtypes are more difficult. 

[ atomtypes ]

    CEL3     6    12.0110     -0.150     A    3.72395664183e-01    2.845120e-01

[ bondtypes ]

[ pairtypes ]

[ angletypes ]

If you know the properties of that atom and if you can find an atom with same properties with a different name then you can change yours into the defined one. If not, you can define a new atom type in Gromacs, but ofcourse you should know how to do it ;)

Thanks everyone for the advices!

Actually i'm doing protein-ligand interaction with silica. The atomtype OC23 that was not found was one of the atom in the silica. And, yes, the silica was not generated using gromacs. That's why some of the atom was not recognized by gromacs.

 I am new to GROMACS, running protein-ligand simulation.

I am having the same problem as Amirah. 

Fatal error:

Atomtype SDMSO not found. 

after the command: gmx grompp -f em.mdp -c solv_13948302.gro -p topol.top -o ions_13948302.tpr

Can you please help me to remove this error? 

Thanks in advance

Hello...I have added the GDP coordinates to the .gro file and now when I run the ionization step by hitting "gmx grompp -f ions.mdp -c solv.gro -p topol.top -o ions.tpr"....it gives me a fatal error as!

Atom type O not found.

Please enlighten

Hello...I generated an itp file and top file for calcite via OBGMX but while adding ions in the grompp command it cant identify atomtype. Can anybody enlighten me on how to define atomtype in gromacs?

Any other alternative solution is appreciable as well.

Thank You!