RDF stricly decreasing to zero

GROMACS version:2016.2
GROMACS modification: Yes/No
Hello Everyone!
I have made a simulation with 400 pure 4-hydroxybenzyl alcohols.
After the MD production, the box seems to be well equilibrated (temperature, density, rmsd).
The length of the final box is 4nm.
I was trying to made an rdf function between C6-C6 atoms (is the group 3 on the .ndx file) using the command:
gmx rdf -f md.trr -n indexrdf.ndx -ref 3 -sel 3 -o g_C6_C6.xvg
but i get the attached strange graph. The rdf does not converges to one. Is that normal?
Any idea about the reason i get this result is really welcomed.
Thank you :)