I would try a ReaxFF potential set, like ReaxFF_CHONSSi_2011 - basically the only one available for Si-C-H in ATK, and perhaps in the literature as well, but if you find another one, please let us know. The other ReaxFF are just extensions of this set, I think.
We have tested ReaxFF_CHONSSi_2011 and it gives fairly good lattice constants for bulk Si, SiC, SiO2, diamond, and graphite, so it can probably be expected to work reasonably well for Si-C-H. But, as always with classical potentials, you had better test it for basic properties of the materials and structures are interested in! For instance, if your work involves graphene, test it for that.
