In looking at your input file one potential problem is that you have pair_coeff * * for the sw potential definition when I'm assuming you want that to be pair_coeff 1 1 to define the interaction btw the Si atoms. That may account for the crashing. I'm not sure if defining the other pair_coeff afterwards overrides the * * definition or not.
Also - when inputting LJ parameters into a simulation with metal units watch out for the unit conversion. You are using sigma = 1.0 angstroms and epsilon = 1.0 eV. This sigma is pretty huge (I think) so it might also be causing some strange interactions. You should be able to find some info out there for the either the Si-Cu LJ parameters or fits for the Si-Si and Cu-Cu LJ parameters which you can then mix.
I had some similar issues a little while ago and Steve sent me a package of files with good examples. I'll send it your way in a separate email.
Hope that helps.