Most potentials are not explicitly optimized for solid/fluid interfaces. More generally, many potentials are optimized only under certain conditions and applying them under different conditions can certainly raise questions. 

This poses am obvious question when you use bulk parameters for confined systems, but even if you use parameters optimized for a solid-liquid interface, they are probably optimized in combination with a specific fluid (often water), at a certain temperature. Those parameters might no longer be strictly valid when you change the temperature, or when you add ions to the water, or want to simulate a different fluid entirely. 

That being said, of course you do need to use some potential to do your simulations as good as you can. Hendrik Heinz published a number of papers (since 2008 I believe) on the INTERFACE potential, which includes a number of metals. I played around with it for copper and gold and was happy with the results. 

Dear Remco,

Thank you very much for your kind help.

In deed atomistic potentials are case-dependent. I will be very careful about this.

Dear Changming,

As Remco points out, there are not always parameters optimised for your desired simulation. So caution is always advised when analysing results. When simulating metals it is ideal to benchmark your results, where possible.

But if I may point out some additional parameters, there was a recent excellent paper by Li, Roberts, Chakravorty, and Merz, Journal of Chemical Theory and Computation, 2013, 9, 2733-2748, which has optimised parameters for a variety of M(II) metal ions based on their computational hydration free energies in water.  In addition, I have previously used the potentials given by Lee and Rasaiah,  Journal of Physical Chemistry, 1996, 100, 1420-1425, and am satisfied the results are good.

Cheers, 

Quinn