I am talking about edge roughness in NiFe nanostructure.

Edge roughness in such processes is typically a direct function of grain size in your thin film. Many things can influence grain size including the vacuum level in your evaporator, temperature, dep rate, etc. If you have a cold stage in your evaporator, that can help or you can try to see if you can tune grain size by changing the deposition rate. Also, the substrate composition and surface preparation can have important effects due to surface energy differences - you can try things like plasma cleaning after EBL.

What's the critical dimention of your patten?  It's difficult to get smooth edge while using lift-off process. Three possible solutions:

1. After lift off, try annealing to smooth the line edge.

2. After lift off, use focused ion beam (at low ion current) to milling the edge parts.

3. Deposit NiFe thin film first and use focused ion beam to fabricate your pattern.

No matter how large the grain size, you can get smooth cross section by focused ion beam fabrication (at low ion current...).

My critical dimensions are ~100 to 200 nm. As you suggested I’ll try your first solution (annealing after lift off). Also, I found that Dr. Myers suggestions are very useful, now I’m trying to install a heat sink on top of my substrate holder so that I can avoid heating of the sample, while evaporating NiFe. As my substrate is Au (50 nm) coated Silicon. Does the Au film contribute any surface roughness ?

Yes, 50nm gold will contribute roughness, depending on how you deposit Au.  Again, slow and low temperature process may reduce the roughness. You can determine the surface roughness by AFM.

I have measured the surface rms roughness of Au which is less then 2 nm. But the problem is NiFe structures edge roughness (spikes are more than the film thickness).

I'm not familiar with NiFe.  Sometimes spikes are related to process temperature. Please check "Thermally induced hillock formation in Al-Cu films" to see if it's helpful.