Roughness on the surface of phase-only micro-optical elements limits their performance. An optical vortex phase element was fabricated, using additive lithography, with an optimized process to achieve minimal surface roughness. Shipley S1827 photoresist was used in order to obtain the appropriate additive lithography dynamic range for the desired phase profile. We investigated the effects of both postapplied and postexposure baking processes, bias exposure dose, as well as the effects of surfactant in the developer. We found the resist surface roughness to be a function of both the temperature and the time of the postapplication baking cycles, as well as the developer surfactant content. Based on our findings, an empirical correlation model was constructed to relate the process parameters with surface roughness measured quantities. The maximum roughness of the optical surface, for the optimized process, was reduced to 40 percent of the value for the unoptimized process and the additive lithography useful exposure range was increased by 10 percent.