Urban Air Mobility (UAM) has the potential to revolutionize transportation. It will exploit the third dimension to help smooth ground traffic in densely populated areas. This new paradigm in mobility will require methods to ensure safety and maximize efficiency. We propose to use mathematical optimization to address tactical deconfliction in UAM. Our approach is envisioned as a way of modeling and solving tactical conflicts, but also as a means for assessing future infrastructures with potential utility in the design phase. We leverage envisioned UAM corridors to provide a mathematical definition of vehicle separation. A mathematical formulation is then proposed, which minimizes the total deviation from flight schedules needed to avoid loss of pairwise separation. The deconfliction is based on both airborne adjustments (through speed changes) and ground delays (holds relative to the scheduled takeoff). Our experimental setup includes three use cases standing for different sources of conflicts and three synthetic UAM network topologies, which represent heterogeneous realistic scenarios. Vehicle and network capabilities are represented through model parameters, which allows us to analyze their impact on the quality of the deconfliction. Finally, insightful comparisons between our approach and both a local deconfliction and a fairness-oriented version are provided.