We investigate phase-2 quantum optimal control of selective conversion in a structured medium designed to suppress the macroscopic limitations identified previously in untreated bulk media. Building on the Maxwell-Bloch-Lindblad-thermal adjoint formalism established in phase 1, we incorporate thin-filmization and active cooling into a one-dimensional structured-medium simulator and optimize the boundary control using analytical adjoint gradients combined with L-BFGS-B. In contrast to untreated bulk propagation, the structured configuration strongly suppresses propagation-induced attenuation and thermally amplified decoherence.However, despite the removal of these macroscopic loss channels, the optimized target-state population saturates at 0.915135 under a realistic control-energy penalty λ_energy = 10u207bu2074. This saturation indicates that, once macroscopic propagation and thermal bottlenecks are relieved, the dominant limitation shifts to the microscopic spontaneous decay of the intermediate state, here characterized by Γ = 0.5. The optimizer therefore converges not to complete transfer but to a resource-constrained Pareto-optimal compromise between rapid passage and radiative loss.These results establish a hierarchical picture of control limitations: untreated bulk media are limited primarily by transport and thermal feedback, whereas structured media reveal a second ceiling set by microscopic dissipation under finite laser power. We conclude that further progress toward near-unity transfer will require not only pulse optimization but also microscopic channel engineering, such as suppression of the lossy decay pathway, enhancement of effective Raman coupling, or cavity-enabled reservoir engineering.