Selective vaporization during laser powder bed fusion (LPBF) can shift the composition of Ni-based superalloys outside specification limits, affecting microstructure, heat-treatment response, and certifiability. This study combines a calibrated normalized enthalpy (NE) framework with melt pool geometry, recoil pressure analysis, and thermodynamic evaporation modeling to interpret selective composition changes in IN738LC across a 23-condition LPBF process matrix. Melt pool geometries were used to determine effective absorptivity, yielding an \(\eta\) -corrected \(\textrm{NE}\) that classified the process window into conduction, stable-keyhole, and unstable-keyhole regimes. The conduction–keyhole transition occurred near \(\Phi \approx 8.5,\) while unstable keyhole behavior emerged above \(\textrm{NE}\approx 12\) . A dimensionless recoil number correlated strongly with melt pool depth, supporting its use as a compact descriptor of cavity morphology. Thermo-Calc activities combined with pure-species vapor pressures and the Hertz–Knudsen relation indicated preferential loss of volatile species, consistent with the measured depletion of Cr, Al, Si, B, and Co. Cr and Al decreased on average by approximately 0.39 and 0.30 at%, respectively, whereas Si and B showed the largest relative losses. Fe and Ni were correspondingly enriched, while Co showed slight average depletion. However, bulk composition showed weak or non-monotonic dependence on \(\textrm{NE}\) for several elements, consistent with near-boiling surface-temperature saturation and dilution by repeated multilayer remelting toward a quasi-steady state. Larger deviations occurred mainly in the unstable keyhole regime, where recoil fluctuations and melt pool instability amplified composition scatter. The combined \(\textrm{NE}\) –recoil–evaporation framework enables systematic LPBF process-window mapping while maintaining printed compositions within specification.
