Characterising the O9–B1 spectral transition in the near-infrared

Near-infrared (NIR) spectral classification of late-O and early-B stars is critical for characterising young, heavily reddened clusters, as accurate spectral types underpin the determination of cluster ages. However, the O9–B1 interval remains particularly problematic, as the hydrogen and helium diagnostics that enable reliable classification at earlier O subtypes partly lose sensitivity in this range. We aim to provide a practical NIR spectral classification framework for the O9--B1 range at intermediate spectral resolution (R = łambda/ ∼ 4000). We employ a hybrid approach. First, we establish qualitative diagnostics anchored by a grid of 11 reference standards spanning the O9--B1 range, with spectral types assigned from optical spectroscopy, observed with VLT/X-shooter and processed to match VLT/KMOS spectral resolution. Second, we characterise the behaviour and consistency of these diagnostics against physical parameters using a homogeneous sample of 109 OB stars in Westerlund 1 (Wd1), observed with VLT/KMOS in the J, H, and K bands. The luminosity class is determined primarily by the wing morphology of the hydrogen lines. The hydrogen line width ratio (W_ 0.25 /W_ 0.75 ) quantifies the morphological sequence of V, III, and I, with class distributions showing 10--30% overlap at class boundaries and a significant correlation with surface gravities derived from independent CMFGEN modelling. For spectral subtype, a hierarchical scheme based on He, łambda1.163 μm, He, łambda2.058 μm, Si, łambda2.426 μm, Si, triplet emission at 1.250--1.260 μm, and hydrogen line strengths establishes the O9-B0 and B0-B1 boundaries across the luminosity classes. ii i iv iii We provide a detailed characterisation of NIR diagnostics for late-O and early-B stars at intermediate resolution, offering guidance for the classification of massive stars in obscured environments where optical spectroscopy is precluded.