Ultraviolet spectroscopy reveals a hot and luminous companion to the Be star + black hole candidate MWC 656

TL;DR

This study re-evaluates the nature of the companion in the Be+BH candidate MWC 656 by combining new high-resolution optical spectroscopy with multi-epoch ultraviolet spectroscopy from HST. A revised orbital solution and spectral disentangling reveal a low-mass companion (approximately $1.1$–$1.9\ M_\odot$) inconsistent with a black hole, while UV diagnostics require a hot, luminous stripped-star companion with $T_{*,2}\approx85\ pm 10$ kK and $\log L_2/\mathrm{L}_\odot\approx4.0$. A joint Be-star + stripped-star composite model successfully reproduces the UV wind features and He II emission, with the stripped companion contributing $\sim10$–$15\%$ of the UV flux and a wind characterized by $\dot M_2\approx10^{-7.5}\ M_\odot\,\mathrm{yr}^{-1}$ and $v_{\infty}\approx400\ \mathrm{km\ s^{-1}}$. The findings place MWC 656 among Be+stripped-star binaries, reduce the tension with BH population models, and reinforce the rarity of confirmed Be+BH systems in the Galaxy. The results have important implications for massive binary evolution, Be-star disc interactions, and the calibration of population-synthesis predictions for compact-object companions.

Abstract

The Galactic Be star binary MWC 656 was long considered the only known Be star + black hole (BH) system, making it a critical benchmark for models of massive binary evolution and for the expected X-ray emission of Be+BH binaries. However, recent dynamical measurements cast doubt on the presence of a BH companion. We present new multi-epoch ultraviolet spectroscopy from the Hubble Space Telescope (HST), combined with high-resolution optical spectra, to reassess the nature of the companion. The far-ultraviolet spectra reveal high-ionisation features, including prominent N V and He II lines, which are absent in the spectra of normal Be stars and are indicative of a hot, luminous companion. Spectral modelling shows that these features cannot originate from the Be star or from an accretion disc around a compact object. Instead, we find that the data are best explained by a hot ($T_\mathrm{eff} \approx 85$ kK), compact, hydrogen-deficient star with strong wind signatures, consistent with an intermediate-mass stripped star. Our revised orbital solution and composite spectroscopic modelling yield a companion mass of $M_2 = 1.54^{+0.57}_{-0.46}\,\mathrm{M}_\odot$, definitively ruling out a BH and disfavouring a white dwarf. MWC 656 thus joins the growing class of Be + stripped star binaries. The system's unusual properties - including a high companion temperature and wind strength - extend the known parameter space of such binaries. The continued absence of confirmed OBe+BH binaries in the Galaxy highlights a growing tension with population synthesis models.

Figures (14)

• Figure 1: Dynamical spectra of MWC 656 shown over two orbital cycles. From left to right, the panels display the Hei $\lambda4026$, Hei $\lambda6678$, and Heii $\lambda4686$ lines. Two epoch spectra near quadrature phases are overplotted in black to further illustrate line profile variations. The orbital RV curves of the Be star (pink) and the companion (white), derived from the best-fit orbit model, are overlaid and trace the associated absorption and emission features.
• Figure 2: Phase-folded RV curves of the Be star and its companion in the MWC 656 binary system, using the best-fit orbital period. RV measurements and uncertainties of the Be star (circles) and the companion (squares) are shown with colour-coded markers corresponding to the instrument used. Maximum-likelihood orbital models derived from RV fitting are overplotted for the Be star (pink) and the companion (green) with thick lines. Thin lines show orbital models for random samples from the posterior to illustrate uncertainties in the inferred parameters. The grey dotted line shows the inferred systemic velocity. Star symbols indicate the predicted orbital phases and velocities of the four HST/STIS FUV observations based on the fitted orbit.
• Figure 3: Disentangled spectra of MWC 656 near quadrature, showing spectral regions where the companion exhibits non-flat continuum features. From left to right, the panels display the Heii $\lambda$4686 emission line, Balmer H$\beta$, and Hei $\lambda$6678. Observed spectra near quadrature phases are plotted in black (top and bottom rows). The individual Doppler-shifted components of the Be star (pink) and the companion (green), obtained from spectral disentangling, are overplotted together with their summed model (light pink).
• Figure 4: HST/STIS UV spectra of MWC 656 obtained on four observing dates. A PoWR stellar atmosphere model with $T_* = 21$ kK, $\log g_* = 3.4$, and $v\sin i = 300$ km s$^{-1}$ is shown in black for comparison. The top and bottom panels show the full spectral range, split into blue and red halves, while the middle row displays zoom-ins on key diagnostic lines: the Nv $\lambda\lambda$1238, 1241 doublet, the Civ $\lambda\lambda$1548, 1551 doublet, and the Heii $\lambda$1640 line. Shaded gray regions mark known ISM absorption features. The zoom-in panels reveal prominent high-ionisation features, not present in the Be star model, suggesting they originate from a hotter source within the system.
• Figure 5: Observed spectrum of MWC 656 (grey) compared to synthetic Be star spectra computed for different X-ray luminosities, from no X-rays up to $10^{32}$ erg s$^{-1}$. Our fiducial model assumes $L_\mathrm{X} = 10^{30}$ erg s$^{-1}$, consistent with observations Ribo+2017. The HST spectrum has been rebinned to 3 pixels for clarity. Only the $10^{32}$ erg s$^{-1}$ model produces Nv emission comparable to the data, but it simultaneously generates Siiv and Civ features that are not observed.