Context. The upper stellar mass limit is a fundamental parameter for simulations of star formation, galactic chemical evolution, and stellar feedback. An empirical bound on this parameter is therefore highly valuable. The most massive stars known to date are R 136 a1, a2, a3, and c, with reported masses in excess of 150–200 M⊙ and initial masses of up to ≈300 M⊙. They are located within the central cluster R 136a of the Tarantula nebula in the Large Magellanic Cloud (LMC), However, the mass estimation of these stars relies on the assumption that they are single.
Aims. Via multi-epoch spectroscopy, we provide, for the first time, constraints on the presence of close stellar companions to the most massive stars known for orbital periods of up to ≈10 yr.
Methods. We collected three epochs of spectroscopy for R 136 a1, a2, a3, and c with the Space Telescope Imaging Spectrograph (STIS) of the Hubble Space Telescope (HST) in the years 2020–2021 to probe potential radial-velocity (RV) variations. We combined these epochs with an additional HST/STIS observation taken in 2012. For R 136 c, we also used archival spectroscopy obtained with the Very Large Telescope (VLT). We used cross-correlation to quantify the RVs and establish constraints on possible companions to these stars up to periods of ≈10 yr. Objects are classified as binaries when the peak-to-peak RV shifts exceed 50 km s−1 and when the RV shift is significant with respect to errors.
Results. R 136 a1, a2, and a3 do not satisfy the binary criteria and are thus classified as putatively single, although formal peak-to-peak RV variability on the level 40 km s−1 is noted for a3. Only R 136 c is classified as a binary, in agreement with the literature. We can generally rule out massive companions (M2 ≳ 50 M⊙) to R 136 a1, a2, and a3 out to orbital periods of ≲1 yr (separations ≲5 au) at 95% confidence, or out to tens of years (separations ≲100 au) at 50% confidence. Highly eccentric binaries (e ≳ 0.9) or twin companions with similar spectra could evade detection down to shorter periods (≳10 days), though their presence is not supported by the relative X-ray faintness of R 136 a1, a2, and a3. We derive a preliminary orbital solution with a 17.2 days period for the X-ray-bright binary R 136 c, though more data are needed to conclusively derive its orbit.
Conclusions. Our study supports a lower bound of 150–200 M⊙ on the upper-mass limit at LMC metallicity.