As a bright, colorful and variable star, Betelgeuse has attracted popular interest from the earliest times. As a prototypical evolved supergiant, it has been investigated with many photometric and spectroscopic techniques. With hundreds of focused studies, many measurements remain puzzling after decades of reflection. We begin our exploration of the Betelgeuse story with an illustrated historical tour of observations and some thoughts on interpretive paths.

We discuss the physics and the evolution of a typical massive star passing through an evolutionary stage similar to that of Betelgeuse. After a brief introduction recalling various observed parameters of Betelgeuse, we discuss the Pre-Main-Sequence phase (PMS), the Main-Sequence (MS) phase, the physics governing the duration of the first crossing of the HR diagram, the red supergiant stage (RSG), the post-red supergiant phases and the final fate of solar metallicity stars with masses between 9 and 25 M⊙. We examine the impact of different initial rotation and of various prescriptions for the mass loss rates during the red supergiant phase. We show that, whatever the initial rotation rate (chosen between 0 and 0.7 × υcrit, υcrit being the surface equatorial velocity producing a centrifugal acceleration balancing exactly the gravity) and the mass loss rates during the RSG stage (varied between a standard value and 25 times that value), a 15 M⊙ star always ends its lifetime as a RSG and explodes as a type II-P or II-L supernova.

We review the significant role played by red supergiants (RSGs) in stellar populations, and some challenges and questions they raise for theoretical stellar evolution. We present how metallicity and rotation modify the way stars go to the red part of the Hertzsprung-Russell diagram or come back from it, and how RSGs might keep a trace of their main-sequence evolution. We compare theoretical population ratios with observed ones.

Mass-loss rates are one of the most relevant parameters determining the evolution of massive stars. In particular, the rate at which the star loses mass during the red-supergiant (RSG) phase is the least constrained by the observations or theory. In this paper, we show how the mass loss during the RSG phase affects the later evolution of the star, as well as the final type of supernova towards which it leads. We also discuss some possibilities to discriminate between blue stars that went through a RSG phase and those which remained in the blue part of the Hertzsprung-Russell diagram.

Red supergiants (RSG) are a crucial phase in the evolution of massive stars. At solar metallicity, they represent the stage where substantial amounts of mass and angular momentum are lost. Here we investigate the role of RSGs at very low metallicity. We have computed evolutionary tracks at Z=0.0004 with and without rotation, using the latest version of the Geneva stellar evolution code. Here, we briefly discuss the evolution of RSGs, their lifetimes, mass ranges for forming RSGs, maximum luminosity of RSGs, and how these are affected by rotation.

Multi-wavelength studies of M supergiants have revealed atmospheric structures with a large range of spatial and temporal scales. Focusing on Betelgeuse, these scales and their perplexing connections from the photosphere to the interstellar medium are reviewed. Of particular current interest is the dynamic origin of the ubiquitous and relatively dust-free mass loss. Is it multiple plumes of convection driven ejecta, episodic ejection of molecular reservoirs, or a more steady and uniform flow? With powerful new facilities such as the VLT and ALMA we may begin to understand the connections and answer such puzzles, but ultimately detailed studies of a sample of M supergiants will be needed to disentangle the physics from the stars’ personalities.

We have re-appraised the temperatures of Red Supergiants (RSGs) in the Magellanic Clouds, by studying their spectral energy distributions (SEDs) from 400–2500 nm using VLT+XSHOOTER, in conjunction with MARCS model atmospheres. We determine temperatures using 3 methods: from model fits to the TiO bands in the optical; from model fits to the SED using the line-free continuum in the near-infrared; and from the integrated fluxes. We find that the temperatures from the TiO fits are systematically lower that those from the other methods by several hundred Kelvin. The TiO fits also dramatically over-predict the flux in the near-IR, and imply extinctions which are anomalously low compared to neighbouring stars. In contrast, the SED temperatures provide good fits to the fluxes at all wavelengths other than the TiO bands, are in agreement with the temperatures from the flux integration method, and imply extinctions consistent with nearby stars. We consider a number of ways to reconcile this discrepancy, concluding that 3-D effects are responsible, and that RSG temperatures are much warmer than previously thought.

Direct images of Betelgeuse were obtained over a span of 4 years with the Faint Object Camera on the Hubble Space Telescope. These images reveal the extended ultraviolet continuum emission (~2 times the optical diameter), the varying overall ultraviolet flux levels and a pattern of bright surface continuum features that change in position and appearance over several months or less. Concurrent photometry and radial velocity measures support the model of a pulsating star, first discovered in the ultraviolet from IUE. Spatially resolved HST spectroscopy reveals a larger extention in chromospheric emissions of Mg II as well as the rotation of the supergiant. Changing localized subsonic flows occur in the low chromosphere that can cover a substantial fraction of the stellar disk and may initiate the mass outflow.

3-D simulations suggest that the atmospheres of red supergiants are subject to large-amplitude convective motions, which are suspected to generate supersonic motions and shocks. We perform tomography of supergiant-star atmospheres, on temporal series of high-resolution spectra and on 3-D hydrodynamical synthetic spectra. The tomographic technique is improved by the computation of the contribution function, which is used to construct numerical spectral masks probing different optical depths. This exploratory work allows us to put spatial and temporal constraints on velocity fields predicted by 3-D model atmospheres.

We present studies of the atmospheric structure and fundamental properties of the red supergiants (RSGs) VY CMa, AH Sco, UY Sct, and KW Sgr based on near-infrared K-band interferometry obtained with the VLTI/AMBER instrument with a spectral resolution of 1500. In our visibility data, we observe the presence of molecular layers of water and CO in extended atmospheres. For a uniform disk modeling, we observe size increases in the water band centered at 1.9 μm and in the CO band at 2.3–2.5 μm, with respect to the near-continuum bandpass (2.20–2.25 μm). With our spectral resolution, we obtain diameters in the near-continuum, that are free from contamination by molecular layers. Using PHOENIX atmosphere models, we estimate Rosseland-mean photospheric angular diameters of VY CMa, AH Sco, UY Sct, and KW Sgr of 11.13 ± 0.3mas, 5.81 ± 0.15mas, 5.48 ± 0.10mas, and 3.91 ± 0.25mas, respectively. We estimate radii and effective temperatures, and place the stars in the HR diagram.

The PHOENIX atmosphere models predict the spectra and the continuum visibility values, but do not predict the molecular layers visibility well: The model atmosphere is too compact when compared with the observations. This may be caused by pulsation and/or convection, which are not included in the models.

Red supergiants with their enormous brightness at J-band are ideal probes of cosmic chemical composition. It is therefore crucial to have realistic models of radiative transfer in their atmospheres, which will permit determination of abundances accurate to 0.15 dex, the precision attainable with future telescope facilities in galaxies as distant as tens of Mpc. Here, we study the effects of non-local thermodynamic equilibrium (NLTE) on the formation of iron, titanium, and silicon lines, which dominate J-band spectra of red supergiants. It is shown that the NLTE radiative transfer models enable accurate derivation of metallicity and effective temperature in the J-band. We also discuss consequences for RSG spectrum synthesis in different spectral windows, including the heavily TiO-blanketed optical region, and atmospheric structure. We then touch upon challenges of NLTE integration with new generation of 3D hydrodynamical RSG models and present the first calculations of NLTE spectra with the mean 3D model of Betelgeuse.

The interpretation of water lines in red supergiant stellar atmospheres has been much debated over the past decade. The introduction of the so-called MOLspheres to account for near-infrared “extra” absorption has been controversial. We propose that non-LTE effects should be taken into account before considering any extra-photospheric contribution.

After a brief introduction on the radiative transfer treatment and the inadequacy of classical treatments in the case of large-scale systems such as molecules, we present a new code, based on preconditioned Krylov subspace methods. Preliminary results suggest that NLTE effects lead to deeper water bands, as well as extra cooling.

The mass-loss mechanism in red supergiants is a long-stand-ing problem. The milliarcsecond angular resolution achieved by infrared long-baseline interferometry provides us with the only way to spatially resolve the region where the material is accelerated. For this goal, the 2.3 μm CO lines are important, because they form in the upper photosphere and the outer atmosphere (so-called MOLsphere). We present high-spatial and high-spectral resolution observations of the 2.3 μm CO lines in the red supergiants Betelgeuse and Antares using the Very Large Telescope Interferometer (VLTI). This has enabled us to spatially resolve the gas dynamics in the photosphere (and the MOLsphere) for the first time other than the Sun. We have detected vigorous motions of large CO gas clumps with velocities of up to 20–30 km s-1. Comparison of the CO line data taken 1 year apart shows a significant change in the dynamics of the atmosphere. In contrast to the CO line data, the continuum data reveal no or only marginal time variations. The observationally estimated gas density in the outer atmosphere at 1.3–1.4 R⋆ is higher than the values predicted by the current 3-D convection simulations by 6 to 11 orders of magnitude. Therefore, at the moment, convection alone cannot explain the detected vigorous gas motions in the extended outer atmosphere of Betelgeuse and Antares.

Although the contribution of a chromosphere to the total flux may be small, it plays a vital role as the interface between the star and interstellar space, as it is where the stellar wind originates. The very outermost layers of a star are expected to be turbulent. Images of the solar chromosphere and corona reveal both small-scale inhomogeneities (prominences and spaces) and large-scale variability (polar plumes near sunspot minima, streamers near sunspot maxima), and something similar but more exaggerated can be expected in cool supergiants. Samplings of the high chromosphere in late-K supergiants show extreme variability in both density and velocity, and can be thought of as analogues of the solar case. Series of chromospheric-eclipse spectra of the Ca ii K line in 31 Cyg, 32 Cyg and ζAur demonstrate (a) the presence in the high chromosphere of discrete, rapidly-moving clumps of gas, (b) that structures are not stable or symmetrical, either from eclipse to eclipse or from ingress to egress in the same eclipse, (c) plenty of empty space, and (d) huge differences between one late-K supergiant and another. What information can this evidence offer as regards the outer structure and wind of Betelgeuse?

The small-scale surface granulation on cool main-sequence stars and white dwarfs influences the overall appearance of these objects only weakly. And it is only indirectly observable by analyzing e.g. line-shapes or temporal fluctuations – except for the Sun. The large-scale and high-contrast convective surface cells and accompanying sound waves on supergiants and low-gravity AGB stars on the other hand have a strong impact on the outer atmospheric layers and are directly detectable by interferometric observations.

Necessary to interpret modern observations with their high resolution in frequency, time, and/or space are detailed numerical multi-dimensional time-dependent radiation-hydrodynamical simulations. Local simulations of small patches of convective surface layers and the atmosphere of main-sequence stars have matured over three decades and have reached an impressive level of agreement with observations and also between different computational codes. However, global simulations of the entire convective surface and atmosphere of a red supergiants are considerably more demanding – and limited – and have become available only for about one decade.

Still, they show how the surface is shaped by the interaction of small surface granules, that sit on top of large envelope convection cells, and waves, that can travel as shocks into the outer atmosphere. The route to more complete future models will be discussed, that comprise the outer atmosphere of the stars and that could explain some of the little-understood phenomena like chromosphere, molsphere, or wind-formation.

As red supergiants are the largest and brightest stars, they are ideal targets for the new generation of sensitive, high resolution instrumentation that provides spectrophotometric, interferometric, astrometric, and imaging observables. The interpretation of the complex stellar surface images requires numerical simulations of stellar convection that take into account multi-dimensional time-dependent radiation hydrodynamics with realistic input physics. We show the results obtained with the synergy between the radiative hydrodynamics code CO5BOLD and the post-processing radiative transfer code Optim3D. Such simulations support a proper and quantitative analysis of these observations, and the observations provide the theoretical work with strong constraints.

We investigate the occurrence of water vapour signatures in a total of 10 red giants in the solar neighbourhood at mid-infrared wavelengths (12 μm). With the use of high resolution spectra from TEXES and synthesized spectra based on MARCS model atmospheres, we analyse the differences and discuss plausible causes. These include abundance adjustments, the addition of non-photospheric components (MOLspheres) and a different temperature profile.

We report on a long-term monitoring of the cool supergiant Betelgeuse, using the NARVAL and ESPaDOnS high-resolution spectropolarimeters, respectively installed at Telescope Bernard Lyot (Pic du Midi Observatory, France) and at the Canada-France-Hawaii Telescope (Mauna Kea Observatory, Hawaii). The data set, constituted of circularly polarized (Stokes V) and intensity (Stokes I) spectra, was collected between 2010 and 2012. We investigate here the temporal evolution of magnetic field, convection and temperature at photospheric level, using simultaneous measurements of the longitudinal magnetic field component, the core emission of the Ca II infrared triplet, the line-depth ratio of selected photospheric lines and the radial velocity of the star.

We present the results of the analysis of our recent interferometric observations of Betelgeuse, using the AMBER instrument of the VLTI. Using the medium spectral resolution mode (R ~ 1500) we detected the presence of the water vapour and carbon monoxide (CO) molecules in the H and K bands. We also derived the photospheric angular diameter in the continuum. By analysing the depth of the molecular lines and the interferometric visibilities, we derived the column densities of the molecules, as well as the temperature and the size of the corresponding regions in the atmosphere of Betelgeuse (the MOLsphere) using a single shell model around the photosphere. Our results confirm the findings by Perrin et al. (2004) and Ohnaka et al. (2011) that the H2O and CO molecules are distributed around Betelgeuse in a MOLsphere extending to approximately 1.3 times the star’s photospheric radius.