Evidence of Increased Macroturbulence for Betelgeuse during Great Dimming

High-resolution and high signal-to-noise ratio spectrum of Betelgeuse (HD39801) observed with fiber-fed cross dispersed echelle spectrograph HARPS (R = 115 000) installed on the 3.6 m telescope at the European Southern Observatory (ESO) on 12 February 2020 (ADP.2020-02-13T01:08:03.103; Sbordone et al 2020) was compared with two spectra (ADP.2013-12-10T15:51:42.627, ADP.2017-10-25T14:11:45.978) observed with ESO VLT UVES on 2013 October 5 and 2002 September 13, respectively. All spectra are downloaded from the ESO Science Archive. The observed spectra are fitted with synthesized examples which were generated using MARCS spherical model atmospheres for T_eff = 3800, 3700, 3600, 3500, and 3400 K. Best fits in the wavelength region around the TiO bandhead at 6158 Å are given in Figure 1.

Zoom In Zoom Out Reset image size

First, we compared three observed spectra with each other and with the spectrum calculated for nominal atmospheric parameters of Betelgeuse over entire observed wavelength region with the main goal to localize level of continuum which is hardly depressed by large number of TiO lines. We adopted the nominal atmospheric parameters (T_eff = 3600 K, log(g) = 0.0 (cgs), ξ_t = 4 km s⁻¹; Tsuji 2000) and the solar abundances for all elements except¹ $\mathrm{log}\,\epsilon ({\rm{C}},{\rm{N}},{\rm{O}})$ = 8.4, 8.6, 8.8 (Lambert et al 1984). The synthetic spectra are calculated with code SYNTH3² (Kochukhov 2007) and are convolved with Gausian profile to mimicry the line broadening. The MARCS spherical LTE model atmospheres for the stellar mass of 15M_☉ and ξ_t = 5 km s⁻¹ are downloaded from the MARCS homepage³ (Gustafsson et al 2008). We included in the line list all lines above the intensity threshold 0.0001 (0.01 percent) from the VALD3 database using the "extract stellar" option (Piskunov et al 1995; Kupka et al 1999).

Comparison of the observed and synthesized spectra over entire wavelength region shows that the continuum is not seen in the observed high-resolution spectra due to quasicontinuous opacity of TiO lines. Less contaminated regions in the spectra (so-called points of pseudo-continuum) located blueward of TiO bandheads were selected for the continuum setting, e.g., at 4802.3, 4950.8, 5443.7, 5755.8, 6131.2, 6144.6, and 6467.8 Å. We normalized all the spectra in a similar way by polinomal approximation of these points of pseudo-continuum. The pseudo-continuum in the wavelength region from 6120 to 6400 Å, for example, rested on three points due to depression by broad TiO band redward of about 6158 Å. The shape of synthesized spectrum for the nominal atmospheric parameters of Betelgeuse in general agrees well with the spectra observed before the Great Dimming; we detected only minor changes between spectra observed on 2002 and 2013 (see, for example, Panel A; Figure 1). However, changes in the intensity and shape of lines are found to be significant between spectra observed on 2013 October and 2020 February during the Great Dimming Event (Panel B; see Figure 1). The atomic lines (those dominate blueward of the TiO bandhead) are shallow and broad during the Great Dimming in comparison with those observed on 2013. On the other hand, the TiO lines (those dominate redward of about 6158 Å) are stronger in comparison with those observed on 2013 October. The profile of less blended Ti I line at 6126.22 Å in radial velocity scale is shown on the right corner of Panel B, illustrating changes in the line profile.

Second, we compared the spectrum observed on 2013 October 5 with synthesized spectra calculated for MARCS models with T_eff = 3800, 3700, 3600 K and found the best fit for T_eff = 3600 K (Panel C; see Figure 1). The synthesized spectrum, which was convolved with Gaussian of R = λ/Δλ = 17,000, reproduces well the shape and intensity of selected (less blended) atomic lines. However, the intensity of TiO lines those dominate redward of the bandhead are too strong for T_eff = 3600 K model in support of slightly higher (between 3600 and 3700 K) temperature. Approximation of the less blended line profiles leads to the macroturbulent velocity of ν_macro ≃ 14 km s⁻¹, adopting $v\,\sin \,i=5.47$ km s⁻¹ (Kervella et al 2018). In addition, we inspected the profile of [O I] line at 6363.776 Å in all spectra. The oxygen line is clearly seen in the spectra observed on 2002/2013 and the fit for MARCS model with T_eff = 3600 K and $\mathrm{log}\,\epsilon (O)=8.8$ is shown on the right corner of Panel C. The oxygen abundance calculated by Lambert et al (1984) seems to be correct.

Finally, the effort was made to reproduce the spectrum of Betelgeuse observed during the Great Dimming Event. The best fit was found for the MARCS model with T_eff = 3500 K (Panel D; see Figure 1). The calculated spectrum was convolved with Gaussian of R = 12,000. We conclude that the intensity and shape of selected less blended atomic lines are reproduced well with this model. In addition, we found a good agreement for the intensity of TiO lines redward of the bandhead. The difference between spectra calculated for T_eff = 3500 and 3600 K is significant (Panel D; see Figure 1). Thus, T_eff of Betelgeuse during the Great Dimming seems to be lower than calculated before, T_eff = 3600 ± 25 K (Levesque & Massey 2020), in support of lower value predicted using photometry by Harper et al (2020). The resolution of spectra used by Levesque & Massey (2020) is too low for quantitative analysis of the crowded spectra of cool stars. Approximation of the less blended profiles of atomic lines in the high-resolution (R = 115,000) spectrum used by us leads to the macroturbulent velocity ν_macro ≳ 23 km s⁻¹, adopting $v\,\sin \,i=5.47$ km s⁻¹. Larger uncertainty in the estimated macroturbulent velocity is because of increased blending of line profiles by weak TiO lines. Note, that the oxygen line at 6363.776 Å is not measurable (smoothed) in the spectrum observed on February 2020 due to increased both macrourbulence and blending by weak TiO lines.