Large-scale variability in macroturbulence driven by pulsations in the rapidly rotating massive star Zeta Oph from high-cadence ESPRESSO spectroscopy and TESS photometry

TL;DR

This study investigates the origin of macroturbulence in the rapidly rotating massive star Zeta Oph by combining high-cadence ESPRESSO spectroscopy with TESS photometry. The authors show that pulsations induce large, time-dependent, line-specific macroturbulent broadening, with peak-to-peak variations up to $v_{\rm macro} \approx 88$ km s$^{-1}$ on 30 s timescales and substantial biases for longer integrations. They identify multiple non-radial pulsation periods that are reflected in both spectroscopic diagnostics ($v\sin i$, $v_{\rm macro}$, and RV) and in the TESS light curve, including periods near $3.33$ h, $2.43$ h, $4.63$ h, and $5.34$ h, with the $2.01$ h signal interpreted as an alias of the true $2.43$ h period. The work highlights that time-dependent macroturbulence can bias atmospheric parameter inferences and stresses the need for pulsation-informed, time-dependent macroturbulence models in the analysis of massive stars.

Abstract

Despite their importance, the dynamical properties of massive stars remain poorly understood. Rotation is a key driver of internal mixing and angular momentum transport, significantly influencing massive star evolution. However, constraining rotation from spectroscopy is challenging, as spectral lines often exhibit excess broadening beyond rotation. The origin of this additional broadening, typically attributed to large-scale velocity fields and commonly referred to as macroturbulence, remains uncertain and unconstrained. Here, we present the combined analysis of TESS photometry and rapid time-series spectroscopy using the high-resolution ESPRESSO instrument at the Very Large Telescope of the European Southern Observatory for the rapidly rotating and pulsating massive star Zeta Ophiuchi. Leveraging excellent temporal coverage, our analysis demonstrates that pulsation-induced variability leads to peak-to-peak scatter as large as 88 km/s in the observed macroturbulent velocity time series. We also demonstrate that time-averaged macroturbulent velocities are spectral line specific and can exceed 100 km/s . Furthermore, the macroturbulent velocities from shorter integration times are systematically lower than those derived from stacked spectra mimicking longer exposure times typically needed for fainter stars. These results highlight the role of pulsations in driving variable macroturbulence in massive stars, while also pointing to a potential bias in spectroscopic estimates of macroturbulence for fainter massive stars.

Figures (12)

• Figure 1: Left panels: Three spectra from 9 June 2024 showing LPV for the He i+ ii$\lambda$4026, He ii$\lambda$4200, and He i$\lambda$4922 lines from top to bottom, respectively, with the stacked mean spectrum overplotted in dark blue. Right panels: Residual dynamic spectra for the corresponding left panels.
• Figure 2: RV time series for three helium lines in $\zeta$ Oph extracted using cross correlation. The peak-to-peak RV variations exceed 20 km s$^{-1}$ in a multi-periodic manner because of pulsations rather than binarity.
• Figure 3: Left panel: Fourier transform of the line profile for different wavelength windows. The black line corresponds to the first minimum of the Fourier transform for the wavelength window shown in the right-hand panel. The coloured vertical dashed lines represent the first minimum of the Fourier transform obtained by adjusting the base window by $\pm0.1$Å the line profile. Right panel: line profile for the He i+ ii$\lambda$4026 line in the stacked spectrum, with the selected wavelength window highlighted in orange.
• Figure 4: Left panel: $v\,\sin\,i$ time series for different spectral lines with peak-to-peak scatter up to 15 km s$^{-1}$. Right panel: Kernel density estimation of the scatter used to estimate the formal uncertainty in $v\,\sin\,i$.
• Figure 5: Macroturbulent velocity (i.e. $v_{\rm macro}$) for three helium lines in $\zeta$ Oph, with an x-axis normalised to the time stamp of the first spectrum (i.e. MJD = 60468.983771). Top panel: time series of all 1706 spectra with 30-s exposure times (in grey), which shows a peak-to-peak variability in $v_{\rm macro}$ of up to 88 km s$^{-1}$. Blue dashed lines are the mean of the $v_{\rm macro}$ values for each night, which are consistent within their uncertainties as indicated by the shaded region. Bottom panel: $v_{\rm macro}$ time series for stacked spectra emulating 1-h (black), 30-min (blue) and 15-min (red) integration times. The peak-to-peak variability is diminished for longer integration times, but the mean $v_{\rm macro}$ is larger for longer integration times, with the nightly mean values denoted as solid lines.