Interferometric view into RT Pav's long secondary period. binary vs oscillatory convective modes

TL;DR

RT Pav’s long secondary period is tested against binary versus oscillatory convective hypotheses using multiwavelength VLTI interferometry (H, K, L, M bands) and Gaia DR3 astrometry. Parametric modeling and image reconstruction favor an oscillatory convective dipole with a ≈200 K temperature contrast over a dusty bound companion, as Gaia bounds and wavelength-dependent CP trends are inconsistent with a coherent binary signal. The H-band photosphere is resolved and supports intrinsic surface variability, while CO band-heads reveal MOLsphere structure rather than a point-like companion. The study concludes that RT Pav’s LSP is more plausibly intrinsic to the stellar surface, with time-resolved spectro-interferometry across the LSP cycle as a natural next step. The results have implications for understanding mass-loss and dust formation in evolved stars, emphasizing non-radial surface modes over binarity in this case.

Abstract

Long secondary periods (LSPs) occur in about one-third of evolved stars, yet their origin remains unclear. The leading explanations are oscillatory convective modes and a binary companion embedded in dust. We investigate the LSP of the red giant RT Pav using multi-wavelength VLTI interferometry (PIONIER, GRAVITY, MATISSE; 1.5-5.0 microns), obtained near the phase where a companion would appear most separated. These data, combined with photometry and Gaia DR3 astrometry, constrain possible companion masses, orbits, and photometric effects. We model the interferometric observables using uniform-disk, limb-darkened, ellipse, binary, and oscillatory convective dipole representations, supported by Monte Carlo simulations. Gaia limits any companion to a mass whose Roche-lobe volume is too small to hold the obscuring or scattering material required to reproduce the observed LSP modulation. While binary fits can yield low chi-squared values, the inferred positions are inconsistent across wavelength, closure phases do not increase with wavelength as dusty companions predict, and significant detections occur in only two of four bands. Theoretical estimates show that a roughly 1 percent flux companion at LSP-like separations should be consistently detectable for typical O-rich AGB dust, but is not consistently observed. In contrast, an oscillatory convective dipole with a temperature contrast of about 200 K reproduces the H-band morphology and visible-light amplitude without violating Gaia or photometric constraints, and binary-like residuals vanish when dipole models are fitted. Our results therefore favor oscillatory convective modes over a binary origin for the LSP in RT Pav. Time-resolved spectro-interferometry across the LSP cycle is a natural next step.

Figures (15)

• Figure 1: Folded light curve of RT Pav from V-band ASAS photometry.The green curve shows the detrended and phase-bin averaged curve over multiple LSP epochs.
• Figure 2: $(u,v)$ plane coverage of RT Pav observations on the VLTI for each instrument. Points are colored based on the square visibility measured at the respective $(u,v)$ coordinate
• Figure 3: Predicted maximum angular separation of a companion to RT Pav as a function of companion mass under the binary hypothesis assuming RT Pav's 757day LSP, and a primary mass of 1$M_\odot$. Curves are shown for orbital eccentricities of 0.0, 0.3, and 0.6. The thick dashed line marks the adopted Gaia DR3 astrometric reflex motion detection threshold of 0.3 mas. Companion configurations above this threshold would typically induce a detectable astrometric signal and are therefore excluded by Gaia's classification of RT Pav as a single star.
• Figure 4: Measured (center column) versus Monte Carlo simulations of the closure measured by the VLTI configurations of (left) thermal variations across the photosphere with temperature gradients $\pm$ 100K, and (right) a binary enshrouded in a large dust cloud with a temperature between 1000-1500K and a radius up to 26% the diameter of the primary. Monte Carlo simulations assume a fixed zenith target geometry without Earth rotation synthesis. The light gray shaded region highlights $\pm1\sigma$ envelope of wavelength binned closure phases across the different MC simulations. Gray colors in the center column are measurements with uncertainties.
• Figure 5: Square visibilities versus angular frequency for RT Pav on PIONIER (H band), GRAVITY (K band), MATISSE (L band), and MATISSE (M band) data.