Abstract

Aims. We present a unique multi-epoch infrared interferometric study of the oxygen-rich Mira variable RR Aql in comparison to radiative transfer models of the dust shell. We investigate flux and visibility spectra at 8-13 mu m with the aim of better understanding the pulsation mechanism and its connection to the dust condensation sequence and mass-loss process. Methods. We obtained 13 epochs of mid-infrared interferometry with the MIDI instrument at the VLTI between April 2004 and July 2007, covering minimum to pre-maximum pulsation phases (0.45-0.85) within four cycles. The data are modeled with a radiative transfer model of the dust shell where the central stellar intensity profile is described by a series of dust-free dynamic model atmospheres based on self-excited pulsation models. We examined two dust species, silicate and Al2O3 grains. We performed model simulations using variations in model phase and dust shell parameters to investigate the expected variability of our mid-infrared photometric and interferometric data. Results. The observed visibility spectra do not show any indication of variations as a function of pulsation phase and cycle. The observed photometry spectra may indicate intracycle and cycle-to-cycle variations at the level of 1-2 standard deviations. The photometric and visibility spectra of RR Aql can be described well by the radiative transfer model of the dust shell that uses a dynamic model atmosphere describing the central source. The best-fitting model for our average pulsation phase of (Phi(V)) over bar = 0.64 +/- 0.15 includes the dynamic model atmosphere M21n (T-model = 2550 K) with a photospheric angular diameter of theta(Phot) = 7.6 +/- 0.6 mas, and a silicate dust shell with an optical depth of tau(V) = 2.8 +/- 0.8, an inner radius of R-in = 4.1 +/- 0.7 R-Phot, and a power-law index of the density distribution of p = 2.6 +/- 0.3. The addition of an Al2O3 dust shell did not improve the model fit. However, our model simulations indicate that the presence of an inner Al2O3 dust shell with lower optical depth than for the silicate dust shell can not be excluded. The photospheric angular diameter corresponds to a radius of R-phot = 520(-140)(+230)R(circle dot) and an effective temperature of T-eff similar to 2420 +/- 200 K. Our modeling simulations confirm that significant intracycle and cycle-to-cycle visibility variations are not expected for RR Aql at mid-infrared wavelengths within our uncertainties. Conclusions. We conclude that our RR Aql data can be described by a pulsating atmosphere surrounded by a silicate dust shell. The effects of the pulsation on the mid-infrared flux and visibility values are expected to be less than about 25% and 20%, respectively, and are too low to be detected within our measurement uncertainties.