Abstract

Aims. We present a 0'5 -resolution 17.7 mu m image of the red supergiant Antares. Our aim is to study the structure of the eircumstellar envelope in detail. Methods. Antares was observed at 17.7 mu m with the VLT mid infrared instrument VESIR. Taking advantage of the BURST mode, in which a large number of short exposure frames are taken, we obtained a diffraction limited image with a spatial resolution of 0'5. Results. The VISIR image shows six clumpy dust clouds located at 0''8-1''8 (43-96 R-* = 136-306 AU) away from the star. We also detected compact emission within a radius of 0''5 around the star. Comparison of our VISIR image taken in 2010 and the 20.8 mu m image taken in 1998 with the Keck Telescope reveals the outward motions of four dust clumps. The proper motions of these dust dumps (with respect to the central star) amount to 0''2-0''6 in 12 years. This translates into expansion velocities (projected onto the plane of the sky) of 13-40 km s(-1) with an uncertainty of +/- 7 km s(-1). The inner compact emission seen in the 2010 VISIR image is presumably newly formed dust. because it is not detected in the image taken in 1998. If we assume that the dust is ejected in 1998, the expansion velocity is estimated to be 34km s(-1). in agreement with the velocity of the outward motions of the clumpy dust clouds. The mass of the dust clouds is estimated to be (3-6) x 10(-9) Ma. These values are lower by a factor of 3-7 than the amount of dust ejected in one year estimated from the (gas dust) mass-loss rate of 2 x 10(-6) M-circle dot yr(-1), suggesting that the continuous mass loss is superimposed on the clumpy dust cloud ejection. Conclusions The clumpy dust envelope detected in the 17.7 mu m diffraction limited image is similar to the clumpy or asymmetric circumstellar environment of other red supergiants. The velocities of the dust clumps cannot be explained by a simple accelerating outflow, implying the possible random nature of the dust cloud ejection mechanism.