2002MNRAS.336..637L

We present the results of an attempt to detect the hottest `hot Jupiter' planets directly in the thermal infrared. A simple method based upon high signal-to-noise ratio spectroscopy of the central star at low spectral resolution is described. In the 2-4 µm region the contrast ratio between planet and star is expected to be relatively low and the planetary spectrum should appear as a faint signal on top of the stellar spectrum, distinguished by edges of H₂O absorption. No water edges were found to 3σ limits of one part in a few hundred in each case. These upper limits are compared with the irradiated planetary atmosphere models of Barman, Hauschildt & Allard to derive upper limits on the size of the hot Jupiters, which are expected to be somewhat larger than Jupiter. If reasonably strong H₂O absorption occurs in these objects then typical upper limits of R < 3 R_Jupare derived, the precision being limited by the stability of telluric transmission. Only a modest improvement in precision is needed (e.g. with space-based instruments) to reach the range of greatest interest (1 < R < 2 R_Jup).