2016A&A...589A..99L

Close-in extrasolar giant planets are expected to cool their thermospheres by producing H₃⁺ emission in the near-infrared (NIR), but simulations predict H₃⁺ emission intensities that differ in the resulting intensity by several orders of magnitude. We want to test the observability of H₃⁺ emission with CRIRES at the Very Large Telescope (VLT), providing adequate spectral resolution for planetary atmospheric lines in NIR spectra. We search for signatures of planetary H₃⁺ emission in the L' band, using spectra of HD 209458 obtained during and after secondary eclipse of its transiting planet HD 209458 b. We searched for H₃⁺ emission signatures in spectra containing the combined light of the star and, possibly, the planet. With the information on the ephemeris of the transiting planet, we derive the radial velocities at the time of observation and search for the emission at the expected line positions. We also apply a cross-correlation test to search for planetary signals and use a shift and add technique combining all observed spectra taken after secondary eclipse to calculate an upper emission limit. We do not find signatures of atmospheric H₃⁺ emission in the spectra containing the combined light of HD 209458 and its orbiting planet. We calculate the emission limit for the H₃⁺ line at 3953.0nm [Q(1,0)] to be 8.32x10¹⁸W and a limit of 5.34x10¹⁸W for the line at 3985.5nm [Q(3,0)]. Comparing our emission limits to the theoretical predictions suggests that we lack 1 to 3 magnitudes of sensitivity to measure H₃⁺ emission in our target object. We show that under more favorable weather conditions the data quality can be improved significantly, reaching 5x10¹⁶W for star-planet systems that are close to Earth. We estimate that pushing the detection limit down to 10¹⁵W will be possible with ground-based observations with future instrumentation, for example, the European Extremly Large Telescope.