Revisiting the potassium feature of WASP-31b at high resolution

The analysis and interpretation of exoplanet spectra from time-series observations remains a significant challenge to our current understanding of exoplanet atmospheres, due to the complexities in understanding instrumental systematics. Previous observations of the hot Jupiter WASP-31b using transmission spectroscopy at low-resolution have presented conflicting results. Hubble Space Telescope (HST) observations detected a strong potassium feature at high significance (4.2\sigma), which subsequent ground-based spectro-photometry with the Very Large Telescope (VLT) failed to reproduce. Here, we present high-resolution observations (R>80,000) of WASP-31b with the UVES spectrograph, in an effort to resolve this discrepancy. We perform a comprehensive search for potassium using differential transit light curves, and integration over the planet's radial velocity. Our observations do not detect K absorption at the level previously reported with HST, consistent with the VLT observations. We measure a differential light curve depth $\Delta F = 0.00031 \pm 0.00036$ using 40\AA\ bins centred on the planet's K feature, and set an upper limit on the core line depth of $\Delta F \leq 0.007$ (3\sigma) at a few times the resolution limit ($\approx0.24\AA$). These results demonstrate that there are still significant limitations to our understanding of instrumental systematics even with our most stable space-based instrumentation, and that care must be taken when extracting narrow band signatures from low-resolution data. Confirming exoplanet features using alternative instruments and methodologies should be a priority, and confronting the limitations of systematics is essential to our future understanding of exoplanet atmospheres.