HD 209458b in New Light: Evidence of Nitrogen Chemistry, Patchy Clouds and Sub-Solar Water

Interpretations of exoplanetary transmission spectra have been undermined by apparent obscuration due to clouds/hazes. Debate rages on whether weak H$_2$O features seen in exoplanet spectra are due to clouds or inherently depleted oxygen. Assertions of solar H$_2$O abundances have relied on making a priori model assumptions, e.g. chemical/radiative equilibrium. In this work, we attempt to address this problem with a new retrieval paradigm for transmission spectra. We introduce POSEIDON, a two-dimensional atmospheric retrieval algorithm including generalised inhomogeneous clouds. We demonstrate that this prescription allows one to break vital degeneracies between clouds and prominent molecular abundances. We apply POSEIDON to the best transmission spectrum presently available, for the hot Jupiter HD 209458b, uncovering new insights into its atmosphere at the day-night terminator. We extensively explore the parameter space with an unprecedented 10$^8$ models, spanning the continuum from fully cloudy to cloud-free atmospheres, in a fully Bayesian retrieval framework. We report the first detection of nitrogen chemistry (NH$_3$ and/or HCN) in an exoplanet atmosphere at 3.7-7.7$\sigma$ confidence, non-uniform cloud coverage at 4.5-5.4$\sigma$, high-altitude hazes at $>$3$\sigma$, and sub-solar H$_2$O at $\gtrsim$3-5$\sigma$, depending on the assumed cloud distribution. We detect NH$_3$ at 3.3$\sigma$ and 4.9$\sigma$ for fully cloudy and cloud-free scenarios, respectively. For the model with the highest Bayesian evidence, we constrain H$_2$O at 5-15 ppm (0.01-0.03$\times$ solar) and NH$_3$ at 0.01-2.7 ppm, strongly suggesting disequilibrium chemistry and cautioning against equilibrium assumptions. Our results herald new promise for retrieving cloudy atmospheres using high-precision HST and JWST spectra.