Limits on the Post-eclipse Emission Spectrum of HD 80606 b From High-Resolution Spectroscop

TL;DR

HD 80606 b offers a unique testbed for atmospheric dynamics during extreme stellar heating near periastron. The authors apply high-resolution cross-correlation spectroscopy in the Keck/NIRSPEC K-band to post-eclipse observations, employing retrievals with both fixed and flexible P–T profiles and forward models to probe CH$_4$, CO, and H$_2$O abundances. The results yield a marginal, absorption-dominated signal with weak CH$_4$ and CO indications and little H$_2$O evidence, broadly consistent with JWST post-eclipse spectra that also disfavour a strong thermal inversion. While not a definitive detection, the analysis supports a non-inverted atmosphere with weak molecular features and highlights the challenges of HRCCS detections for long-period, eccentric exoplanets, underscoring the need for higher-resolution or broader-wavelength data to robustly constrain HD 80606 b’s atmosphere.

Abstract

We present Keck/NIRSPEC $K$-band observations of HD 80606 b, one of the most eccentric known exoplanets. HD 80606 b was observed after secondary eclipse, close to periastron, when the planet passes within 0.03 AU of HD 80606 and the rapid heating of the atmosphere may lead to extreme chemical changes and a temporary thermal inversion. The rapid change in the planetary radial velocity near periastron is sufficient to enable high-resolution cross-correlation spectroscopy (HRCCS) analysis, which produces a tentative detection ($\rm SNR\sim4$) of HD 80606 b. Injection-recovery tests appear to reject strong thermal inversions near periastron, consistent with recent results from JWST. We also perform atmospheric retrievals with free parameters for the Pressure-Temperature ($P-T$) profile and with a profile matched to the JWST results, which suggest the presence of absorption features from CH$_4$ and CO. While HD 80606 b is not definitively detected in these data, these results are consistent with JWST observations, which found the post-eclipse atmosphere of HD 80606 b shows weak absorption features from these species. Future observations with higher spectral resolution and/or wider wavelength coverage are needed for a confident atmospheric detection of HD 80606 b via high-resolution spectroscopy alone, but such observations are a challenge to schedule due to the 111-day orbital period.

Figures (13)

• Figure 1: Opacities of the dominant species (H$_2$O, CO, and CH$_4$) in the NIRSPEC K-band spectral region. NIRSPEC contains 9 spectral orders in the K-band region, 6 of which can be calibrated and used in our analysis (3 are highly contaminated). H$_2$O, CO, and CH$_4$ all contain major bandheads within this spectral region and so we expect to be able to detect them, if they are present.
• Figure 2: (left) The blue curve shows the separation of HD 80606 b from its host star in units of the stellar radius as a function of time immediately before and after periapsis. The eclipse midpoint is a few hours earlier and is denoted by a dashed black line. The red curve is an illustrative equilibrium temperature for the planet driven only by instantaneous insolation based on the square root of its distance from the star consistent with the observations of laughlin2009. The gray shaded region shows the coverage of the sikora2024JWST observations, and the green shaded region shows the approximate times of the NIRSPEC observations presented here. right) The velocity of the planet in km s$^{-1}$. Values are derived from the model and parameters described in pearson2022.
• Figure 3: Retrieved parameters and priors for the three $P-T$ cases omitting 6 principal components. Additionally, we require the atmospheric temperature stay below 3500 K at all pressure levels. For nearly all parameters, the marginalized posterior spans the prior range. The full corner plots are shown in Appendix \ref{['app:corner']}.
• Figure 4: Retrieved maximum-likelihood (top) and median (bottom) spectral models from the retrievals with six omitted components for each of the considered $P-T$ priors. The expected flux ratio is plotted in dashed black, assuming the star and planet are blackbodies at 5500 K and 1500 K, respectively. The fixed profiles cases roughly matches the nominal flux ratio by construction, while the median retrieved free $P-T$ results prefer slightly higher fluxes at short wavelengths and the maximum-likelihood models show significantly stronger fluxes. Such flux offsets are common in short-bandpass HRCCS observations finnerty2023finnerty2025c. The apparent emission features are a result of absorption features in the stellar model, as all plotted planet models lack a thermal inversion. The spectra show some absorption from CO and CH$_4$, but these features are relatively weak.
• Figure 5: Cross-correlation versus frame number (top) and coadded in the nominal reference frame of HD 80606 b (bottom) for the maximum-likelihood inversion-permitted models from the retrievals omitting 4, 6, and 8 PCs from left to right. The nominal planet rest frame is indicated in dashed red, and the stellar and telluric reference frame in solid and dashed black, respectively. Individual orders are shown in the lower panel in blue, and the average of all orders in black. Values are converted to signal-to-noise ratio by dividing by the standard deviation of the region with $|\Delta v_{sys} - v_{pl}| > 50$$\rm km\,s^{-1}$. The planet model is reprocessed with the removed PCs by assuming the nominal $K_{\rm p}$ and varying $\Delta v_{\rm sys}$. Residuals in the stellar/telluric reference frame are clear in the 4 PC case, suggesting uncorrected non-planetary features are contributing to the cross-correlation, but are much less prominent in the 6 and 8 PC cases. The tentative planet peak is weakly visible after coadding at $\sim10$$\rm km\,s^{-1}$, but does not dominate the cross-correlation signal. Note the lower SNR compared with the $K_{\rm p} - \Delta v_{\rm sys}$ diagrams in Figure \ref{['fig:kpvsys']} due to differences in the noise estimation.