No TiO detected in the hot Neptune-desert planet LTT-9779 b in reflected light at high spectral resolution

TL;DR

This study targets the reflected light spectrum of the hot Neptune LTT-9779 b at high spectral resolution to break the degeneracy between clouds and metallicity and to probe its mass-loss history. Using HRCCS with ESPRESSO in 4-UT mode, the authors build a comprehensive modeling and injection framework, including self-consistent JWST-informed atmospheres and single-species grids, to search for TiO and other species across three planetary dayside segments. They report a non-detection but demonstrate the method’s sensitivity and provide quantitative constraints on cloud-top pressures and atmospheric composition via injection tests, highlighting TiO depletion as a plausible explanation for the non-detection. The work also establishes a practical path toward future detections with the ELT, emphasizing target selection criteria and the critical role of deep spectral lines in enabling high-resolution reflected-light characterization.

Abstract

LTT-9779 b is an inhabitant of the hot Neptune desert and one of only a few planets with a measured high albedo. Characterising the atmosphere of this world is the key to understanding what processes dominate in creating the hot Neptune desert. We aim to characterise the reflected light of LTT-9779 b at high spectral resolution to break the degeneracy between clouds and atmospheric metallicity. This is key to interpreting its mass loss history which may illuminate how it kept its place in the desert. We use the high resolution cross-correlation spectroscopy technique on four half-nights of ESPRESSO observations in 4-UT mode (16.4-m effective mirror) to constrain the reflected light spectrum of LTT-9779 b. We do not detect the reflected light spectrum of LTT-9779 b despite these data having the expected sensitivity at the level 100 ppm. Injection tests on the post-eclipse data indicate that TiO should have been detected for a range of different equilibrium chemistry models. Therefore this non-detection suggests TiO depletion in the western hemisphere however, this conclusion is sensitive to temperature which impacts the chemistry in the upper atmosphere and the reliability of the line list. Additionally, we are able to constrain the top of the western cloud deck to $P_{\text{top, western}}<10^{-2.0}$ bar and the top of the eastern cloud deck $P_{\text{top, eastern}}<10^{-0.5}$ bar, which is consistent with the predicted altitude of MgSiO$_3$ and Mg$_2$SiO$_4$ clouds from JWST NIRISS/SOSS. While we do not detect the reflected light spectrum of LTT-9779 b, we have verified that this technique can be used in practice to characterise the high spectral resolution reflected light of exoplanets so long as their spectra contain a sufficient number of deep spectral lines. Therefore this technique may become an important cornerstone of exoplanet characterisation with the ELT and beyond.

Figures (15)

• Figure 1: Spectrum of a hydrogen-helium atmosphere with a uniform volume mixing ratio of Fe of 0.0001 (see Section \ref{['sec:searchgridtemplates']}). Each line represents a model with a grey cloud deck at a different altitude which is represented by the cloud top pressure.
• Figure 2: Top panel: phase-folded radial velocity curve comprised of radial velocity measurements from this work (ESPRESSO eclipse), ESPRESSO transit observations and HARPS. The vertical extent of each point represents the error. So that the radial velocities appear as a continuous curve, the systemic velocity and the fitted instrumental offset have been subtracted. Bottom panel: phase-folded short cadence TESS light curve (light points) with the darker points showing a binned version of the curve. In both panels, the best fit curve for a circular orbit is shown (orange).
• Figure 3: Phases of LTT-9779 b, assuming the ephemeris from the circular orbit fit in Table \ref{['tab:orbitfit']}, covered by these observations. Each circle represents an exposure and the shaded area indicates the secondary eclipse of the planet. The gaps in the observations taken on night one were due to technical difficulties and night three's observations had to be halted early due to precipitation.
• Figure 4: Schematic of the data reduction process for the spectral time-series of order 76 night two. Each of the six top panels show the time series of spectra with each row representing one spectrum in the series. Top panel: Original spectra. Second panel from the top: Spectral time-series after continuum normalisation. Third panel from the top: Spectral time-series after the mean spectrum has been subtracted. Fourth to sixth panels from the top: Spectral time-series after one, two and three sysrem iterations are run on the time series. Bottom panel: Difference in the standard deviation of the residuals between consecutive sysrem iterations as a function of the number of sysrem components. The shaded region highlights the one sigma deviation in the last four points (those to the right of the dashed line) and the black dot indicates the chosen number of iterations (the smaller number) as described in the text.
• Figure 5: High spectral resolution ($R=70,000$) models computed using the eastern-dayside, dayside, and western-dayside mean temperature-profiles from Coulombe2025 (T.1) assuming $10\times$ solar metallicity. Each panel shows two spectra, one that assumes equilibrium chemistry for all species (Ch.1) and one that removes TiO from the atmosphere (Ch.2). The contrast ratio measured by Coulombe2025 (see Fig. 3 of their work) for each segment is shown as the dashed line with the shaded region indicating the one sigma errors.