X-SHYNE Library: Young Substellar Spectra

• X-SHYNE Library is a homogeneous spectral dataset of young, low-mass brown dwarfs and planetary-mass companions observed with VLT/X-Shooter.
• It spans 0.65–2.5 μm, capturing key diagnostic features like TiO, VO, water vapor, and CO that are essential for atmospheric characterization.
• By combining semi-empirical and Bayesian synthetic approaches, the library benchmarks atmospheric models and highlights limitations in deriving T_eff and log g.

The X-SHYNE Library is a homogeneous, medium-resolution (R ≃ 8000) spectral library of 43 young (<500 Myr), low-mass (<20 M_Jup), and cool (T_eff = 600–2000 K) isolated brown dwarfs and wide-separation planetary-mass companions observed with the VLT/X-Shooter instrument. It spans a broad wavelength range from the optical (0.65 μm) to the near-infrared (2.5 μm) and is designed to provide a self-consistent, systematics-minimized dataset for comparative analysis of exoplanet analog atmospheres. By combining semi-empirical and Bayesian synthetic approaches, the library enables robust constraints on atmospheric properties and highlights both the strengths and current limitations of atmospheric models for young, low-gravity substellar objects (Petrus et al., 22 Jul 2025).

1. Spectral Content and Diagnostic Features

The X-SHYNE library’s spectra encompass 0.65–2.5 μm, probing spectral regions critical for identifying atomic and molecular features that diagnose physical conditions in young ultracool dwarfs:

• 0.60–0.85 μm (VIS): Prominent molecular bands (TiO, VO) in warmer types; disappearance at cooler L types marks cloud formation.
• 0.85–1.10 μm (Y band): Evolution of key atomic lines with spectral type.
• 1.10–1.35 μm (J band): Water vapor absorption, potassium (K I) doublets at 1.17 and 1.25 μm (surface gravity indicators).
• 1.42–1.81 μm (H band): Strong H₂O absorption shaping the continuum.
• 1.96–2.48 μm (K band): CO overtone bands near 2.3 μm (C/O ratio sensitivity).

Spectral zooms facilitate identification and measurement of features essential for determining T_eff, log g, [M/H], and C/O. The evolving morphology across bands allows systematics studies in cloud opacity and pressure broadening effects across the sample.

2. Physical Characterization Methodology

Two complementary methods are used for deriving atmospheric and physical parameters:

(a) Semi-Empirical Analysis:

• Spectral Energy Distribution (SED) Completion: Observed X-Shooter spectra are combined with broadband photometry to construct SEDs.
• Bolometric Luminosity (L_bol): Integrated numerically from the SED.
• Evolutionary Model (COND03): Inputting L_bol and age to derive mass, radius, T_eff, and surface gravity.
• Effective Temperature Relation: Using

$$L_{\text{bol}} = 4\pi R^2\,\sigma\,T_{\text{eff}}^4$$

for consistency checks or direct estimation.

• Equivalent Width (EW) Calculation: For features such as K I, defined as

$$EW = \sum_{\lambda} \left[1 - \frac{F_{\text{line}}(\lambda)}{F_{\text{cont}}(\lambda)}\right] \Delta\lambda$$

with pseudo-continuum estimation bracketing each feature.

(b) Synthetic (Forward-Modeling) Analysis:

• Atmospheric Model Grids: Three independent self-consistent model grids (Exo-REM, Sonora Diamondback, ATMO) are adopted, each varying in cloud physics and chemistry.
• ForMoSA Bayesian Inference Code: Observed segments (J, H, K, J+H+K) are simultaneously fit using synthetic spectra; parameters are sampled from their posterior distributions, with error inflation (iteratively increasing uncertainties until reduced $\chi^2 \approx 1$) to account for model imperfections.

This approach allows a direct comparison of empirically derived and model-dependent atmospheric properties, as well as an assessment of systematics.

3. Comparison of Semi-Empirical and Synthetic Results

The paper finds strong agreement in bolometric luminosities (L_bol) across both approaches, confirming that integrated energy output is robustly constrained even with incomplete or model-dependent SEDs. However, effective temperatures (T_eff) and surface gravities (log g) show model-dependent discrepancies:

• Cloudy models (Exo-REM, Sonora Diamondback) systematically underestimate T_eff (especially in J and H bands for earlier L dwarfs), a likely result of missing absorbers relevant to cloud opacity.
• The cloudless ATMO grid, which parametrizes diabatic convection via an adiabatic index, tends to yield T_eff estimates more consistent with evolutionary predictions.
• Log g displays significant scatter (3.5–5.5 dex for mid-L dwarfs across models), interpreted as resulting from a range of rotational velocities, cloud migration toward equators, and a variety of viewing geometries that modulate observed pressures and cloud column densities.

This highlights the challenge: while L_bol can often be robustly recovered, T_eff and log g values remain sensitive to model assumptions on cloud physics and chemical equilibrium.

4. Atmospheric Model Grids: Cloud Treatment and Limitations

The choice of synthetic atmosphere grid has a direct, significant impact on inferred physical and chemical properties:

• Exo-REM: Cloudy, non-equilibrium chemistry; lacks opacity contributors (e.g., Al₂O₃), leading to underestimated T_eff in warmer L dwarfs.
• Sonora Diamondback: Cloudy, equilibrium chemistry; cloud effects encoded via sedimentation efficiency (f_sed). K band is well reproduced, but systematic issues persist elsewhere.
• ATMO: Cloudless, with diabatic convection mimicking cloud suppression. Yields higher T_eff estimates and aligns better with evolutionary L_bol, but may inadequately capture true cloud physics or their spectral signatures.

The absence of key absorbers or incomplete opacity databases in the cloudy models leads to forced parameter compensation during spectral fitting, directly highlighting the need for continued improvements in model accuracy.

5. Trends in Atmospheric Composition and Implications for Formation

Analysis of the sample's metallicity ([M/H]) and carbon-to-oxygen ratio (C/O) finds both are close to solar on average. This suggests, within current modeling uncertainties, that these young low-mass objects share a stellar-like formation process (gravitational collapse) rather than exhibiting the C/O enhancement signatures predicted for core accretion–dominated formation. While robust individual [M/H] and C/O determinations remain challenging—reflecting both model and data limitations—the global inference is that the X-SHYNE sample supports a star-like, rather than planet-like, compositional origin.

6. Advantages of Homogeneity for Comparative Atmospheric Studies

The entire X-SHYNE dataset was acquired with a consistent instrument (VLT/X-Shooter), under similar observational setups, and uniformly reduced and analyzed. Such homogeneity eliminates instrument-dependent systematics and enables:

• Cross-comparisons that directly attribute spectral trends and anomalies to intrinsic physical parameters rather than data reduction or calibration artifacts.
• More secure identification of trends (e.g., in T_eff or log g versus spectral type or age) and model–data discrepancies.
• Greater sensitivity to subtle physical drivers such as rotational modulation, viewing angle, and local variations in cloud coverage or atmospheric structure.

This approach avoids over-interpretation based on inhomogeneous data and makes the X-SHYNE library a robust reference for model development and validation.

7. Impact and Outlook

The X-SHYNE library provides a foundation for benchmarking and improving atmospheric models of young exoplanet analogs. By systematically comparing semi-empirical and model-based analyses, it exposes both where models succeed (L_bol) and where they fail (cloud physics, absorber completeness). The globally solar [M/H] and C/O ratios emerge as a population metric rather than robust individual diagnostics, reflecting current model uncertainties. The work underscores the need for further incorporation of missing opacity sources and improved cloud treatments in model development. The homogeneous nature of the dataset sets a standard for comparative atmospheric studies and will be crucial for the analysis and interpretation of upcoming large samples of directly imaged exoplanets and ultracool dwarfs (Petrus et al., 22 Jul 2025).