betaFeedback

WorksPhysics and Astronomy

Dwarfs of Fire and Ice: UV-Irradiation-Driven Thermal Inversion, Disequilibrium Chemistry, and a Cloud-Free Atmosphere in a Strongly Irradiated Brown Dwarf Probed by JWST/NIRSpec Phase-Resolved Spectroscopy

Abstract

We present time-resolved JWST/NIRSpec PRISM spectrophotometry of the white dwarf (WD)-brown dwarf (BD) binary SDSS J141126.20+200911.1 (SDSS1411). Close WD-BD binaries provide powerful laboratories for studying the effects of stellar irradiation on substellar atmospheres because of their high companion-to-host flux ratios, short orbital periods, and intense UV irradiation from the WD. We measure strong wavelength-dependent variability, with day-night relative flux variations reaching ~280%. The SDSS1411B phase curves are well described by a second-order Fourier model, and correlations among the Fourier amplitudes suggest larger flux and temperature variations in the upper atmosphere. Non-irradiated forward models under chemical equilibrium do not reproduce the observed phase-resolved SDSS1411B spectra, indicating substantial departures from isolated-BD atmospheric structures. In contrast, atmospheric retrievals successfully reproduce both the dayside and nightside spectra. The retrievals reveal a prominent dayside temperature inversion, favor cloud-free atmospheres, and provide evidence for CO2 disequilibrium chemistry. An energy-budget analysis indicates low day-to-night heat redistribution efficiency and constrains the BD cooling age to ~3 Gyr, while the inferred BD radius is significantly more compact than predicted by BD evolutionary models. These results reveal the rich atmospheric dynamics and disequilibrium chemistry of a UV-irradiated BD and demonstrate the power of full-orbit JWST phase-resolved spectroscopy for constraining the atmospheric physics of strongly irradiated substellar objects.

Introduction

Brown dwarfs (BDs) offer insights into ultracool atmospheres (T<2,000 KT<2{,}000\,\mathrm{K}) without the contrast challenges posed by the host stars of exoplanets. The more favorable observing conditions often allow efficient acquisition of higher-quality data than what is possible for exoplanets. Time-resolved observations probe the atmospheres in different rotational phases, thereby providing important spatial constraints on the atmospheres (e.g., E. Buenzli et al. [30]; S. A. Metchev et al. [89]; T. Karalidi et al. [71]; B. A. Biller et al. [18]; N. Fuda et al. [55]; A. M. McCarthy et al. [86]). These include constraints on condensate cloud distribution (e.g., D. Apai et al. [8]), pressure-dependent cloud structures (E. Buenzli et al. [31]; H. Yang et al. [129]; J. M. Vos et al. [124]; B. A. Biller et al. [18]; B. E. Miles et al. [90]), particle sizes (e.g., B. W. P. Lew et al. [74]; J. M. Vos et al. [124]), and non-equilibrium chemistry (A. M. McCarthy et al. [86]; B. E. Miles et al. [90]). Multi-epoch high-precision monitoring also revealed that cloud cover evolves rapidly, even on time scales as short as a few hours (e.g., E. Artigau et al. [10]; M. Gillon et al. [60]; D. Apai et al. [9]). It was shown that condensate clouds in L/T transition brown dwarfs often trace atmospheric zonal circulation and are shaped by planetary-scale waves (e.g., D. Apai et al. [9, 7]; Y. Zhou et al. [136]; N. Fuda et al. [55]). Longer-timescale observations also identified spectrally and dynamically distinct regions at high latitudes, evidence for the presence of polar vortices (D. Apai et al. [7]; N. Fuda & D. Apai [54]).

However, the majority of the BDs studied differ from giant exoplanets in that they are not irradiated (or at negligible levels). It is, then, not trivial to transfer the abundant knowledge gained from non-irradiated BDs to irradiated giant exoplanets.

This difference in irradiation is important. Stellar irradiation impacts atmospheric physics and chemistry (e.g., A. P. Showman et al. [110]). In giant planets, external irradiation influences at least three key atmospheric processes: (1) Energy and material transport (longitudinally and vertically), (2) Cloud and haze formation and dispersal, and (3) Photochemistry (e.g., X. Tan & A. P. Showman [117]; V. Parmentier et al. [99, 98]; M. E. Steinrueck et al. [114]).

Close white dwarf-brown dwarf systems (WD–BDs) bridge the gap between non-irradiated BDs and irradiated exoplanets (A. P. Showman [109]; A. P. Showman et al. [110]): WD–BD systems provide excellent laboratories for studying the effects of stellar irradiation on substellar atmospheres (e.g., S. P. Littlefair et al. [78]; R. C. Amaro et al. [5]; J. R. French et al. [53]). Compared with hot Jupiters, WD–BD binaries can produce BD infrared phase curves with signal-to-noise ratios up to ∼10\sim10× higher because of the small radii and blue spectral energy distributions of hot WDs. Their orbital periods are also typically ∼10\sim10–20×20\times shorter than those of hot Jupiters, allowing complete phase-curve observations through the expenditure of only modest JWST observing time (R. C. Amaro et al. [5]). In addition, WD emission is generally featureless and stable across the near-infrared, aside from systems containing pulsating WDs (e.g., R. C. Amaro et al. [6]), enabling a clean separation of the BD spectrum from the WD contribution.

Hubble Space Telescope (HST) phase-curve observations of irradiated BDs have revealed irradiation-driven atmospheric circulation and disequilibrium chemistry in a unique regime of irradiation and atmospheric conditions (e.g., B. W. P. Lew et al. [75]; R. C. Amaro et al. [4]; J. D. Lothringer et al. [82]; R. C. Amaro et al. [6]). JWST phase-resolved spectroscopy of irradiated BDs, with its broad wavelength coverage, high sensitivity, and high temporal cadence, now provides a timely opportunity to characterize these atmospheres and gain new insights into the physics of irradiated substellar atmospheres (D. Broski-Laing et al. [28]).

JWST program GO-4967 (PI: Y. Zhou) provides JWST/NIRSpec phase-resolved spectrophotometry of five close WD–BD binaries spanning equilibrium temperatures of ∼\sim750–3300 K, enabling systematic studies of irradiated BDs across a broad range of irradiation environments. D. Broski-Laing et al. [28] presented the first atmospheric analysis from this program for ZTF0038B, the least strongly irradiated BD in the sample. Through light-curve modeling, they identified asymmetric CO2\mathrm{CO}_2 band light curves as signatures of irradiation-driven disequilibrium chemistry. Using both self-consistent forward modeling (M. S. Marley et al. [85]; C. V. Morley et al. [92]) and atmosphere retrievals (P. Mollière et al. [91]; E. Nasedkin et al. [95]), they further revealed phase-dependent variations in atmospheric structure and molecular abundances and constrained day–night energy redistribution.

We present an analysis of the second target, SDSSJ141126.20+200911.1 (SDSS1411), which experiences an intermediate level of stellar irradiation (Firr∼108.6 erg s−1 cm−2F_{\rm irr} \sim10^{8.6}\ {\rm erg\,s^{-1}\,cm^{-2}}) within the GO-4967 sample. SDSS1411 was first identified as an eclipsing WD–BD binary by K. Buermann et al. 2013 in the Catalina Sky Survey (A. J. Drake et al. [42]). Subsequent ground-based observations constrained the system parameters and basic properties of the BD (SDSS1411B) (S. P. Littlefair et al. [78]; S. L. Casewell et al. [35, 34]; A. Yates et al. [130]), including a WD effective temperature of $\sim$13,000 K, a BD dayside equilibrium temperature of ∼\sim1400 K, a BD mass of ∼\sim0.050,M_{\odot}$, and an orbital period of ∼2.03\sim2.03 hr. Based on these results, F. Sainsbury-Martinez et al. [105] investigated the atmospheric dynamics of SDSS1411B using 3D global circulation models and suggested that rapid rotation suppresses atmospheric heat transport, particularly downward energy transport into the deep atmosphere, potentially explaining the lack of significant radius inflation. In addition, near-infrared HST phase-curve observations (B. W. P. Lew et al. [75]) further characterized day–night variations in the BD atmosphere, including changes in thermal emission and brightness temperature. However, the role of stellar irradiation in shaping the atmospheric thermal structure and disequilibrium chemistry has not yet been investigated in detail using broad-wavelength spectroscopy extending beyond ≳2\gtrsim2 μ\mum.

In this paper, we study the effects of irradiation on the atmosphere of SDSS1411B using JWST/NIRSpec phase-resolved spectroscopy, covering 0.6–5.3 μ\mum wavelengths. We analyze the complete phase and multi-wavelength light curve, use atmospheric (forward) modeling, and perform atmospheric retrievals. The paper is organized as follows: § 2 describes the JWST observations and the data reduction process. In § 3, we analyze the phase curves and constrain the binary system parameters. § 4 presents the atmospheric modeling and energy transport analyses of SDSS1411B. Finally, we discuss the implications of our results in § 5 and summarize our conclusions in § 6.

Observations and Data Reduction

JWST Observations

We observed SDSS1411 on 2025 May 27 with JWST/NIRSpec as part of program GO-4967 (PI: Y. Zhou), using the PRISM/CLEAR grating-filter combination with a broad wavelength coverage of 0.6–5.3 μ\mum and a low spectral resolving power of R∼100R\sim100 at 3 μ\mum ([19, 69]). We performed high-throughput, stable time-resolved spectroscopy in the Bright Object Time Series (BOTS) mode ([44]) using the 1.6″ × 1.6″ S1600A1 slit without dithering. We covered the full ∼2.02\sim2.02 hr orbital period of the SDSS1411 system with 2.27 hr of observations. The time series consists of 75 integrations, each with 120 groups. We used the NRS readout pattern on the SUB512 subarray of the NRS1 detector.

JWST/NIRSpec Data Reduction

The JWST/NIRSpec data reduction follows the procedures described in D. Broski-Laing et al. [28], with additional background-subtraction steps summarized below. To extract flux-calibrated time-series spectra, we reduced the JWST/NIRSpec spectroscopic phase curve data of SDSS1411 using the Eureka! pipeline (T. Bell et al. [15]) and the standard jwst data reduction pipeline (H. Bushouse et al. [33]). We divided each integration into five subgroups to improve the time resolution of the light curve while maintaining sufficient signal-to-noise ratio. We then generated the high-cadence spectrophotometric light curve using the first three stages of the Eureka! pipeline. We constructed the 2D spectral image by fitting the ramps of nondestructive reads using a maximum-likelihood estimator (T. D. Brandt [26]), adopting a jump-detection threshold of 15σ15\sigma. We derived 1D spectra by performing background subtraction and optimal spectral extraction (K. Horne [66]) using an aperture half-width of 3 pixels and a background half-width of 5 pixels.

To calibrate the 1D spectra from the Eureka! pipeline, we also generated absolute-flux-calibrated spectra for each integration using the standard jwst pipeline. These 1D spectra were obtained through background subtraction using Stage 3 of Eureka!, because the standard jwst pipeline does not perform background subtraction for BOTS observations, followed by optimal spectral extraction based on Stage 2 of the jwst pipeline. We derived the flux calibration factor for the Eureka!-produced spectra from the ratio of the integration-averaged Eureka! spectra to the corresponding flux-calibrated jwst spectrum. Applying this calibration factor to the high-cadence Eureka! spectra, we constructed the final absolute-flux-calibrated, high-cadence spectrophotometric light curve of SDSS1411.

Results

Lightcurve Analysis

The reduced phase-resolved spectra of SDSS1411 show strong wavelength-dependent flux variations. Figure 1 presents the combined WD+BD spectra along with the corresponding light curves integrated over several representative wavelength intervals. In Figure 1(a), the spectra at wavelengths shorter than ∼2 μm\sim2\ \mu\mathrm{m} are dominated by WD emission, exhibiting prominent Hα\alpha, Paβ\beta, and Paγ\gamma hydrogen absorption features at 0.656, 1.095, and 1.284 μm\mu\mathrm{m}, respectively. At wavelengths longer than ∼2 μm\sim2\ \mu\mathrm{m}, the BD emission contributes significantly to the total system flux, and its day–night flux modulation drives the large observed variability amplitudes. We also detect significant flux variations in the observed system spectrum associated with the primary eclipse (BD eclipses the WD) at an elapsed time of ∼1.55\sim1.55 hr and the secondary eclipse (WD transits the BD) at ∼0.55\sim0.55 hr (Figure 1b). The flux decrease during the secondary eclipse is small relative to the phase-curve amplitude because the WD, which is smaller than the BD, obscures only a fraction of the BD’s projected disk. In Figure 1(b), the light curve integrated at 0.6–1.47 μm\mu\mathrm{m} varies by less than 2%, excluding the eclipse phases. In contrast, the phase curves at wavelengths 4.40–5.20 μm\mu\mathrm{m} show larger variability amplitudes up to ∼54%\sim54\% total variation, possibly driven by rotational/orbital modulation of the BD emission.

Figure 1. Panel (a): The phase-resolved spectra of the SDSS1411 WD+BD system, shown for every fifth epoch. The shaded regions indicate the wavelength bins used to construct the phase curves shown in panel (b). Panel (b): Normalized spectrophotometric light curves of the SDSS1411 system. The colors correspond to the shaded wavelength channels in Panel (a). The primary and secondary eclipses occur at elapsed times of ∼1.55\sim1.55 and ∼0.55\sim0.55 hr, respectively. At wavelengths ≲2\lesssim2 μ\mum, the system light curves are dominated by the stable WD emission, whereas at wavelengths ≳2\gtrsim2 μ\mum, the BD contribution produces strong day–night flux variations.

We modeled the primary and secondary eclipses using the phase-curve data spanning 0.6–5.3 μm\mu\mathrm{m}. The eclipse fits provide mid-eclipse time measurements, allowing us to determine the robust orbital phase of each integration, and constrain the near-infrared limb-darkening parameters. The remaining system parameters are derived in § 4.1 using the latest values reported in the literature. We computed the eclipse models with the batman transit modeling package (L. Kreidberg [72]). Because batman does not include emission from the secondary companion, we first removed the companion contribution from the spectra used for each eclipse fit. For the primary eclipse, we subtracted the average nightside spectrum, measured during t=1.535t = 1.535–1.5651.565 hr. For the secondary eclipse, we subtracted the WD spectrum, which we estimated by subtracting the nightside spectrum from the system spectra adjacent to the primary eclipse at t=1.45t = 1.45–1.491.49 and 1.611.61–1.651.65 hr. We then integrated the corrected spectra over 0.60.6–5.35.3 µm for the primary eclipse and 2.02.0–5.35.3 µm for the secondary eclipse to construct the corresponding light curves. The eclipse fits were performed over elapsed-time intervals of 1.451.45–1.651.65 hr and 0.450.45–0.600.60 hr for the primary and secondary eclipses, respectively, assuming that the intrinsic BD and WD flux variations are negligible over these short intervals.

We fitted the batman transit model using the emcee package, which implements the Markov Chain Monte Carlo (MCMC) fitting algorithm (D. Foreman-Mackey et al. [51]). We used four free parameters: mid-eclipse time t0t_0, the ratio of the BD radius over the WD radius rp/rsr_{\mathrm{p}}/r_{\mathrm{s}}, a linear limb-darkening coefficient uu, and an offset for the light-curve normalization. To reduce model complexity, we fixed the remaining orbital parameters, orbital period PP, semi-major axis ratio a/rWDa/r_{\mathrm{WD}}, and inclination ii, to the least-χ2\chi^2 best-fit values reported by B. W. P. Lew et al. [75]. Table 1 summarizes the adopted priors for the eclipse fit. To properly account for the uncertainties in the phase curves, we inflated the pipeline-produced uncertainties to achieve a reduced χ2=1\chi^2 = 1 for each phase curve model fit, following D. Broski-Laing et al. [28].

ParameterPriors for Eclipse FitBest-fit Parameters
P (days)*2.02864 (fixed)2.02864 (fixed)
i (deg)*86 (fixed)86 (fixed)
a/rWD*43 (fixed)43 (fixed)
Primary Eclipse
t₀1.51 – 1.581.547445 ± 0.000035
u−3 – 30.560⁺⁰·⁰⁷⁵₋₀·₀₈₀
rBD/rWD4 – 54.7321 ± 0.0042
Offset0.9 – 1.11.003 ± 0.0014
Secondary Eclipse
t₀0.5 – 0.560.5306⁺⁰·⁰⁰⁵⁷₋₀·₀₀₄₅
u−4 – 41.41⁺⁰·⁵²₋₁·⁷⁸
rWD/rBD0.1 – 0.30.164⁺⁰·⁰²⁷₋₀·⁰³⁰
Offset0.9 – 1.11.004 ± 0.005

Table 1. Priors and Best-Fit Parameters for the Eclipse Fit. *Adopted best-fit values from B. W. P. Lew et al. [75].

Figure 2 shows the observed eclipse light curves together with the best-fit batman models. The modeled light curves are generally consistent with the observed primary and secondary eclipses. The best-fit eclipse parameters are summarized in Table 1. The best-fit primary mid-eclipse time, t0=1.547445±0.000035t_0 = 1.547445 \pm0.000035 hr, corresponds to a barycentric-corrected eclipse time of 2460822.8508442(15) MJDTDB_{\mathrm{TDB}}. This timing is consistent with the linear ephemeris of A. Yates et al. [130] at epoch 57,149 within ∼0.6\sim0.6 s. The time interval between the primary and secondary mid-eclipse times, Δt0=1.0168±0.0051\Delta t_0 = 1.0168 \pm0.0051 hr, is also consistent with half of the orbital period within the 1σ1\sigma uncertainty.

Figure 2. Panel (a): Primary eclipse light curve of SDSS1411 with the best-fit model. The red points show the observed phase curve, and the black solid line represents the best-fit batman transit model. The lower panel shows the residuals between the model and the observed light curve. Panel (b): Same as Panel (a), but for the secondary eclipse.

White Dwarf Spectrum Subtraction

To construct the phase-resolved spectra of the BD, we separated the WD spectrum from the observed spectra of the unresolved SDSS1411 WD-BD system. We assumed that the combined spectra consist of a stationary WD component and a time-dependent BD component (see Figure 1 and § 3.1). Spectra obtained near but outside the primary eclipse contain contributions from both the WD and the BD nightside, whereas the primary-eclipse spectra contain only the BD emission. We therefore extracted the WD spectrum by subtracting the primary-eclipse spectra from the nightside system spectra, based on the assumption that the BD nightside emission remains constant during the night and eclipse phase. We defined the eclipse-phase intervals as t=1.535t = 1.535–1.5651.565 (ϕ∼0\phi\sim0–0.0080.008 and 0.9940.994–11), and the night-phase intervals as t=1.45t = 1.45–1.491.49 and 1.611.61–1.651.65 hr (ϕ∼0.03\phi\sim0.03–0.050.05 and 0.950.95–0.970.97). Figure 3 shows the re-

Figure 3. Separated WD spectrum of the SDSS1411 system and the best-fit model WD spectrum using the model grid from P.-E. Tremblay et al. [119]. The bottom panel shows the ratio between the observed WD spectrum and the best-fit model. The observed spectrum is generally consistent with the model within 2±14%.

resulting WD spectrum, whose flux is approximately ∼23\sim23 times higher than the time-averaged BD flux.

To assess the agreement with theoretical WD models and evaluate the quality of the absolute flux calibration, we compared the separated WD spectrum to pure-hydrogen WD model spectra from P.-E. Tremblay et al. [119]. We linearly interpolated between the models with log⁡g=7.5\log g = 7.5 and 8.08.0 to obtain a model spectrum at Teff=11,616T_{\mathrm{eff}} = 11{,}616 K and log⁡g=7.93\log g = 7.93, corresponding to the best-fit WD parameters reported by O. Vincent et al. [123]. To match the spectral resolution, we convolved the model spectra with the dispersion curve of NIRSpec/PRISM. Figure 3 presents the separated WD spectrum and the best-fit WD model. The fitted scaling factor corresponds to a WD radius of 0.0134±0.00030.0134 \pm0.0003 R⊙_\odot, using the Gaia DR3 distance of 177±4177 \pm4 pc ([57]), consistent with 0.0130±0.00030.0130 \pm0.0003 R⊙_\odot reported by B. W. P. Lew et al. [75]. The observed and model spectra agree to within relative differences of 2±14%2 \pm14\%.

Phase-dependent Brown Dwarf Spectra

After subtracting the phase-independent WD component, we derived the phase-resolved BD spectra by using a Fourier model to describe the BD emission modulation (e.g., [75, 28]). The Fourier modeling produces robust phase-resolved BD spectra based on the spectrophotometric data spanning a broad range of orbital phases. We masked the eclipse phases to prevent fitting the eclipse features as intrinsic BD flux variations. We then modeled the phase curve in each wavelength bin using a second-order Fourier series,

F(λ,t)=A0(λ)+A1(λ)cos⁡(ωt+ϕ1)+A2(λ)cos⁡(2ωt+ϕ2).(1)F(\lambda,t)=A_0(\lambda)+A_1(\lambda)\cos(\omega t+\phi_1)+A_2(\lambda)\cos(2\omega t+\phi_2). \tag*{(1)}

We fixed ω\omega using the orbital period of 2.028786 hr reported by A. Yates et al. [130] and defined tt as the elapsed time from the mid-eclipse time, t0=1.547445t_0 = 1.547445, measured in § 3.1. We then fitted the amplitudes and phases of Fourier components in each wavelength bin. Similar to the phase-curve modeling described in § 3.1, we inflated the pipeline-produced uncertainties such that the reduced χ2\chi^2 of each Fourier model fit is unity, resulting in an average error-inflation factor of ∼1.86\sim1.86.

Figure 4(a) shows the best-fit Fourier amplitudes at each wavelength bin. A0A_0 represents the constant (time-independent) component of the BD emission, while A1A_1 and A2A_2 quantify the temporal variation amplitudes of the BD emission at each wavelength. The wavelengths with strong molecular absorption generally correspond to the trough of

Figure 4. Panel (a): Amplitudes of the individual Fourier components as a function of the central wavelength of each bin. A0A_0 corresponds to the constant component of the BD emission. A1A_1 and A2A_2 represent the amplitudes of the time-varying BD emission components at each wavelength. Shaded regions indicate wavelengths with strong molecular opacities, which generally coincide with troughs in ∣A0∣\lvert A_0\rvert and peaks in ∣A1∣\lvert A_1\rvert. Panel (b): Same as Panel (a), but for phases of the Fourier components. The best-fit phases are generally consistent with zero phase shift within 1σ1\sigma of the phase distributions. The phase scatter is larger at wavelengths where the Fourier amplitudes are smaller. Panel (c): Example of a second-order Fourier fit to the BD light curve in the wavelength bin centered at 3.133.13 μ\mum. The red line indicates the best-fit model light curve, and colored dashed lines show individual Fourier components. The gray shaded regions mark integrations masked due to eclipse features.

relation between photospheric pressures and flux variation amplitudes. Figure 4(b) shows the best-fit phases for the corresponding Fourier components. The fitted phases are generally consistent with zero, with ϕ1=−0.01±0.30\phi_1 = -0.01 \pm0.30 and ϕ2=−0.03±0.44\phi_2 = -0.03 \pm0.44. Larger phase scatter is observed at wavelengths where the corresponding Fourier amplitudes are smaller. The first- and second-order Fourier components show a phase difference of ϕ1−ϕ2∼0\phi_1 - \phi_2 \sim0, indicating the symmetry of the light curve.

We reconstructed phase-resolved BD spectra using the best-fit Fourier model. We estimated the flux density uncertainties by propagating the uncertainties of the best-fit model parameters. Figure 5 presents the reconstructed BD spectra at different orbital phases. The integrated dayside BD flux is approximately 280% brighter than the nightside BD flux. The BD flux density varies by ≲\lesssim 30% at ≲1.3\lesssim1.3 and ∼1.6 μm\sim1.6\,\mu\mathrm{m}, with the largest variation of ≳\gtrsim 2,000% at ∼3.4\sim3.4 and 5 μm5\,\mu\mathrm{m}. The troughs in the BD nightside spectrum generally coincide with opacity peaks of common atmospheric absorbers, including H2_2O, CO2_2, and CO. In particular, H2_2O, CH4_4, and CO exhibit strong opacity at ∼3\sim3 and 5 μm5\,\mu\mathrm{m}, which correspond to the spectral regions with the largest variability (see § 5.3). These regions show prominent absorption features on the nightside, but strong emission features on the dayside, possibly indicating a significant

Figure 5. Top: Opacities of common absorbers in exoplanet atmospheres, calculated at a pressure of 0.1 mbar and a temperature of 1,000 K. The y-axis is inverted to better compare the wavelength-dependent opacity trends with the phase-resolved BD spectra shown in the middle panel. Middle: BD spectra at different orbital phases reconstructed using the best-fit Fourier model (Equation (1)). Different colors indicate orbital phase: orange, green, purple, and blue correspond to the Day (ϕ=0\phi= 0), Morning and Evening (ϕ=π/2\phi= \pi/2, 3π/23\pi/2), and Night (ϕ=π\phi= \pi) phases, respectively. The phase-resolved spectra show wavelength-dependent flux variation, with an average variation of ∼\sim450% relative to the nightside spectrum. Bottom: Brightness temperatures of SDSS1411B with different orbital phases. The brightness temperatures span ∼\sim600−-2000 K, with an average day-night temperature variation of ∼\sim580 K.

temperature inversion in the dayside atmosphere.

To study the thermal structure of SDSS1411B, we calculated the brightness temperature at each wavelength, defined as the temperature of a blackbody that produces the same specific intensity as the observed flux. The brightness temperature provides an approximate measure of the atmospheric temperature at the layer probed by each wavelength. When combined with a temperature–pressure (T–P) profile without a temperature inversion, the brightness temperature also provides a useful proxy for the relative pressure levels probed at different wavelengths. We derived the brightness temperatures using the Planck function, adopting published values for the BD radius and system distance of 0.70 ±\pm 0.04 RJup_{\mathrm{Jup}} and 177 ±\pm 4 pc, respectively ([78, 84, 57]).

Figure 5(c) shows the brightness temperature as a function of wavelength at different orbital phases. The brightness temperature has strong wavelength dependence, spanning ∼\sim600−-2,000 K. The average day–night temperature difference is ∼\sim610 K. At ∼3\sim3 and 5 μm5\,\mu\mathrm{m}, where the nightside brightness temperature is lowest, the temperature variation is largest, reaching ∼1,000\sim1{,}000 K. In contrast, spectral regions at λ≲1.3\lambda\lesssim1.3, ∼1.6\sim1.6, 2.1, and 4 μ\mum show much smaller temperature variations of ≲400\lesssim400 K.

Modeling Day-Night Spectral Variation

We investigated the atmospheric structure of SDSS1411B by comparing the observed phase-resolved spectra to atmospheric model spectra. We first updated the system parameters required for the atmospheric modeling using the latest values reported in the literature in § 4.1. We then performed one-dimensional (1D) forward modeling in § 4.2, followed by atmospheric retrieval analysis in § 4.3. Finally, based on the spectral modeling results, we investigated the energy balance between the day- and nightside hemispheres of SDSS1411B in § 4.4.

Updating System Parameters

We adopted or derived the system parameters required for the atmospheric modeling using measurements reported in previous studies. We first adopted the Gaia DR3 distance of 177 ±\pm 4 pc ([57]). We then derived a BD mass of 54.7 ±\pm 2.1 MJM_{\mathrm{J}} using the binary mass function,

PK132πG=MBD3sin⁡3i(MWD+MBD)2(2)\frac{P K_{1}^{3}}{2\pi G} = \frac{M_{\mathrm{BD}}^{3}\sin^{3}i}{\left(M_{\mathrm{WD}}+M_{\mathrm{BD}}\right)^{2}} \tag*{(2)}

adopting a WD mass of MWD=0.562±0.028 M⊙M_{\mathrm{WD}} = 0.562 \pm0.028\,M_{\odot} from O. Vincent et al. [123], a radial-velocity amplitude of K1=35.0±0.7K_{1} = 35.0 \pm0.7 km s−1^{-1} and orbital inclination of i=86.5±0.4∘i = 86.5 \pm0.4^{\circ} from S. P. Littlefair et al. [78], and an orbital period of P=0.08453275006(4)P = 0.08453275006(4) day from A. Yates et al. [130]. We then estimated the orbital separation to be a=0.69±0.01 R⊙a = 0.69 \pm0.01\,R_{\odot} by applying Kepler’s third law using the adopted orbital period and the WD and BD masses. Combining this separation with the eclipse-fit constraint RBD/a=0.107±0.005R_{\mathrm{BD}}/a = 0.107 \pm0.005 from S. P. Littlefair et al. [78], we inferred a BD radius of RBD=0.717±0.035 RJR_{\mathrm{BD}} = 0.717 \pm0.035\,R_{\mathrm{J}}. Finally, using the derived BD mass and radius, we calculated the surface gravity as log⁡g=log⁡(GMBD/RBD2)≈5.42\log g = \log(GM_{\mathrm{BD}}/R_{\mathrm{BD}}^{2}) \approx5.42, which we fixed in the atmospheric modeling.

Forward Modeling

Based on the derived system parameters, we first compared the observed BD spectra with the Sonora Bobcat models ([85]), a 1D cloudless grid assuming rainout chemical equilibrium. The model grid spans 2.5≤log⁡g≤5.52.5 \leq\log g \leq5.5 and 200≤Teff≤2,400200 \leq T_{\mathrm{eff}} \leq2{,}400 K. We interpolated the model spectra and fitted the observed spectra using the species package ([116]). We implemented the UltraNest nested sampling algorithm ([111, 29]) with 400 live points. We used a uniform prior on the effective temperature of 1,000≤Teff≤2,4001{,}000 \leq T_{\mathrm{eff}} \leq2{,}400 K and metallicity of −0.5≤[Fe/H]≤0.5-0.5 \leq[\mathrm{Fe}/\mathrm{H}] \leq0.5.

We also compared the BD spectra with 1D cloudy Sonora Diamondback models ([92]), which include effects of refractory cloud species. The model grid spans surface gravities from log⁡g=3.5\log g = 3.5 to 5.5 and effective temperatures from 900 to 2400 K. We adopted the same interpolation and fitting procedure used for the Sonora Bobcat modeling to fit the observed BD spectra.

The best-fit Sonora Bobcat model for the nightside spectrum has Teff=1268T_{\mathrm{eff}} = 1268 K and [Fe/H] = 0.5 (Figure 6a). The model overpredicts the J-band peak by ∼20%\sim20\% and does not reproduce the positive spectral slope at λ>4.5\lambda> 4.5 μ\mum. The best-fit Sonora Diamondback model (Teff=1218T_{\mathrm{eff}} = 1218 K, [Fe/H] = 0.5, fsed=8f_{\mathrm{sed}} = 8) provides similar spectral slope at λ>4.5\lambda> 4.5 μ\mum, but predicts weaker blueside peaks within the H2_2O absorption bands (∼1\sim1–22 μ\mum) than observed, possibly due to cloud-induced reddening. The discrepancy between the observed nightside spectrum and the cloudy Diamondback model with the least reddening and vertically thinnest clouds (fsed=8f_{\mathrm{sed}} = 8; see Figure 15b) suggests that the atmospheric conditions of SDSS1411B deviate from equilibrium cloud model predictions (see § 5.7 for further discussion of cloud abundances).

Figure 6. Nightside (Panel a) and dayside (Panel b) BD spectra compared with the best-fit cloudless Sonora Bobcat model (purple) and cloudy Sonora Diamondback model (cyan). The lower panels show the residuals between the observed BD spectra and the best-fit models. The discrepancies between the observed spectra and the equilibrium forward models may reflect departures from equilibrium atmospheric structures driven by strong external WD irradiation.

Figure 15. Panel (a): Cloud abundance profiles of MgSiO3_3 (solid) and Fe (dashed) for the Day (orange), Morning (green), and Night (blue) phases from the cloudy retrieval model with free CO2_2 chemistry, with shaded regions indicating the abundance uncertainties. MgSiO3_3 condensates form near ∼30\sim30 bar, whereas Fe clouds form at deeper atmospheric pressures of ∼100\sim100 bar. Panel (b): Cumulative cloud optical depth at 3 μ\mum for each longitude. The purple dashed line shows the cloud optical-depth profile of the Sonora Diamondback model with Teff=1200T_{\mathrm{eff}} = 1200 K and fsed=8f_{\mathrm{sed}} = 8 for comparison. Panel (c): Corresponding contribution functions integrated over 0.6–5.3 μ\mum.

The Sonora Bobcat model does not provide a satisfactory fit to the dayside spectrum, as expected because the Sonora models do not account for external irradiation. The model does not reproduce both the rising continuum slope and the prominent emission features at ∼3\sim3 and 5 μ\mum, with flux deviations exceeding ∼60%\sim60\% of the observed value. The cloudy Diamondback model produces weaker molecular absorption features, in better qualitative agreement with the observed dayside spectrum. However, the Diamondback model likewise does not reproduce the overall spectral shape, particularly near ∼5\sim5 μ\mum, where the predicted flux differs from the observations by more than ∼30%\sim30\%.

Atmospheric Retrieval Analysis

We further performed multi-phase joint retrievals of SDSS1411B to investigate the detailed atmospheric structures. Because the observed morning- and eveningside spectra are consistent (Figure 5), we considered three representative atmospheric longitudes: nightside (ϕ=0\phi= 0), morning (ϕ=π/2\phi= \pi/2), and dayside (ϕ=π\phi= \pi). Our joint retrieval framework simultaneously analyzes these atmospheric longitudes with multiple 1D models (e.g., D. Broski-Laing et al. [28]; S. Mukherjee et al. [94]). We constructed each 1D model using petitRADTRANS (v3; P. Mollière et al. [91]; E. Nasedkin et al. [95]), which supports a variety of opacity sources and flexible T–P structures. Following the procedures described in J. D. Lothringer et al. [83], we first adopted 80 vertical grid layers evenly spaced in a logarithmic scale of pressure, spanning from 10310^3 to 10−610^{-6} bars. We then calculated opacities using tabulated correlated-K tables with the model spectral resolution of R=100R = 100. Based

on the thermal profile construction and opacity calculation, we ran the retrievals with the MultiNest sampling algorithm (F. Feroz & M. P. Hobson [48]; F. Feroz et al. [49, 50]) using 1,000 live points. The details of modeling the T–P profiles and opacity sources are summarized below.

We modeled the 1D thermal profile of each atmospheric longitude of SDSS1411B using a gradient T–P profile (Z. Zhang et al. [135]). We first divided the vertical BD atmosphere into six primary pressure layers evenly spaced in log-pressure. We fitted the temperature at the bottom layer, TbottomT_{\mathrm{bottom}}, and the temperature gradients, d log T / d log P, defined at the boundaries between adjacent layers (namely, 10310^3, ∼16\sim16, ∼0.3\sim0.3, ∼4×10−3\sim4 \times10^{-3}, ∼6×10−5\sim6 \times10^{-5}, and 10−610^{-6} bar). A quadratic interpolation was then applied to construct the detailed temperature structure across 80 evenly spaced subgrid layers spanning the six primary layers, given a set of d log T / d log P gradients (Z. Zhang et al. [135]). Based on the theoretical prediction from general circulation models (GCMs) (V. Parmentier et al. [99]; S. Lines et al. [77]), the phase-dependent T–P profiles were required to converge at the deepest atmosphere by adopting phase-independent parameters for TbottomT_{\mathrm{bottom}} and d log T / d log P for the deepest layer at P = 10310^3 bar.

We then computed the abundances of opacity sources at each atmospheric layer. We included common absorbers in exoplanet and BD atmospheres assuming chemical equilibrium: H2_2O (O. L. Polyansky et al. 2018), CH4\mathrm{CH}_{4} [64], NH3\mathrm{NH}_{3} [38], HCN [13], H2S\mathrm{H}_{2}\mathrm{S} [11], FeH [126], PH3\mathrm{PH}_{3} [113], Na [3], K [2], TiO [87], VO [91], and SiO [132]. To investigate the disequilibrium processes suggested in D. Broski-Laing et al. [28], we performed retrievals with free chemistry for CO (G. Li et al. 2015) and CO2\mathrm{CO}_{2} [131], implementing vertically constant abundances. Collision-induced absorption from H2\mathrm{H}_{2}–H2\mathrm{H}_{2} [24, 21] and H2\mathrm{H}_{2}–He [25, 23, 22], and Rayleigh scattering from H2\mathrm{H}_{2} and He were included as opacity sources [41, 37]. We also added H−\mathrm{H}^{-} bound-free and free-free absorption following D. F. Gray [63]. Lastly, for cloudy models, we included cloud species Fe and MgSiO3\mathrm{MgSiO}_{3}

with crystalline structures, following the cloud model of A. S. Ackerman & M. S. Marley [1]. To avoid interpreting potential scattered or reflected WD irradiation as thermal emission from the BD, we excluded wavelengths shorter than 1 µm from the retrieval analysis.

We performed retrievals for absorber abundances both in chemical equilibrium and disequilibrium parameterizations. The absorber abundances in equilibrium were determined by adding [Fe/H] and C/O as free parameters. We accounted for the chemical disequilibrium using the logarithmic quench pressure, log⁡Pquench\log P_{\mathrm{quench}} (e.g., K. J. Zahnle & M. S. Marley [134]). Cloud opacity was parameterized by the width of the cloud particle size distribution, σg\sigma_g, the vertical cloud mixing coefficient, KzzK_{\mathrm{zz}}, and the sedimentation efficiency, fsedf_{\mathrm{sed}} (A. S. Ackerman & M. S. Marley [1]). We also determine the condensate abundance at the cloud base using a free parameter X0X_0 to trace the cloud abundances in each orbital phase. Table 2 summarizes the model parameter priors adopted in the retrieval analysis.

ParameterPrior
log gg (cgs)5.42 (fixed)
DBDD_{\mathrm{BD}} (pc)177 (fixed)
RBDR_{\mathrm{BD}} (RJupR_{\mathrm{Jup}})N(μ=0.70,σ=0.04)\mathcal{N}(\mu= 0.70, \sigma= 0.04)
TbottomT_{\mathrm{bottom}} (K)U(0,8000)\mathcal{U}(0,8000)
d log T / d log PU(−1,1)\mathcal{U}(-1,1)
log Pquench_{\mathrm{quench}} (bar)U(−6,3)\mathcal{U}(-6,3)
[Fe/H]U(−1,2)\mathcal{U}(-1,2)
C/OU(0.05,1.45)\mathcal{U}(0.05,1.45)
log XCO2,global_{\mathrm{CO_2,global}}*U(−10,0)\mathcal{U}(-10,0)
log XCO2,phase_{\mathrm{CO_2,phase}}/XCO2,global_{\mathrm{CO_2,global}}*U(−5,5)\mathcal{U}(-5,5)
log XCO,global_{\mathrm{CO,global}}*U(−10,0)\mathcal{U}(-10,0)
log XCO,phase_{\mathrm{CO,phase}}/XCO,global_{\mathrm{CO,global}}*U(−5,5)\mathcal{U}(-5,5)
σlnorm\sigma_{\mathrm{lnorm}}**U(1.05,3)\mathcal{U}(1.05,3)
log Kzz_{\mathrm{zz}}**U(5,13)\mathcal{U}(5,13)
fsedf_{\mathrm{sed}}**U(0,10)\mathcal{U}(0,10)
log X0,Fe_{\mathrm{0,Fe}}**U(−10,0)\mathcal{U}(-10,0)
log X0,MgSiO3_{\mathrm{0,MgSiO_3}}**U(−10,0)\mathcal{U}(-10,0)
* For free chemistry models
** For cloudy models

Table 2. Priors for the Joint Retrieval.

We introduced additional flexibility in the chemistry model because initial fits assuming equilibrium chemistry struggled to reproduce the 4.2 µm CO2_2 feature in the nightside spectrum (Figure 7; see also the discussion of disequilibrium CO2_2 chemistry in ZTF0038B in D. Broski-Laing et al. [28]). To assess the robustness of our retrieval setup, we compared models with different treatments of CO and CO2_2 chemistry, as well as models with and without cloud opacity. Among the tested chemistry schemes, the model with free CO2_2 retrieval has the highest marginal likelihood (ln Z = 43965.2), providing strong evidence (R. Trotta [120]) over the free-CO model (ln Z = 43911.9), the model with both free CO and CO2_2 retrievals (ln Z = 43946.4), and the model without free retrievals (ln Z = 43914.4). We also found that including cloud components does not significantly improve the spectral fit relative to the cloudless model (Figure 7), despite the increased model complexity. Instead, the cloudy model is statistically disfavored, providing a lower marginal evidence (ln Z = 39452.1) than the cloudless model (ln Z = 43965.2). We therefore adopt the cloudless retrieval model with free CO2_2 chemistry as the preferred model for SDSS1411B. The best-fit parameters of the preferred model from the joint retrieval analysis are summarized in Table 3.

Figure 7. Phase-resolved BD spectra for the Day (orange), Morning (green), and Night (blue) phases, compared with the best-fit retrieval models without clouds and with free CO2\mathrm{CO}_{2} chemistry (black), with clouds and free CO2\mathrm{CO}_{2} chemistry (magenta), and without clouds and with equilibrium chemistry only (lightgreen). The lower panels show the residuals between the observed and model spectra. Relative to the best-fit cloudless model with free CO2\mathrm{CO}_{2} chemistry, including cloud opacity does not significantly improve the spectral fit, whereas the equilibrium-chemistry model does not reproduce the nightside CO2\mathrm{CO}_{2} features near 4 μ\mum.

Global Parameters
TbottomT_{\mathrm{bottom}}(dlog⁡Tdlog⁡P)1\left(\frac{\mathrm{d}\log T}{\mathrm{d}\log P}\right)_1[Fe/H]C/ORBDR_{\mathrm{BD}}
4874±2440.411±0.0280.226±0.0400.512±0.0120.693±0.004
Phase-dependent T–P Profile Properties
Phase(dlog⁡Tdlog⁡P)2\left(\frac{\mathrm{d}\log T}{\mathrm{d}\log P}\right)_2(dlog⁡Tdlog⁡P)3\left(\frac{\mathrm{d}\log T}{\mathrm{d}\log P}\right)_3(dlog⁡Tdlog⁡P)4\left(\frac{\mathrm{d}\log T}{\mathrm{d}\log P}\right)_4(dlog⁡Tdlog⁡P)5\left(\frac{\mathrm{d}\log T}{\mathrm{d}\log P}\right)_5
(dlog⁡Tdlog⁡P)6\left(\frac{\mathrm{d}\log T}{\mathrm{d}\log P}\right)_6CO2_2log⁡Pquench\log P_{\mathrm{quench}}
Night0.1701±0.00190.395±0.0550.50±0.32-0.01±0.53
Morning0.1342±0.00210.0501±0.0046-0.331±0.016-0.04±0.14
Day0.1061±0.0024-0.0746±0.0036-0.3773±0.00930.32±0.16
0.03±0.54-4.364±0.0670.551±0.033
0.08±0.54-4.514±0.134-1.19±0.40
-0.02±0.56-8.77±1.74-3.9±1.0

Table 3. Best-Fit Parameters for the Joint Retrieval.

The atmospheric retrievals provide generally good fits to the observed multi-phase spectra (Figure 7). The retrieved nightside spectrum reproduces the observed positive continuum slope at long wavelengths (≳4\gtrsim4 µm). Furthermore, the retrieved dayside spectrum successfully reproduces the emission features at ∼\sim3 and 5 µm, as well as the shallow H2_2O absorption bands between 1 and 2 µm, with relative discrepancies of ≲20%\lesssim20\%.

We plot the wavelength-dependent contribution functions of the best-fit retrieval model in Figure 8 to trace the relative contribution of each pressure layer to the observed spectra. At P∼1P \sim1–100 bar, the photospheric pressures vary inversely with molecular absorption strength (Figure 5), with stronger absorption features probing lower-pressure layers. The dayside and morningside show additional contributions from low-pressure layers at P∼10−3P \sim10^{-3}–10−210^{-2} bar, possibly probing upper atmospheric layers strongly affected by external irradiation. Our results show that, at wavelengths ≳4\gtrsim4 µm, a relatively broad range of pressures ≲1\lesssim1 bar contributes to the emerging dayside and morningside spectra, compared to those probed at shorter wavelengths (see also Figure 9c and Figure 16c).

Figure 8. Contribution functions of the best-fit joint retrieval model. Colors indicate the relative normalized contribution of each pressure layer to the observed spectra at the given wavelength. The nightside contribution is concentrated at high pressures of P ∼ 1–100 bar, while the dayside and morningside show bimodal contributions from both the deep photosphere (P ∼ 1–100 bar) and the inversion region (P∼10^{-3}–10^{-2} bar).

Figure 9. Panel (a): Best-fit T–P profiles for the Day (orange), Morning (green), and Night (blue) phases. The dashed curves show the Fe (cyan) and MgSiO3_3 (yellow) condensation curves, indicating potential cloud formation at P∼102P \sim10^2 bar. The black dash-dotted curve shows the Sonora Bobcat model (Teff=1276T_{\mathrm{eff}} = 1276 K, log⁡g=5.42\log g = 5.42), which is consistent with the retrieved nightside T–P profile at P≳1P \gtrsim1 bar. Panel (b): Radiative timescales derived from the retrieved T–P profiles. At P≳1P \gtrsim1 bar, the radiative timescale increases monotonically with pressure. Panel (c): Corresponding contribution functions integrated over 0.6–5.3 µm.

Figure 16. Panel (a): CO2\mathrm{CO_2} abundance profiles for retrievals with equilibrium chemistry (dashed lines) and free CO2\mathrm{CO_2} chemistry (solid lines). The best-fit free-CO2\mathrm{CO_2} retrieval favors a lower CO2\mathrm{CO_2} abundance on the dayside than on the morning- and nightsides, while the nightside CO2\mathrm{CO_2} abundance exceeds that predicted by the equilibrium model. Panel (b): Temperature-gradient profiles derived from the retrieved T–P profiles at each longitude (Figure 9a). The dashed line indicates the adiabatic gradient adopted from V. Parmentier et al. [97]. The nightside profile exceeds the adiabatic gradient over ∼0.001\sim0.001–11 bar, suggesting potential convective instability. Panel (c): Corresponding contribution functions integrated over 44–5.35.3 µm, including the CO2\mathrm{CO_2} band.

Figure 9a shows the retrieved T–P profiles and corresponding contribution functions, obtained by integrating the wavelength-dependent contribution functions in Figure 8. The nightside exhibits a monotonic thermal structure, whereas the morning and dayside profiles display a prominent temperature inversion at ∼\sim0.1 and 1 bar, respectively. The inversion occurs at higher pressure on the dayside than on the morningside, potentially due to stronger WD irradiation on the dayside. The MgSiO3_3 and Fe condensation curves intersect the retrieved T–P profiles at ∼\sim10 and 10210^2 bar, respectively, suggesting cloud formation in the deep atmosphere. At the pressures probed by the nightside observation (∼\sim10 bar), where the nightside contribution function peaks, the retrieved nightside T–P profile is consistent with that of a Sonora Bobcat model with Teff=1268T_{\mathrm{eff}} = 1268 K and log⁡g=5.42\log g = 5.42, adopted from the best-fit model from § 4.2.

To investigate the origin of the wavelength-dependent spectral modulations in the phase-resolved BD spectra, we calculated the radiative time constant at each pressure level based on the retrieved T–P profiles. Following [61], the radiative time constant is given by τ=1/N=cpP/(4σT3g)\tau= 1/N = c_{\mathrm{p}}P/(4\sigma T^3g), where N is the planetary mean relaxation rate, cpc_{\mathrm{p}} is the heat capacity at constant pressure, and g is the surface gravity. Figure 9b shows the radiative time constant profiles derived from the retrieved T–P profiles. The dayside and morning profiles exhibit increasing radiative time constants with pressure, indicating slower thermal relaxation in deeper atmospheric layers. The nightside profile shows a similar trend at P≳1P \gtrsim1 bar, where the contribution function peaks (Figure 7). This decreasing trend in timescales at lower pressures indicates a rapid thermal response, large flux and temperature variations in the upper atmosphere. Similar pressure-dependent radiative responses are also found in GCMs of hot Jupiters (e.g., [98]).

Energy Transport Balance Analysis

To investigate the energy balance of SDSS1411B based on the retrieved model, we calculated the energy budgets of the nightside and dayside hemispheres under a set of assumptions. We modeled the BD as four equal longitudinal quarters representing the dayside, nightside, morningside, and eveningside, assuming identical spectra for the morning and evening regions. The dayside hemisphere therefore consists of one quarter emitting the dayside spectrum plus one-half of each of the morning- and eveningside quarters, while the nightside hemisphere consists of one quarter emitting the nightside spectrum plus the remaining halves of the morning- and eveningside quarters. We further assumed that the same amount of internal energy is emitted from both hemispheres, that WD irradiation is absorbed only on the dayside hemisphere, and that a fraction of the absorbed energy is redistributed to the nightside. Finally, imposing global energy balance, we solved for the heat-redistribution efficiency, defined as the ratio of the energy transported from the dayside to the nightside to the total absorbed irradiation energy.

We first constrained the emitted energy from the two hemispheres based on the observed phase-resolved spectra. We derived the bolometric luminosity of each hemisphere by accounting for flux emitted outside the NIRSpec/PRISM wavelength coverage. We performed the bolometric correction of the observed spectra using the best-fit petitRADTRANS retrieval model spectra spanning 0.50.5–5050 μ\mum. Based on the ratio between the observed and the model fluxes, we estimated that the observed bandpass (0.60.6–5.35.3 μ\mum) captures ∼84\sim84, 9292, and 95%95\% of the BD’s total bolometric flux at day, morning, and night, respectively. The resulting effective temperatures are 1703±581703 \pm58 K, 1446±501446 \pm50 K, and 1272±481272 \pm48 K for the dayside, morningside, and nightside, respectively.

We then constrained the three components of the BD’s energy budget: the irradiation, internal, and circulation luminosity. We first calculated the irradiation luminosity (LirrL_{\rm irr}), defined as the WD luminosity absorbed by the BD:

Lirr=(1−A)×Finc×πRBD2(3)L_{\rm irr} = (1-A) \times F_{\rm inc} \times\pi R_{\rm BD}^{2} \tag*{(3)}

where AA is the Bond albedo, FincF_{\rm inc} is the incident WD flux, and RBDR_{\rm BD} is the BD radius. Assuming a global energy balance between the absorbed and emitted energy, the BD internal luminosity is given by

Lint=Lbol,day+Lbol,night−Lirr(4)L_{\rm int} = L_{\rm bol,day} + L_{\rm bol,night} - L_{\rm irr} \tag*{(4)}

where Lbol,dayL_{\rm bol,day} and Lbol,nightL_{\rm bol,night} are the measured bolometric luminosities of the dayside and nightside hemispheres, respectively. The circulation luminosity (LcirL_{\rm cir}), representing the energy transported from the dayside to the nightside, is then

Lcir=Lint2+Lirr−Lbol,day=Lbol,night−Lint2.(5)L_{\rm cir} = \frac{L_{\rm int}}{2} + L_{\rm irr} - L_{\rm bol,day} = L_{\rm bol,night} - \frac{L_{\rm int}}{2}. \tag*{(5)}

Finally, we defined the heat transport efficiency, η=Lcir/Lirr\eta= L_{\rm cir}/L_{\rm irr}, which quantifies the fraction of absorbed irradiation redistributed from the dayside to the nightside. We note that our simple model does not specifically consider the potential complication that some UV flux is absorbed and thermally re-emitted in the upper atmosphere; this would have an effect similar to a higher Bond Albedo even though it would not be from scattering. We also neglect energy transport through chemical processes, such as photodissociation on the dayside and subsequent recombination on the nightside.

We obtained each component of the energy budget based on our spectral analysis and system parameters derived in § 4.2. To estimate the irradiation luminosity from the WD, we adopted a WD effective temperature of 11,616±29911{,}616 \pm299 K, WD radius of $0.0135 \pm 0.0006 R⊙_{\odot} (O. Vincent et al. [123]), and orbital separation of 0.69±0.010.69 \pm0.01 R⊙_{\odot}. We estimated the day- and nightside bolometric luminosities by integrating the best-fit retrieval model spectra over 0.50.5–5050 μ\mum and adopting the Gaia DR3 distance of 177±4177 \pm4 pc (Gaia Collaboration et al. [57]). We then derived the internal and circulation luminosities based on Equations (3) and (5).

Figure 10 presents the derived energy budgets as a function of BD Bond albedo. The bolometric luminosity of the dayside hemisphere is ∼1.5×10−5\sim1.5 \times10^{-5} L⊙_{\odot}, approximately two times larger than that of the nightside hemisphere (∼8×10−6\sim8 \times10^{-6} L⊙_{\odot}). The absorbed irradiation luminosity decreases linearly with increasing Bond albedo, while the BD internal luminosity increases linearly with Bond albedo. The absorbed stellar irradiation remains comparable in magnitude to the nightside bolometric luminosity. The bottom panel of Figure 10 shows the heat transport efficiency between the dayside and nightside hemispheres. The maximum efficiency is ∼0.1\sim0.1 for zero Bond albedo. The heat transport efficiency decreases with increasing Bond albedo and becomes negative for Bond albedo values greater than ∼0.22\sim0.22, providing the upper limit of LintL_{\mathrm{int}} as ∼1.6×10−5\sim1.6 \times10^{-5} L⊙_{\odot}.

Figure 10. (Top) Energy budget components of SDSS1411B as a function of Bond albedo. (Bottom) Heat transport efficiency between the dayside and nightside hemispheres as a function of Bond albedo. The gray shaded region marks the unphysical regime where the inferred heat transport efficiency becomes negative.

The estimated internal luminosity provides a constraint on the evolutionary age of SDSS1411B. In Figure 11, we compare the inferred internal luminosity with thermal evolutionary model tracks. The figure shows

Figure 11. Sonora evolutionary track corresponding to the mass of SDSS1411B, 54.7±2.154.7 \pm2.1 MJ_{\mathrm{J}}. The horizontal shaded region indicates the 1σ1\sigma range of LintL_{\mathrm{int}} inferred from the energy transport analysis (see Figure 10 and § 4.4). The vertical dotted lines mark the BD cooling-age range corresponding to the overlap between the inferred LintL_{\mathrm{int}} and the Sonora evolutionary track. The gray shaded region marks the age constraint of ≳3\gtrsim3 Gyr from previous studies (S. P. Littlefair et al. [78]; B. W. P. Lew et al. [75]), which overlaps with our inferred cooling age of ∼3\sim3 Gyr.

the thermal evolutionary track from the Sonora Bobcat evolutionary models (M. S. Marley et al. [85]) for a BD with a mass of 54.7±2.154.7 \pm2.1 MJup_{\mathrm{Jup}}, derived in § 4.2. We linearly interpolated the evolutionary track with [Fe/H] of 0 and 0.5 to obtain the track with [Fe/H]=0.226=0.226, the best-fit metallicity from the retrieval analysis (Table 3). The inferred internal luminosities for the range of assumed Bond albedos constrain the BD cooling age to ∼2.5\sim2.5 Gyr. Considering the uncertainties in LintL_{\mathrm{int}} and the Sonora model parameters, this estimate is marginally consistent with the ∼3\sim3–10 Gyr ages inferred from the BD mass–radius relation in previous studies (S. P. Littlefair et al. [78]; B. W. P. Lew et al. [75]).

We further compared the derived BD surface gravity and radius with predictions from evolutionary models. Figure 12 shows the evolution of BD log gg and radius from the Sonora Bobcat models (M. S. Marley et al. [85]). The derived values of log⁡g=5.42±0.05\log g = 5.42 \pm0.05 and RBD=0.717±0.035RJR_{\mathrm{BD}} = 0.717 \pm0.035 R_{\mathrm{J}} from § 4.2 correspond to an age older than 10 Gyr, exceeding the upper limit of the model grid. This discrepancy between the age inferred from the internal luminosity and BD compactness may indicate either a more compact structure or a higher internal luminosity than expected for an isolated BD of the same mass.

Sonora Bobcat evolutionary tracks for BD radius and log g

Figure 12. Sonora Bobcat evolutionary tracks for BD radius (Panel a) and log gg (Panel b) corresponding to the mass of SDSS1411B, 54.7±2.154.7 \pm2.1 MJ_{\mathrm{J}}. The horizontal shaded regions indicate the 1σ1\sigma ranges of BD radius and log gg derived in § 4.2. The inferred radius and log gg do not overlap with the evolutionary model predictions, implying either a very old BD age of ≳10\gtrsim10 Gyr or a more compact BD structure than expected for an isolated BD of the same mass.

Discussion

Lightcurve Interpretation

We confirmed significant spectrophotometric variations in the SDSS1411 system (see Figure 1 and Figure 5), consistent with previous HST and ground-based observations. The BD flux varies by ∼48%\sim48\% from the nightside to the dayside at 1.11.1–1.6 μm1.6\ \mu\mathrm{m}, broadly consistent with the ∼38%\sim38\% variability in the HST synthetic photometry by B. W. P. Lew et al. [75]. We found larger flux variations at longer wavelengths, reaching ∼260%\sim260\% in the K band and ≳700%\gtrsim700\% at λ>4.5 μm\lambda> 4.5\ \mu\mathrm{m}, which may be attributed to the dayside temperature inversion or UV-induced fluorescence, as suggested by S. L. Casewell et al. [35, 34] based on ground-based observations.

The best-fit mid-eclipse time shows no significant phase offset from the trough of the Fourier model of the BD flux variation within the uncertainties (ϕ1∼ϕ2∼0\phi_1 \sim\phi_2 \sim0 in Figure 4, see § 3.3). The expected secondary-eclipse phase (orbital phase of 0.5) is also consistent with the peak of the best-fit Fourier model. This near-zero phase offset is consistent with the results of B. W. P. Lew et al. [75], suggesting that the hottest region of the BD is centered near the substellar point. The absence of a significant phase offset indicates that the radiative cooling timescale is shorter than the horizontal advection timescale, implying weak or insignificant equatorial (zonal) wind transport (see, e.g., X. Tan & A. P. Showman [117]; E. K. H. Lee et al. [73]).

White Dwarf and System Parameters

We find that the extracted WD spectrum presented in § 3.2 generally agrees with the theoretical WD model, with a relative difference of 2±14%2 \pm14\%. This agreement supports both the best-fit WD parameters of T=11,616 KT = 11{,}616\ \mathrm{K} and log⁡g=7.93\log g = 7.93 adopted from O. Vincent et al. [123] and the quality of the absolute flux calibration described in § 2. The WD radius adopted from O. Vincent et al. [123], 0.0135±0.0006R⊙0.0135 \pm0.0006 R_{\odot}, is also consistent within 1σ1\sigma uncertainties with the WD radius corresponding to the best-fit scale factor from § 3.2.

The WD temperature adopted in this study, derived from Gaia low-resolution spectrophotometry, is lower than the 13,000±300 K13{,}000 \pm300\ \mathrm{K} value reported by S. P. Littlefair et al. [78] based on VLT/X-shooter spectroscopy. This discrepancy may arise from the different distance estimates (190 pc190\ \mathrm{pc} from S. P. Littlefair et al. [78]) and systematic differences among WD parameter-estimation methods. However, the resulting change in the incident WD flux at the BD is only ∼0.2\sim0.2 dex, leaving our main conclusions of inefficient heat redistribution and a BD cooling age of ∼2.5\sim2.5 Gyr largely unchanged. Moreover, the differences in the adopted WD parameters are not expected to significantly affect our primary conclusions regarding the atmospheric physics of the irradiated BD, which are based directly on the observed BD spectra.

We also find that the observed mid-eclipse time is consistent with the linear ephemeris from A. Yates et al. [130] at an epoch of ∼6×104\sim6 \times10^{4} orbital cycles. Combined with the mid-eclipse timing data from A. Yates et al. [130], we found no significant statistical evidence for orbital evolution over the current observational baseline of ∼10\sim10 years, despite the very short orbital period of ∼2\sim2 hr. To further assess the stability of the SDSS1411 system, we calculated the secular evolution of the orbital period for a circular orbit given by (L. Blanchet [20]):

P˙=−192π5c5(2πGP)5/3MWDMBD(MWD+MBD)1/3∼−4.6×10−14.(6)\dot{P}=-\frac{192\pi}{5c^{5}}\left(\frac{2\pi G}{P}\right)^{5/3}\frac{M_{\mathrm{WD}}M_{\mathrm{BD}}}{(M_{\mathrm{WD}}+M_{\mathrm{BD}})^{1/3}}\sim-4.6\times10^{-14}. \tag*{(6)}

Substituting the parameters for the SDSS1411 system, we estimate a cumulative phase shift of Δt=12P˙tepoch=571492∼0.5 s\Delta t = \frac{1}{2}\dot{P}t_{\mathrm{epoch}=57149}^{2} \sim0.5\ \mathrm{s}, which would be difficult to robustly detect from a single mid-eclipse measurement. The corresponding orbital decay timescale due solely to gravitational wave emission (P. C. Peters [102]) is

T=5c5a4256G31MWDMBD(MWD+MBD)∼1.9 Gyr,(7)T=\frac{5c^{5}a^{4}}{256G^{3}}\frac{1}{M_{\mathrm{WD}}M_{\mathrm{BD}}(M_{\mathrm{WD}}+M_{\mathrm{BD}})}\sim1.9\ \mathrm{Gyr}, \tag*{(7)}

which is comparable to the estimated evolutionary age of SDSS1411B of ∼2.5\sim2.5 Gyr. Thus, although gravitational-wave emission should drive secular orbital decay in this short-period and close binary, the expected timing signal over the current observational baseline of ∼10\sim10 yr is too small to be detected.

Phase-dependent Brown Dwarf Spectra

The Fourier model in § 3.3 robustly reproduces the observed spectra, and the Fourier amplitudes trace the flux variability of different atmospheric layers. The correlations between the Fourier amplitudes of the stationary (A0A_0) and time-varying components (A1A_1 and A2A_2) reflect the interplay between photospheric pressure at different wavelengths and the corresponding radiative cooling timescale. The opposite wavelength dependence of ∣A0∣|A_0| and ∣A1∣|A_1| shown in Figure 4(b) indicates that spectral regions with stronger absorption exhibit larger variability amplitudes. Wavelengths with high molecular opacity generally probe higher altitudes in the atmosphere (e.g., H. Yang et al. [129]) and correspond to lower photospheric pressures. As shown in Figure 9(b), lower pressure is associated with shorter radiative timescales, which tend to indicate larger flux and temperature variations. This naturally explains the observed correlation between molecular absorption strength and variability amplitude and is also consistent with similar pressure-dependent trends in phase-curve offsets found in recent hot-Jupiter GCMs (e.g., L.-P. Coulombe et al. [40]).

Brightness temperature measurements provide an additional probe of flux variability across different atmospheric pressure levels. The ∼3\sim3 and 5 μm5\ \mu\mathrm{m} bands exhibit the lowest nightside brightness temperatures (Figure 5), suggesting that they probe relatively low-pressure layers at ∼1\sim1 bar given the monotonic nightside TT–PP profile (Figure 8 and Figure 9a). These layers have relatively short nightside radiative timescales of ∼1\sim1 hour (Figure 9b) and are significantly influenced by external irradiation, consistent with the large flux and temperature variations observed at ∼3\sim3 and 5 μm5\ \mu\mathrm{m}. On the other hand, spectral regions with weaker molecular opacity, such as λ≲1.3\lambda\lesssim1.3, ∼1.6\sim1.6, 2.12.1, and 4 μm4\ \mu\mathrm{m}, probe relatively deeper atmospheric layers and exhibit smaller variations in flux and brightness temperature.

Nightside Spectrum: Field Brown Dwarfs and Atmospheric Models

The forward modeling (Figure 6a) and retrieval analysis (Figure 7) for the observed nightside spectrum provide different atmospheric properties and TT–PP profiles. The best-fit Sonora Bobcat forward model has Teff=1268 KT_{\mathrm{eff}} = 1268\ \mathrm{K}, in good agreement with the retrieved effective temperature of ∼1271 K\sim1271\ \mathrm{K}. However, the forward model converges on the highest metallicity value in the model grid ([Fe/H]=0.5[\mathrm{Fe}/\mathrm{H}] = 0.5), exceeding the retrieved value of [Fe/H]∼0.3[\mathrm{Fe}/\mathrm{H}] \sim0.3 (Table 3), suggesting departures of the retrieval model from the equilibrium assumptions of the forward model. Furthermore, the retrieved and forward-model TT–PP profiles are consistent near P∼1P \sim1 bar but diverge at lower pressures. The steeper retrieved temperature gradient in the upper atmosphere may reflect the effects of external irradiation on the atmospheric structure of SDSS1411B, or systematics of the gradient TT–PP profile parameterization in pressure regions with small contribution function (Figure 9c).

Surprisingly, the nightside spectrum of the irradiated BD SDSS1411B resembles that of isolated BDs. The retrieved nightside effective temperature of ∼1270 K\sim1270\ \mathrm{K} corresponds to an early-T spectral type based on the spectral type–effective temperature relation of D. C. Stephens et al. [115]; T. J. Dupuy & M. C. Liu [43]. Figure 13 compares the JWST NIRSpec/PRISM spectra of T5, T2, and L9 field dwarfs from Z. Tu et al. (2025). The T2 spectrum closely matches the nightside spectrum of SDSS1411B, reproducing both the H2O\mathrm{H_2O} absorption bands at ∼1\sim1–2 μm2\ \mu\mathrm{m} and the CO2\mathrm{CO_2} feature at ∼4.2 μm\sim4.2\ \mu\mathrm{m}. In contrast, the T5 and L9 spectra do not reproduce the relative peak heights within the H2O\mathrm{H_2O} bands and the widths of the spectral features. The agreement between the isolated T2 dwarf spectrum and the SDSS1411B nightside spectrum suggests that stellar irradiation has a limited influence on the nightside atmosphere, likely due to inefficient day-to-night heat transport (see § 4.4) and the limited impact of common-envelope evolution on the BD spectral features (D. Broski-Laing et al. [28]).

Figure 13. Comparison of the nightside spectrum of SDSS1411B with field BD spectra from Z. Tu et al. (2025). The gray shaded region marks the reference wavelength range used to normalize each spectrum based on its integrated flux. Among the comparison spectra, the T2 dwarf provides the closest match to the observed nightside spectrum of SDSS1411B.

Dayside model fits

The non-irradiated Sonora forward models do not reproduce the observed dayside emission features at ∼3\sim3 and 5 μm5\ \mu\mathrm{m} in Figure 6. The lack of irradiation prevents the 1D models from forming any upper-atmosphere temperature inversion (M. S. Marley et al. [85]; C. V. Morley et al. [92]).

In contrast, the best-fit retrieval model provides a good fit to the dayside spectrum (Figure 7). The retrieved dayside T–P profile exhibits a hotter upper atmosphere than the corresponding night- and morningside profiles. The T–P profile also shows a significant temperature inversion at ∼ 10−110^{-1} bar, similar to those observed in highly irradiated hot Jupiters (e.g., T. M. Evans et al. [45]), explaining the prominent ∼2.8 μ\mum H2_2O and 4.6 μ\mum CO emission features. However, unlike any previous spectra of irradiated planets or BDs with inversions, SDSS1411B also exhibits molecular features in absorption, namely, H2_2O at 1.4 μ\mum. This can only be explained by a temperature inversion that exists at pressures probed by the 2.8 μ\mum H2_2O band but not at pressures probed by the 1.4 μ\mum band, placing a strict constraint on the shape of the T–P profile. Interestingly, the apparent absence of strong absorption or emission from the 1.9 μ\mum H2_2O band is consistent with it probing the isothermal “transition” region of the inversion.

While this behavior has never before been seen in an atmosphere, such a scenario for CO was hypothesized to explain the low K-band eclipse depth of WASP-18b compared to the high Spitzer 4.5 μ\mum eclipse depth (K. B. Sheppard et al. [108]). However, subsequent observations of WASP-18b with JWST/NIRISS/SOSS showed no evidence for a low eclipse depth at 2 μ\mum and were instead consistent with both fundamental and overtone CO bands in emission (L.-P. Coulombe et al. [39]).

SDSS1411B represents the lowest-temperature strongly irradiated substellar object known to exhibit a temperature inversion. Previous theoretical modeling has identified objects above irradiation temperatures of about 2000 K as being susceptible to temperature inversion due to the absorption of the intense incoming flux by short-wavelength absorbers like TiO, VO, Fe, and other metal species, present only in the gas phase for temperatures above refractory condensation temperatures ([67, 52, 80, 58]). With an equilibrium temperature around 1600 K, we expect the canonical optical absorbers to be condensed in the atmosphere of SDSS1411B.

Rather than the inversion being driven by such optical absorbers, we hypothesize that SDSS1411B’s temperature is driven by UV absorbers. More than 60% of the irradiation from its Teff∼12,000T_{\rm eff} \sim12{,}000 K WD host is emitted at wavelengths shorter than the optical, with the peak flux at about 0.25 μ\mum. Flux at such wavelengths is preferentially absorbed at lower pressures in the object’s atmosphere (J. D. Lothringer et al. [80]) and is more capable of driving upper atmosphere heating. This is supported by theoretical modeling of planets around different host stars (J. D. Lothringer & T. Barman [79]), as well as observations of such planets (e.g., F. Yan et al. [128]). At some point, the atmosphere at low pressures can become hot enough that H2_2O is dissociated. The dissociation can further drive a temperature inversion due to reduced ability of the atmosphere to cool via radiation ([100, 80]).

This scenario is supported by 1D PHOENIX atmosphere models including irradiation. Using the same setup as the models from (J. D. Lothringer & S. L. Casewell [81]), we irradiate SDSS1411B with a Teff∼T_{\rm eff} \sim 13000 K WD host star and assume dayside heat redistribution. We also include rainout, which prevents TiO, VO, Fe, and other metal species from existing in the gas phase at low pressures. Figure 14 shows that this model exhibits many of the same qualitative features as the observed dayside spectrum of SDSS1411B, including a strong temperature inversion, as well as the mix of absorption and emission features. We also compute an identical model but changed the irradiation spectrum to represent a Teff∼6000T_{\mathrm{eff}} \sim6000 K main-sequence host star. In this comparison model, no strong inversion forms and all major molecular features are found to be in absorption, supporting our hypothesis of a previously unobserved UV-driven temperature inversion in SDSS1411B.

Figure 14. Panel (a): Model spectra from 1D BD atmosphere models irradiated by a WD (blue) and a G-type star (orange), compared with the observed SDSS1411B dayside spectrum (gray). Similar to SDSS1411B, the WD-irradiated model exhibits both absorption bands at short wavelengths and prominent emission features at ∼3 and 4.5 μ\mum. Panel (b): Corresponding T–P profiles for the two irradiated models, along with the best-fit retrieval profile for the SDSS1411B dayside. Both the WD-irradiated model and the retrieved SDSS1411B dayside profile exhibit a pronounced temperature inversion near 1 bar.

Condensate Clouds

The atmospheric retrieval analysis in § 4.3 well reproduces the observed SDSS1411B spectra without cloud components. Including cloud opacity in the retrieval modeling does not provide a significant improvement in the spectral fit shown in Figure 7(a), providing lower marginal evidence. This perhaps unexpected absence of clouds can be understood considering additional modeling results: The retrieved T–P profile (Figure 9b) suggests that condensates may form in the deep atmosphere (between ∼30\sim30 and 300 bar pressures), whereas most of our observations probe pressures lower than 30 bar. Furthermore, the condensate abundance profile of the cloudy model (Figure 15a) indicates a negligible cloud layer near ∼30\sim30 bar, with a cloud-base condensate mass fraction less than ∼10−5\sim10^{-5}. The low cumulative cloud optical depth, <0.01< 0.01 at 100 bar, further supports the negligible effect of clouds on the observed spectra of SDSS1411B (Figure 15b). Together, these retrieval results suggest that cloud opacity is not required to reproduce the observed spectra of SDSS1411B.

Given the low nightside temperature, this may be surprising. We speculate that the strong external irradiation may impact cloud formation in SDSS1411B. In this scenario, irradiation would heat the upper atmosphere and evaporate condensates. This is shown in Figure 9(b), where the heated dayside T–P profile does not cross the MgSiO3_3 condensation curve beyond the 1σ1\sigma uncertainty, suppressing cloud formation on the dayside. This model result is consistent with the GCM studies of hot Jupiters predicting that refractory condensates from the cooler nightside can evaporate as they are transported toward the hotter dayside (e.g., [99, 96, 94]). In addition, the nightside cloud distribution may be affected by day-to-night material transport and disequilibrium chemistry (e.g., [88, 16]). Such irradiation-driven processes could alter the abundances and chemical phases of cloud-forming species, potentially further suppressing the formation or survival of extensive and homogeneous cloud layers.

One caveat in the above scenario is that the observed spectrum does not directly rule out the presence of clouds on the nightside. Rather, our results show that the observed spectra can be adequately reproduced by a cloudless model and that including cloud opacity is not statistically favored, because the additional cloud model parameters do not provide a significant improvement in the spectral fit. Future mid-infrared observations, such as with JWST/MIRI, will be important for testing the presence of clouds and constraining their properties in SDSS1411B by probing the characteristic cloud features at ≳5\gtrsim5 µm (e.g., D. Grant et al. [62]; L. Heinke et al. [65]).

CO2 Disequilibrium Chemistry

Our retrieval analysis statistically favors the model with a free, vertically constant CO2\mathrm{CO_2} mass fraction (see § 4.3). To further investigate potential CO2\mathrm{CO_2} disequilibrium chemistry, we compared the CO2\mathrm{CO_2} abundance profiles of the best-fit cloudless model with free CO2\mathrm{CO_2} chemistry and the fully equilibrium cloudless model in Figure 16a. The free-CO2\mathrm{CO_2} model retrieves a lower CO2\mathrm{CO_2} abundance on the dayside than on the morning and nightside hemispheres. This result remains broadly consistent with the equilibrium model because the dayside contribution function across the CO2\mathrm{CO_2} band is concentrated in the upper atmosphere (Figure 8 and Figure 16c), where the equilibrium CO2\mathrm{CO_2} abundance is already low. In contrast, the free-CO2\mathrm{CO_2} model retrieves a higher CO2\mathrm{CO_2} abundance on the nightside than predicted by the equilibrium model, which alleviates the discrepancy near the 44 µm CO2\mathrm{CO_2} feature (Figure 7a).

In fact, there are reasons why CO2\mathrm{CO_2} may be out of equilibrium: For example, the irradiation by the WD may reduce the CO2\mathrm{CO_2} abundance of the BD through thermal

and photochemical dissociation. Similar to the thermal dissociation of molecules in highly irradiated hot Jupiters (e.g., V. Parmentier et al. [100]; S. Gandhi et al. [59]), the high temperatures and prominent temperature inversion on the dayside can drive CO2\mathrm{CO_2} dissociation and deplete its abundance in the upper atmosphere. In addition, strong external UV irradiation from the WD can photochemically dissociate CO2\mathrm{CO_2} into CO in the dayside upper atmosphere (F. Tian [118]; K. Zahnle et al. [133]). Together, these irradiation-driven processes can drive the atmospheric composition away from thermochemical equilibrium and may contribute to the low CO2\mathrm{CO_2} abundance retrieved on the dayside.

Similarly, vertical convection and horizontal advection may also contribute to the departure of the nightside CO2\mathrm{CO_2} abundance from chemical equilibrium. To assess potential convective instability in SDSS1411B, we calculated the temperature gradients of the retrieved T–P profiles (Figure 9a) and compared them with the adiabatic gradient from V. Parmentier et al. [97], based on a fit to the high-pressure equation of state of D. Saumon et al. [107], ∇ad∼0.32−0.1(T/3000 K)\nabla_{\mathrm{ad}} \sim0.32 - 0.1(T/3000\ \mathrm{K}). The retrieved temperature gradient, ∇ret\nabla_{\mathrm{ret}}, is smaller than ∇ad\nabla_{\mathrm{ad}} at ∼1\sim1–100100 bar, consistent with the HST-based modeling results of B. W. P. Lew et al. [75]. In contrast, the nightside profile exhibits ∇ret>∇ad\nabla_{\mathrm{ret}} > \nabla_{\mathrm{ad}} over ∼10−3\sim10^{-3}–11 bar, indicating potential convective instability in the same pressure range where the retrieved CO2\mathrm{CO_2} abundance departs from equilibrium. Although the nightside temperature gradients remain highly uncertain in the upper atmosphere, this stability analysis suggests that atmospheric dynamics may contribute to the observed disequilibrium chemistry in SDSS1411B.

Thermal and Radius Evolution of SDSS1411B

Our inferred cooling age for SDSS1411B (§ 4.4) is marginally consistent with the ≳3\gtrsim3 Gyr age range estimated in previous studies [78, 75], although discrepancies remain among age estimates derived using different methods. The ≳3\gtrsim3 Gyr constraint from S. P. Littlefair et al. [78] is based on the BD mass–radius relation, yet their inferred BD radius is more than 3σ3\sigma smaller than predicted by the 3 Gyr I. Baraffe et al. [12] isochrone. Similarly, the Sonora evolutionary models (§ 4.4; M. S. Marley et al. [85]) predict a larger radius than the value derived in § 4.1. This compact radius is also inconsistent with the radius inflation predicted for irradiated BDs [93]. The discrepancy may reflect differences in atmospheric properties or irradiation conditions [14], including the much bluer incident spectrum of the WD compared with the cooler main-sequence stellar irradiation considered by S. Mukherjee et al. [93]. Furthermore, the rapid rotation of SDSS1411B may suppress vertical energy transport, potentially limiting irradiation-driven radius inflation [105, 106]. On the other hand, the inferred radius of SDSS1411B, ∼0.72 RJ\sim0.72\,R_{\mathrm{J}}, lies within the observed radius distribution of weakly irradiated transiting BDs [14]. The SDSS1411 system may have completed common-envelope evolution on a relatively short timescale, potentially ≲103\lesssim10^{3} yr [68, 70], after which WD irradiation may have only a limited effect on thermal and radius evolution of the BD during its post-common-envelope phase.

Despite the inferred BD cooling age from well-constrained LintL_{\mathrm{int}}, constraining the total system age remains challenging. Common-envelope evolution and subsequent tidal interactions may alter the BD radius and thermal state [104, 125, 75]. In addition, accretion simulations coupled with stellar evolution models predict that a substellar companion may accrete substantial mass during common-envelope evolution [36, 70], potentially obscuring its pre-common-envelope thermal history. Studies of close WD–BD binaries therefore suggest that the BD cooling age may provide only a lower limit on the system age if the BD was significantly reheated during common-envelope evolution. [32] (M. Zorotov & M. Schreiber 2022). Future detailed modeling of common-envelope and tidal evolution, combined with system parameters constrained by phase-resolved bolometric observations, will be essential for fully understanding the energy budget and evolutionary history of the SDSS1411 system.

SDSS1411B and Other Irradiated Brown Dwarfs

We compare SDSS1411B with other irradiated BDs to examine three key properties: (1) strong temperature inversion in the dayside, (2) the cloudless hemispheres, and (3) CO2\mathrm{CO}_{2} disequilibrium chemistry.

The strong dayside temperature inversion identified in § 3 and § 4 places SDSS1411B as the least strongly irradiated BD known to exhibit a significant dayside inversion. Other irradiated BDs with reported dayside inversions, including EPIC212227B, SDSS1557B, and WD0137B [82, 5], receive incident irradiation values of ≳109\gtrsim10^{9} erg s−1^{-1} cm−2^{-2}, higher than that of SDSS1411B, ∼108.6\sim10^{8.6} erg s−1^{-1} cm−2^{-2}. In contrast, less strongly irradiated systems such as NLTT5306B, GD1400B, and ZTF0038B [4, 6] (D. Broski-Laiing et al. 2026) show no evidence of dayside temperature inversions in their HST or JWST spectra. SDSS1411B therefore occupies an important irradiation regime for investigating the conditions under which stellar irradiation drives thermal inversions, particularly the role of short-wavelength heating in the upper atmosphere.

Previous studies revealed a substantial diversity in cloud distribution among WD-irradiated BDs. Atmospheric modeling of HST observations of SDSS1557B, which receives a higher irradiation flux than SDSS1411B, favors a cloudy nightside and cloud-free dayside [5]. On the other hand, HST spectra of the highly irradiated EPIC2122B are well reproduced without including cloud opacity [82]. Among more weakly irradiated systems, WD1032+011B exhibits a cloudy nightside and clear dayside [53], suggesting dayside cloud dissipation for weaker irradiation, whereas GD1400B, with the weakest irradiation flux, shows evidence for global cloud coverage on both hemispheres [46, 47, 6]. These contrasting results indicate no clear correlation between cloud coverage and irradiation strength, suggesting that additional parameters, such as BD mass, effective temperature, or spectral type, may also regulate cloud formation. Systematic cloud analyses based on broad-wavelength spectroscopy of a larger sample of irradiated BDs will therefore be essential for determining how external irradiation influences phase-dependent cloud formation.

D. Broski-Laing et al. [28] presented the first JWST/NIRSpec phase curve of the irradiated BD ZTF0038B, which orbits an ∼\sim11,000 K WD at a separation of ∼\sim2 R⊙_{\odot} and therefore receives a lower irradiation flux (∼107.6\sim10^{7.6} erg s−1^{-1} cm−2^{-2}) than SDSS1411B (∼108.6\sim10^{8.6} erg s−1^{-1} cm−2^{-2}). Their observations revealed asymmetric CO2_2 absorption-band phase curves and nearly homogeneous CO2_2 abundances across orbital phase. Similar to SDSS1411B, ZTF0038B shows departures from chemical-equilibrium predictions, potentially driven by vertical convection and horizontal advection (T. J. Bell et al. [16]; V. Parmentier et al. [101]). However, unlike SDSS1411B, ZTF0038B exhibits relatively homogeneous CO2_2 abundances across longitude, possibly reflecting more efficient horizontal advection or weaker external irradiation that is insufficient to drive substantial thermal or photochemical dissociation on the dayside.

Studying a broader sample of irradiated BDs will provide a foundation for understanding irradiation-driven atmospheric dynamics and disequilibrium chemistry in the high-gravity, rapidly rotating, and high-internal-luminosity regime. Analysis of the three remaining irradiated BDs in JWST program GO-4967, WD1202, WD0137, and WD0137 (French et al., Broski-Laing et al., and Shin et al., in prep.), will enable a more systematic investigation of the atmospheric physics across different irradiation environments. Future discoveries of additional WD–BD systems through optical photometric surveys such as ZTF and ATLAS (e.g., J. van Roestel et al. 2021; K. W. Smith et al. [112]) and infrared surveys (e.g., R. F. Wilson et al. [127]; J. Gagné et al. [56]) will further expand the sample and improve its statistical homogeneity.

Conclusions

We presented time-resolved JWST/NIRSpec PRISM spectrophotometry of SDSS1411B, which occupies an intermediate irradiation regime among five close WD–BD binaries in the GO-4967 program (PI: Yifan Zhou). Our atmospheric analysis reveals atmospheric dynamics, thermal inversion, and disequilibrium chemistry in this strongly UV-irradiated BD. The key findings of this study are as follows:

  • The SDSS1411 system shows an average variability amplitude of ∼\sim14% of the total flux, increasing from a trough-to-peak variation of ∼\sim2% at 0.6–1.47 μ\mum to ∼\sim54% at 4.40–5.20 μ\mum.

  • The second-order Fourier model successfully reproduces the observed BD light curve. Wavelength bins with significant correlations between Fourier amplitudes of stationary and time-varying components coincide with wavelength regions dominated by H2_2O, CH4_4, and CO absorption features at ∼\sim1–2.4 and ∼\sim4 μ\mum, suggesting larger flux variations in the upper atmosphere than in deeper atmospheric layers.

  • The BD brightness temperatures range from ∼\sim600 to 2,000 K on the nightside and from ∼\sim1,500 to 2,000 K on the dayside. The spectral regions near ∼\sim3 and 5 μ\mum exhibit the lowest nightside brightness temperatures (∼\sim600 K) and probe relatively low-pressure atmospheric layers (∼\sim1 bar) with short radiative timescales (≲\lesssim2 hr), consistent with the large flux and brightness temperature variations observed at these wavelengths.

  • The BD nightside spectrum is inconsistent with the non-irradiated 1D forward models, showing a deviation from the equilibrium assumptions possibly due to the irradiation effects. In contrast, the atmospheric retrieval analysis successfully reproduces the nightside BD spectra with a cloud-free model and free CO2_2 chemistry. The retrieved nightside model shows a monotonic T–P profile, showing departure from that of the thermochemical-equilibrium model at ≲\lesssim1 bar.

  • The BD dayside spectrum is not well fit by non-irradiated 1D forward models. In contrast, the atmospheric retrieval analysis provides a cloudless best-fit model spectrum, in agreement with the observed dayside spectra. The retrieved dayside T–P profile shows a prominent temperature inversion near ∼\sim0.1 bar, suggesting that SDSS1411B is the lowest-temperature irradiated object known to exhibit a thermal inversion.

  • The phase-resolved spectra of SDSS1411B can be adequately reproduced by both cloudy and cloudless models, although the retrieval statistically favors the cloudless model. The best-fit cloudy model expects negligible condensate mass fractions (≲10−4\lesssim10^{-4}) and low cumulative cloud optical depths (≲10−2\lesssim10^{-2}) across all hemispheres, indicating that clouds have little effect on the observed spectra.

  • The free-CO2_2 chemistry model provides a better fit to the observed spectra than the equilibrium-chemistry model, suggesting irradiation-driven

disequilibrium chemistry in SDSS1411B, potentially including thermal and photochemical CO2CO_2 dissociation on the dayside and vertical mixing on the nightside.

  • The bolometric correction of the observed phase-resolved BD spectra provides effective temperatures of 1703±581703 \pm58 K for the dayside and 1272±481272 \pm48 K for the nightside. Our energy transport analysis indicates inefficient heat redistribution between the two hemispheres, with a transport efficiency of <0.1< 0.1. The inferred internal luminosity constrains the BD cooling age to ∼3\sim3 Gyr, while the derived BD radius is significantly smaller than predicted by evolutionary models.

  • Comparisons between SDSS1411B and other irradiated BDs in terms of (1) dayside temperature inversions, (2) cloud coverage, and (3) CO2CO_2 disequilibrium chemistry establish SDSS1411B as an important laboratory in a unique irradiation regime for investigating how external irradiation shapes substellar atmospheres.

We present high-quality, phase-resolved JWST spectrophotometry of the close WD–BD binary SDSS1411. Our multi-phase joint retrieval framework successfully reproduces the observed spectra and provides robust constraints on the atmospheric properties across different orbital phases. We identify a prominent dayside temperature inversion, largely cloud-free hemispheres, and CO2CO_2 disequilibrium chemistry. These results highlight the potential of JWST observations of close WD–BD systems, providing new insights into how strong external irradiation shapes substellar atmospheres.

ACKNOWLEDGMENTS

We thank the anonymous referee for constructive comments that improved the manuscript. This work is based on observations made with the NASA/ESA/CSA James Webb Space Telescope. National Aeronautics and Space Administration supported this work through a grant under the JWST-GO-4967 program from the Space Telescope Science Institute and through agreement No. 80NSSC21K0593 for the program “Alien Earths.” The data were obtained from the Mikulski Archive for Space Telescopes at the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 5-03127 for JWST. The specific observations analyzed in this study can be accessed via doi: 10.17909/zrd9-d176. SX is supported by the international Gemini Observatory, a program of NSF NOIRLab, which is managed by the Association of Universities for Research in Astronomy (AURA) under a cooperative agreement with the U.S. National Science Foundation, on behalf of the Gemini partnership of Argentina, Brazil, Canada, Chile, the Republic of Korea, and the United States of America. AI tools were used to suggest language improvements within the manuscript.

AUTHOR CONTRIBUTIONS

LS reduced and analyzed the data, carried out the forward modeling and retrieval studies, interpreted the model results, and drafted the manuscript. DA, YZ, and DB-L contributed significantly to the data analysis and modeling strategy, interpretation, and provided comments to the manuscript draft. YZ led the proposal and planned the observations. DA advised LS. JL contributed significantly to the atmospheric modeling studies and drafted the discussion of the dayside model fits. SC, SX, and BL contributed to the characterization and interpretation of the WD-BD system parameters and provided comments to the manuscript draft. MSM, LM, XT, JF, and VP contributed to the atmospheric modeling and interpretation of the BD spectrum. XT, JF, and VP also reviewed the manuscript and provided comments.

References

  1. [1]Ackerman, A. S., & Marley, M. S. 2001, ApJ, 556, 872, doi: 10.1086/321540DOI
  2. [2]Allard, N. F., Spiegelman, F., & Kielkopf, J. F. 2016, A&A, 589, A21, doi: 10.1051/0004-6361/201628270DOI
  3. [3]Allard, N. F., Spiegelman, F., Leininger, T., & Molliere, P. 2019, A&A, 628, A120, doi: 10.1051/0004-6361/201935593DOI
  4. [4]Amaro, R. C., Apai, D., Zhou, Y., et al. 2023, ApJ, 948, 129, doi: 10.3847/1538-4357/acbfb3DOI
  5. [5]Amaro, R. C., Apai, D., Lew, B. W. P., et al. 2024, ApJ, 966, 4, doi: 10.3847/1538-4357/ad354cDOI
  6. [6]Amaro, R. C., Apai, D., Zhou, Y., et al. 2025, ApJ, 979, 231, doi: 10.3847/1538-4357/ada295DOI
  7. [7]Apai, D., Nardiello, D., & Bedin, L. R. 2021, ApJ, 906, 64, doi: 10.3847/1538-4357/abcb97DOI
  8. [8]Apai, D., Radigan, J., Buenzli, E., et al. 2013, ApJ, 768, 121, doi: 10.1088/0004-637X/768/2/121DOI
  9. [9]Apai, D., Karalidi, T., Marley, M. S., et al. 2017, Science, 357, 683, doi: 10.1126/science.aam9848DOI
  10. [10]Artigau, É., Bouchard, S., Doyon, R., & Lafrenière, D. 2009, ApJ, 701, 1534, doi: 10.1088/0004-637X/701/2/1534DOI
  11. [11]Azzam, A. A. A., Tennyson, J., Yurchenko, S. N., & Naumenkо, O. V. 2016, MNRAS, 460, 4063, doi: 10.1093/mnras/stw1133DOI
  12. [12]Baraffe, I., Chabrier, G., Barman, T. S., Allard, F., & Hauschildt, P. H. 2003, A&A, 402, 701, doi: 10.1051/0004-6361:20030252DOI
  13. [13]Barber, R. J., Strange, J. K., Hill, C., et al. 2014, MNRAS, 437, 1828, doi: 10.1093/mnras/stt2011DOI
  14. [14]Beatty, T. G. 2026, arXiv e-prints, arXiv:2607.27352, doi: 10.48550/arXiv.2607.27352DOI
  15. [15]Bell, T., Arher, E.-M., Brande, J., et al. 2022, The Journal of Open Source Software, 7, 4503, doi: 10.21105/joss.04503DOI
  16. [16]Bell, T. J., Crouzet, N., Cubillos, P. E., et al. 2024, Nature Astronomy, 8, 879, doi: 10.1038/s41550-024-02230-xDOI
  17. [18]Biller, B. A., Vos, J. M., Zhou, Y., et al. 2024, MNRAS, 532, 2207, doi: 10.1093/mnras/stae1602DOI
  18. [19]Birkmann, S. M., Ferruit, P., Giardino, G., et al. 2022, A&A, 661, A83, doi: 10.1051/0004-6361/202142592DOI
  19. [20]Blanchet, L. 2014, Living Reviews in Relativity, 17, 2, doi: 10.12942/lrr-2014-2DOI
  20. [21]Borysow, A. 2002, A&A, 390, 779, doi: 10.1051/0004-6361:20020555DOI
  21. [22]Borysow, A., & Frommhold, L. 1989, ApJ, 341, 549, doi: 10.1086/167515DOI
  22. [23]Borysow, A., Frommhold, L., & Moraldi, M. 1989, ApJ, 336, 495, doi: 10.1086/167027DOI
  23. [24]Borysow, A., Jorgensen, U. G., & Fu, Y. 2001, JQSRT, 68, 235, doi: 10.1016/S0022-4073(00)00023-6DOI
  24. [25]Borysow, J., Frommhold, L., & Birnbaum, G. 1988, ApJ, 326, 509, doi: 10.1086/166112DOI
  25. [26]Brandt, T. D. 2024a, PASP, 136, 045005, doi: 10.1088/1538-3873/ad38daDOI
  26. [28]Broski-Laing, D., Zhou, Y., Lothringer, J. D., et al. 2026, arXiv e-prints, arXiv:2606.30112, doi: 10.48550/arXiv.2606.30112
  27. [29]Buchner, J. 2021, The Journal of Open Source Software, 6, 3001, doi: 10.21105/joss.03001DOI
  28. [30]Buenzli, E., Apai, D., Radigan, J., Reid, I. N., & Flateau, D. 2014, ApJ, 782, 77, doi: 10.1088/0004-637X/782/2/77DOI
  29. [31]Buenzli, E., Apai, D., Morley, C. V., et al. 2012, ApJL, 760, L31, doi: 10.1088/2041-8205/760/2/L31DOI
  30. [32]Burleigh, M. R., Hogan, E., Dobbie, P. D., Napiwotzki, R., & Maxted, P. F. L. 2006, MNRAS, 373, L55, doi: 10.1111/j.1745-3933.2006.00242.xDOI
  31. [33]Bushouse, H., Eisenhamer, J., Dencheva, N., et al. 2023, JWST Calibration Pipeline, 1.11.2 Zenodo, doi: 10.5281/zenodo.8140011
  32. [34]Casewell, S. L., Littlefair, S. P., Parsons, S. G., et al. 2018, MNRAS, 481, 5216, doi: 10.1093/mnras/sty2599DOI
  33. [35]Casewell, S. L., Lawrie, K. A., Maxted, P. F. L., et al. 2015, MNRAS, 447, 3218, doi: 10.1093/mnras/stu2721DOI
  34. [36]Chamandy, L., Frank, A., Blackman, E. G., et al. 2018, MNRAS, 480, 1898, doi: 10.1093/mnras/sty1950DOI
  35. [37]Chan, Y. M., & Dalgarno, A. 1965, Proceedings of the Physical Society, 85, 227, doi: 10.1088/0370-1328/85/2/304
  36. [38]Coles, P. A., Yurchenko, S. N., & Tennyson, J. 2019, MNRAS, 490, 4638, doi: 10.1093/mnras/stz2778DOI
  37. [39]Coulombe, L.-P., Benneke, B., Challener, R., et al. 2023, Nature, 620, 292, doi: 10.1038/s41586-023-06230-1DOI
  38. [40]Coulombe, L.-P., Parmentier, V., Stevenson, K. B., et al. 2026, arXiv e-prints, arXiv:2609.16115, doi: 10.48550/arXiv.2609.16115arxiv.org/abs/2609.16115
  39. [41]Dalgarno, A., & Williams, D. A. 1962, ApJ, 136, 690, doi: 10.1086/147428DOI
  40. [42]Drake, A. J., Beshore, E., Catelan, M., et al. 2010, arXiv e-prints, arXiv:1009.3048, doi: 10.48550/arXiv.1009.3048DOI
  41. [43]Dupuy, T. J., & Liu, M. C. 2017, ApJS, 231, 15, doi: 10.3847/1538-4365/aa5e4cDOI
  42. [44]Espinoza, N., Úbeda, L., Birkmann, S. M., et al. 2023, PASP, 135, 018002, doi: 10.1088/1538-3873/aca3d3DOI
  43. [45]Evans, T. M., Sing, D. K., Kataria, T., et al. 2017, Nature, 548, 58, doi: 10.1038/nature23266DOI
  44. [46]Farihi, J., & Christopher, M. 2004, AJ, 128, 1868, doi: 10.1086/423919DOI
  45. [47]Farihi, J., Zuckerman, B., & Becklin, E. E. 2005, AJ, 130, 2237, doi: 10.1086/491707DOI
  46. [48]Feroz, F., & Hobson, M. P. 2008, MNRAS, 384, 449, doi: 10.1111/j.1365-2966.2007.12353.xDOI
  47. [49]Feroz, F., Hobson, M. P., & Bridges, M. 2009, MNRAS, 398, 1601, doi: 10.1111/j.1365-2966.2009.14548.xDOI
  48. [50]Feroz, F., Hobson, M. P., Cameron, E., & Pettitt, A. N. 2019, The Open Journal of Astrophysics, 2, 10, doi: 10.21105/astro.1306.2144DOI
  49. [51]Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman, J. 2013, PASP, 125, 306, doi: 10.1086/670067
  50. [52]Fortney, J. J., Lodders, K., Marley, M. S., & Freedman, R. S. 2008, ApJ, 678, 1419, doi: 10.1086/528370DOI
  51. [53]French, J. R., Casewell, S. L., Amaro, R. C., et al. 2024, MNRAS, 534, 2244, doi: 10.1093/mnras/stae2121DOI
  52. [54]Fuda, N., & Apai, D. 2024, ApJL, 975, L32, doi: 10.3847/2041-8213/ad87e9DOI
  53. [55]Fuda, N., Apai, D., Nardiello, D., et al. 2024, ApJ, 965, 182, doi: 10.3847/1538-4357/ad2c84DOI
  54. [56]Gagné, J., Faherty, J. K., Ruiz Diaz, A., et al. 2026, arXiv e-prints, arXiv:2604.22012, doi: 10.48550/arXiv.2604.22012arxiv.org/abs/2604.22012
  55. [57]Gaia Collaboration, Vallenari, A., Brown, A. G. A., et al. 2023, A&A, 674, A1, doi: 10.1051/0004-6361/202243940
  56. [58]Gandhi, S., & Madhusudhan, N. 2019, MNRAS, 485, 5817, doi: 10.1093/mnras/stz751DOI
  57. [59]Gandhi, S., Madhusudhan, N., & Mandell, A. 2020, AJ, 159, 232, doi: 10.3847/1538-3881/ab845eDOI
  58. [60]Gillon, M., Triaud, A. H. M. J., Jehin, E., et al. 2013, A&A, 555, L5, doi: 10.1051/0004-6361/201321620DOI
  59. [61]Goody, R. M., & Yung, Y. L. 1989, Atmospheric radiation : theoretical basis
  60. [62]Grant, D., Lewis, N. K., Wakeford, H. R., et al. 2023, ApJL, 956, L32, doi: 10.3847/2041-8213/acfc3b10.3847/2041-8213/acfdab
  61. [63]Gray, D. F. 2008, The Observation and Analysis of Stellar Photospheres
  62. [64]Hargreaves, R. J., Gordon, I. E., Rey, M., et al. 2020, ApJS, 247, 55, doi: 10.3847/1538-4365/ab7a1aDOI
  63. [65]Heinke, L., Min, M., Bouwman, J., et al. 2026, A&A, 710, A305, doi: 10.1051/0004-6361/2022557839
  64. [66]Horne, K. 1986, PASP, 98, 609, doi: 10.1086/131801DOI
  65. [67]Hubeny, I., Burrows, A., & Sudarsky, D. 2003, ApJ, 594, 1011, doi: 10.1086/377080DOI
  66. [68]Izzard, R. G., Hall, P. D., Tauris, T. M., & Tout, C. A. 2012, in IAU Symposium, Vol. 283, Planetary Nebulae: An Eye to the Future, 95–102, doi: 10.1017/S1743921312010769
  67. [69]Jakobsen, P., Fertitt, P., Alves de Oliveira, C., et al. 2022, A&A, 661, A80, doi: 10.1051/0004-6361/202142663DOI
  68. [70]Jenkins, S., Vanderburg, A., Mang, J., et al. 2026, ApJ, 1006, 241, doi: 10.3847/1538-4357/ae83b5DOI
  69. [71]Karalidi, T., Apai, D., Marley, M. S., & Buenzli, E. 2016, ApJ, 825, 90, doi: 10.3847/0004-637X/825/2/90DOI
  70. [72]Kreidberg, L. 2015, PASP, 127, 1161, doi: 10.1086/683602DOI
  71. [73]Lee, E. K. H., Casewell, S. L., Chubb, K. L., et al. 2020, MNRAS, 496, 4674, doi: 10.1093/mnras/staa1882DOI
  72. [74]Lew, B. W. P., Apai, D., Zhou, Y., et al. 2016, ApJL, 829, L32, doi: 10.3847/2041-8205/829/2/L32DOI
  73. [75]Lew, B. W. P., Apai, D., Zhou, Y., et al. 2022, AJ, 163, 8, doi: 10.3847/1538-3881/ac3001DOI
  74. [77]Lines, S., Mayne, N. J., Manners, J., et al. 2019, MNRAS, 488, 1332, doi: 10.1093/mnras/stz1788DOI
  75. [78]Littlefair, S. P., Casewell, S. L., Parsons, S. G., et al. 2014, MNRAS, 445, 2106, doi: 10.1093/mnras/stu1895DOI
  76. [79]Lothringer, J. D., & Barman, T. 2019, ApJ, 876, 69, doi: 10.3847/1538-4357/ab1485DOI
  77. [80]Lothringer, J. D., Barman, T., & Koskinen, T. 2018, ApJ, 866, 27, doi: 10.3847/1538-4357/aadd9eDOI
  78. [81]Lothringer, J. D., & Casewell, S. L. 2020, ApJ, 905, 163, doi: 10.3847/1538-4357/abc5bcDOI
  79. [82]Lothringer, J. D., Zhou, Y., Apai, D., et al. 2024, ApJ, 968, 126, doi: 10.3847/1538-4357/ad43daDOI
  80. [83]Lothringer, J. D., Bennett, K. A., Sing, D. K., et al. 2025, AJ, 169, 274, doi: 10.3847/1538-3881/adcl17
  81. [84]Luri, X., Brown, A. G. A., Sarro, L. M., et al. 2018, A&A, 616, A9, doi: 10.1051/0004-6361/201832964DOI
  82. [85]Marley, M. S., Saumon, D., Visscher, C., et al. 2021, ApJ, 920, 85, doi: 10.3847/1538-4357/ac141dDOI
  83. [86]McCarthy, A. M., Vos, J. M., Muirhead, P. S., et al. 2025, ApJL, 981, L22, doi: 10.3847/2041-8213/ad9eafDOI
  84. [87]McKemmish, L. K., Masseron, T., Hoeijmakers, H. J., et al. 2019, MNRAS, 488, 2836, doi: 10.1093/mnras/stz1818DOI
  85. [88]Mendonça, J. M., Malik, M., Demory, B.-O., & Heng, K. 2018, AJ, 155, 150, doi: 10.3847/1538-3881/aaaeeb
  86. [89]Metchev, S. A., Heinze, A., Apai, D., et al. 2015, ApJ, 799, 154, doi: 10.1088/0004-637X/799/2/154DOI
  87. [90]Miles, B. E., Mang, J., Morley, C. V., et al. 2026, arXiv e-prints, arXiv:2609.20664, doi: 10.48550/arXiv.2609.20664DOI
  88. [91]Mollière, P., Wardenier, J. P., van Boekel, R., et al. 2019, A&A, 627, A67, doi: 10.1051/0004-6361/201935470DOI
  89. [92]Morley, C. V., Mukherjee, S., Marley, M. S., et al. 2024, ApJ, 975, 59, doi: 10.3847/1538-4357/ad71d5DOI
  90. [93]Mukherjee, S., Fortney, J. J., Carmichael, T. W., Davis, C. E., & Thorngren, D. P. 2026a, ApJ, 1000, 309, doi: 10.3847/1538-4357/ae3539DOI
  91. [94]Mukherjee, S., Sing, D. K., Fu, G., et al. 2026b, Science, 392, 858, doi: 10.1126/science.adx5903DOI
  92. [95]Nasedkin, E., Mollière, P., & Blain, D. 2024, The Journal of Open Source Software, 9, 5875, doi: 10.21105/joss.05875
  93. [96]Parmentier, V., Fortney, J. J., Showman, A. P., Morley, C., & Marley, M. S. 2016, ApJ, 828, 22, doi: 10.3847/0004-637X/828/1/22DOI
  94. [97]Parmentier, V., Guillot, T., Fortney, J. J., & Marley, M. S. 2015, A&A, 574, A35, doi: 10.1051/0004-6361/201323127DOI
  95. [98]Parmentier, V., Showman, A. P., & Fortney, J. J. 2021, MNRAS, 501, 78, doi: 10.1093/mnras/staa3418DOI
  96. [99]Parmentier, V., Showman, A. P., & Lian, Y. 2013, A&A, 558, A91, doi: 10.1051/0004-6361/201321132DOI
  97. [100]Parmentier, V., Line, M. R., Bean, J. L., et al. 2018, A&A, 617, A110, doi: 10.1051/0004-6361/201833059DOI
  98. [101]Parmentier, V., Stevenson, K. B., Welbanks, L., et al. 2026, Nature Astronomy, 10, 1011, doi: 10.1038/s41550-026-02845-2DOI
  99. [102]Peters, P. C. 1964, Physical Review, 136, 1224, doi: 10.1103/PhysRev.136.B1224DOI
  100. [104]Rasio, F. A., Tout, C. A., Lubow, S. H., & Livio, M. 1996, ApJ, 470, 1187, doi: 10.1086/177941DOI
  101. [105]Sainsbury-Martinez, F., Casewell, S. L., Lothringer, J. D., Phillips, M. W., & Tremblin, P. 2021, A&A, 656, A128, doi: 10.1051/0004-6361/202141637
  102. [106]Sainsbury-Martinez, F., & Tremblin, P. 2024, A&A, 691, A232, doi: 10.1051/0004-6361/202451603
  103. [107]Saumon, D., Chabrier, G., & van Horn, H. M. 1995, ApJS, 99, 713, doi: 10.1086/192204DOI
  104. [108]Sheppard, K. B., Mandell, A. M., Tamburo, P., et al. 2017, ApJL, 850, L32, doi: 10.3847/2041-8213/aa9ae9DOI
  105. [109]Showman, A. P. 2016, Nature, 533, 330, doi: 10.1038/533330aDOI
  106. [110]Showman, A. P., Tan, X., & Parmentier, V. 2020, SSRv, 216, 139, doi: 10.1007/s11214-020-00758-8DOI
  107. [111]Skilling, J. 2004, in American Institute of Physics Conference Series, Vol. 735, Bayesian Inference and Maximum Entropy Methods in Science and Engineering: 24th International Workshop on Bayesian Inference and Maximum Entropy Methods in Science and Engineering, ed. R. Fischer, R. Preuss, & U. V. Toussaint (AIP), 395–405, doi: 10.1063/1.1835238DOI
  108. [112]Smith, K. W., Smartt, S. J., Young, D. R., et al. 2020, PASP, 132, 085002, doi: 10.1088/1538-3873/ab936eDOI
  109. [113]Sousa-Silva, C., Al-Refaie, A. F., Tennyson, J., & Yurchenko, S. N. 2015, MNRAS, 446, 2337, doi: 10.1093/mnras/stu2246
  110. [114]Steinrueck, M. E., Koskinen, T., Lavvas, P., et al. 2023, ApJ, 951, 117, doi: 10.3847/1538-4357/acd4bbDOI
  111. [115]Stephens, D. C., Leggett, S. K., Cushing, M. C., et al. 2009, ApJ, 702, 154, doi: 10.1088/0004-637X/702/1/154DOI
  112. [116]Stolker, T., Quanz, S. P., Todorov, K. O., et al. 2020, A&A, 635, A182, doi: 10.1051/0004-6361/201937159DOI
  113. [117]Tan, X., & Showman, A. P. 2020, ApJ, 902, 27, doi: 10.3847/1538-4357/abb3d4DOI
  114. [118]Tian, F. 2009, ApJ, 703, 905, doi: 10.1088/0004-637X/703/1/905DOI
  115. [119]Tremblay, P.-E., Ludwig, H.-G., Steffen, M., & Freytag, B. 2013, A&A, 559, A104, doi: 10.1051/0004-6361/201322318DOI
  116. [120]Trotta, R. 2008, Contemporary Physics, 49, 71, doi: 10.1080/0010751080266753
  117. [123]Vincent, O., Barstow, M. A., Jordan, S., et al. 2024, A&A, 682, A5, doi: 10.1051/0004-6361/202347694DOI
  118. [124]Vos, J. M., Burningham, B., Faherty, J. K., et al. 2023, ApJ, 944, 138, doi: 10.3847/1538-4357/acab58DOI
  119. [125]Weinberg, N. N., Sun, M., Arras, P., & Essick, R. 2017, ApJL, 849, L11, doi: 10.3847/2041-8213/aa9113DOI
  120. [126]Wende, S., Reiners, A., Seifahrt, A., & Bernath, P. F. 2010, A&A, 523, A58, doi: 10.1051/0004-6361/201015220DOI
  121. [127]Wilson, R. F., Barclay, T., Powell, B. P., et al. 2023, ApJS, 269, 5, doi: 10.3847/1538-4365/acf3dfDOI
  122. [128]Yan, F., Pallé, E., Reiners, A., et al. 2020, A&A, 640, L5, doi: 10.1051/0004-6361/202038294DOI
  123. [129]Yang, H., Apai, D., Marley, M. S., et al. 2016, ApJ, 826, 8, doi: 10.3847/0004-637X/826/1/8DOI
  124. [130]Yates, A., Parsons, S. G., Brown, A. J., et al. 2026, MNRAS, 547, stag358, doi: 10.1093/mnras/stag358DOI
  125. [131]Yurchenko, S. N., Mellor, T. M., Freedman, R. S., & Tennyson, J. 2020, MNRAS, 496, 5282, doi: 10.1093/mnras/staa1874DOI
  126. [132]Yurchenko, S. N., Tennyson, J., Syme, A.-M., et al. 2022, MNRAS, 510, 903, doi: 10.1093/mnras/stab3267DOI
  127. [133]Zahnle, K., Marley, M. S., Freedman, R. S., Lodders, K., & Fortney, J. J. 2009, ApJL, 701, L20, doi: 10.1088/0004-637X/701/1/L20DOI
  128. [134]Zahnle, K. J., & Marley, M. S. 2014, ApJ, 797, 41, doi: 10.1088/0004-637X/797/1/41DOI
  129. [135]Zhang, Z., Mollière, P., Hawkins, K., et al. 2023, AJ, 166, 198, doi: 10.3847/1538-3881/acf768DOI
  130. [136]Zhou, Y., Bowler, B. P., Apai, D., et al. 2022, AJ, 164, 239, doi: 10.3847/1538-3881/ac9905DOI

Paper details

vlnc:W_2pyd3jvt
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

AI tools

View statement
Looking for related papers…
Contents