The JWST Early-release Science Program for Direct Observations of Exoplanetary Systems II: A 1 to 20 μm Spectrum of the Planetary-mass Companion VHS 1256–1257 b

Version 1.7.2 of the standard JWST pipeline was used to reduce each science observation dither into spectral cubes for the NIRSpec IFU data set. The CRDS versions and context used by the JWST pipeline were "11.16.12" and "jwst_0977.pmap", respectively. At the time of the initial analysis, the NIRSpec pipeline used instrument parameters determined from ground-based testing (Rigby et al. 2022). Residual bad pixels and cosmic rays remain in the standard pipeline reductions, and these issues are compounded when dithers are reduced together. Each dither was processed through Stage 1, which performs detector-level corrections and converts detector ramps into slope images. After Stage 1, the Stage 2 step removes other instrument artifacts and creates calibrated slope images. Stage 3 of the pipeline takes slope images to create three-dimensional spectral cubes of a target. Total errors are propagated through all stages of the JWST Pipeline starting with variances estimated in the slope-fitting step in Stage 1. Stage 3 spectral cubes have an associated error array with the same dimensions of the spectral cube.

The cube-building step in Stage 3 was run with outlier detection turned off and all parameters set to default values to produce three-dimensional spectral cubes that are aligned in right ascension and declination. The Stage 3 spectral cubes were used to create one-dimensional spectra using aperture extraction. First, the spectral cube for each dither is collapsed along the wavelength axis to calculate a mean image to mitigate the effects of cosmic rays, then a 2D Gaussian is fit to the source point-spread function (PSF) to find the source's center position. The center of the source's position changes on a subpixel level in both the x and y directions of the image cube for all NIRSpec observation modes. The source center oscillates as a function of wavelength by 0.2–0.3 pixels in the x direction and 0.15–0.2 pixels in the y direction. For each image along the spectral cube wavelength axis, we then extract the flux within a circular aperture centered on this position. We tested several extraction radii to find where the object's flux plateaus with extraction radius. Based on this, we adopted extraction radii of 3–4 pixels (1.2–1.7 FWHM) for VHS 1256 b, while the calibrator star required a radius of 4 pixels across all bands. Aperture correction was not applied to the NIRSpec IFU data and it is not relevant for a well-centered point source. The same aperture used for each image is applied to the error array for every wavelength to calculate the error with standard error propagation. The weighted mean of the dithers is taken to produce a final spectrum.

Before a final weighted average is calculated, a hard cutoff is applied to fluxes that are 10% higher than the spectral energy distribution of the initial median to remove wavelengths affected by cosmic rays. Edges of the reconstructed IFU data cube include spatial elements that are not fully illuminated, and the extracted spectra derived from these parts of the data cube are removed from the final spectrum. After the cutoffs are applied, we take the weighted mean of the dithers to obtain the final spectrum for a filter/grating combination. The error of the spectrum is the propagated error of the weighted average.

As discussed previously, the flux calibration of the spectral cubes is not optimized due to the use of ground-based calibration files in the standard JWST pipeline. To produce a final flux-calibrated spectrum for VHS 1256 b, the extracted spectra must be multiplied by a scale factor, which is the ratio of a calibrated standard spectrum and the response it produces after processing with the JWST pipeline. The A3V star TYC 4433-1800-1, observed during commissioning (Program ID: 1128), has a calibrated flux reference in the CALSPEC Library (Bohlin et al. 2020) and was used as a calibrator to adjust the spectral response of the VHS 1256 b extracted spectrum.

All observation modes used in the ERS program were available for the calibrator star, and we reduced these observations with the same JWST pipeline parameters as the science spectra. As before, each dither was reduced individually, and the extracted spectra of each filter/grating mode were combined with a weighted mean to produce a final spectrum for that mode. There are oscillating features present in both the calibrator star and the VHS 1256 b NIRSpec spectra that are dependent on dither position and brightness. These features are averaged out in the calibrator star by fitting a fourth-order (∼T⁴) polynomial to obtain the "mean" response at each wavelength. The intrinsic spectral shape of the planetary-mass companion is not assumed, and the oscillations are not removed by fitting. The oscillating features change the amplitude of the extracted science spectra by about 2%–6%; however, in the extracted calibrator spectra these amplitudes can vary by as much as 20%, making the fitting step much more important for the calibrator star. In the VHS 1256 b–extracted spectra, the oscillations are visually apparent in dithers from the G140H/F100LP detector 1, G140H/F100LP detector 2, G140H/F100LP detector 1, and G140H/F100LP detector 2 observations. Averaging or taking the mean of several dithers helps to remove some of these oscillating features, but there are not enough dithers to remove the effect entirely. The calibrator star has oscillations in all observation modes for our program. Portions of the final VHS 1256 b spectra that appear to be affected by these oscillations are 1.65–1.75 μm and 2–2.4 μm. We will discuss this in the context of identified molecular features in Section 4. Generally speaking, the crest-to-crest width of the oscillations appearing in the calibrator star extracted spectra vary in width, but are never smaller than .01 μm and can be as large as .029 μm.

The standard spectrum of our calibrator star possesses absorption features that are masked before fitting a fourth-order polynomial to the spectra. We perform each fit for a single filter/grating mode. The best fit of the calibrator spectrum and the best fit of the extracted pipeline spectra are divided to find the scale factor, which is then applied to the VHS 1256 b spectrum that falls within that bandpass. The extracted errors from the calibrator spectra are also included with VHS 1256 b's spectrum errors when the scale factor is applied.

Version 1.8.1 of the standard JWST pipeline was used to reduce the MIRI MRS observation listed in Table 1. The CRDS versions and context used by the JWST pipeline were "11.16.14" and "jwst_1007.pmap." Each dithering sequence within a wavelength bandpass was combined to create a single spectral cube. JWST pipeline stages 1, 2, and 3 were all used to reduce the MIRI MRS data. The MIRI MRS dithers were reduced together for each channel/grating combination, and one-dimensional spectra were extracted from the resulting spectral cubes using aperture photometry.

We chose not to use the background subtraction method implemented in the standard pipeline, as it was unable to account for cosmic-ray showers present in the data. These cosmic-ray showers appear as diffuse, extended structures that do not produce the usual jumps in the detector ramps, which the pipeline is able to detect. The distribution and shape of the cosmic-ray showers vary between science and background exposures, and the impact of these differences produces the main systematic noise source for the faint point source extracted with MIRI. We instead estimated the background using a reference aperture placed off of the target in our calibrated science cubes. The separation of the background apertures varied between channels, with 17 for Channel 1, 22 for Channel 2, and 23 for Channels 3 and 4.

Before performing aperture photometry, as with NIRSpec, each spectral cube is collapsed in the wavelength direction using the mean, and the source center is fit with a 2D Gaussian. This is a necessary step due to slight band-to-band offsets still present in the distortion model of the MRS. We then extracted a flux within a circular aperture centered on our derived source position for each image along the spectral cube wavelength axis. For a given wavelength-dependent extraction radius (2.0 FWHM), the aperture correction is applied to account for the flux of the PSF missed outside of the extraction radius. This correction factor was derived using webbPSF (Perrin et al. 2014) models for the MRS, matching the empirical PSF FWHM measured during commissioning.

After extraction and aperture correction, fringes remain in portions of the MIRI spectrum. Fringing appears as a regular, beating interference pattern in the MIRI spectra (Argyriou et al. 2020), particularly in channels 2C, 3A, 3B, and 3C. The ResidualFringeStep is applied to the one-dimensional spectra to find and remove fringing patterns from the data (P. Kavanagh et al. 2023, in preparation). ⁸¹

Past 18 μm, the sensitivity of the MRS drops significantly due to a combination of the low efficiency of the grating, low quantum efficiency of the detectors, and rising thermal background (Glasse et al. 2015). As a result, the spectrum could not reliably be extracted in channel 4. Instead, the cube was collapsed from 17.71 to 20.94 μm (Channel 4A) to produce a single photometric point. The source center is found by fitting a 2D Gaussian to the collapsed channel 4A cube. A circular aperture (radius of 1.5 FWHM) is placed at the measured PSF center, and the same aperture correction described above is used. We measure the background using an off-target reference aperture with the same radius and subtract the background.

The error in the extracted spectrum was estimated using the propagated error by the jwst pipeline Extract1dStep. The output x1d.fits files contain an ERR array extension that captures the error of the full processing chain, including the photon noise of the source. The extracted spectrum of the pipeline uses an annulus subtraction that introduces systematic errors from the background subtraction, but the overall spectrum matches the flux and shape of our own extraction. We therefore assume that the error reported by the pipeline should be representative of our own extraction. Indeed, the slope of the error follows the expected rising thermal background but seems to be overall a factor of ten lower than what one would expect from the ETC. Additionally, comparing with the noise from a collection of reference apertures around the field of view confirms this discrepancy. We therefore choose to artificially amplify the pipeline estimated error by a factor of 10, yielding a rather conservative estimate of the signal-to-noise ratio(S/N) until the pipeline issue is resolved.

The full reduced spectrum with labeled bandpasses is shown in Figure 2. The NIRSpec (1–5 μm) spectra have an S/N of 50–400 per pixel. The MIRI (4.9–20.94 μm) spectra have an S/N of 7 to 20 per pixel, with hardly any signal in Channel 4 (17.66–28.1 μm). The photometry point from the channel has an S/N of 2.7.

Zoom In Zoom Out Reset image size

There are 557 data points from NIRSpec and 370 data points from MIRI that overlap in wavelength. When the NIRSpec data are binned down and interpolated onto MIRI's wavelength spacing, some portions of the NIRSpec data have a higher baseline flux, more than 3σ away from the MIRI data points. Figure 3 shows the overlap region of NIRSpec IFU and MIRI MRS data at the resolution of each respective instrument.

Zoom In Zoom Out Reset image size

Late M-dwarf stars and L spectral-type brown dwarfs display absorption features from neutral gases such as vanadium oxide (VO), iron hydride (FeH), potassium (K), titanium oxide (TiO), and sodium (Na). These absorption features can be used to infer surface gravity. Broader and deeper absorption features from these molecules and atoms correspond to higher surface gravity and older ages (McGovern et al. 2004; Allers & Liu 2013). The 1 to 1.35 μm portion of the JWST/NIRSpec spectrum (top, Figure 4) holds two K doublets and a Na line that indicate a relatively low surface gravity, which is consistent with VHS 1256 b's placement on a color–magnitude diagram. Figure 6 shows a comparison of VHS 1256 b's Na and K lines with those of field brown dwarfs. Both absorption lines appear narrower than the same lines in a similar spectral-type field brown dwarf, indicating a low surface gravity for VHS 1256 b. The first published near-infrared spectrum of VHS 1256 b in Gauza et al. (2015) showed no K doublet features and no detection of the 1.134 μm Na line, possibly due to insufficient resolution. Follow-up work by Petrus et al. (2023) obtained medium-resolution (R ∼ 8000) spectra of VHS 1256 b at a resolution higher than the JWST spectrum presented here. The Na line at 1.134 μm and K doublets at 1.173 and 1.248 μm are easier to distinguish from the continuum in the JWST spectrum; however, the published spectrum from Petrus et al. (2023) likely has the resolution to capture a broader range of absorption features such as Fe, FeH, and TiO. However, the standard NIRSpec pipeline procedures are still in development. Once these parameters have been refined, future reductions of the JWST spectrum may lead to the detection of finer features with higher S/Ns.

Zoom In Zoom Out Reset image size

All spectral types of brown dwarfs possess absorption features due to water vapor in the near and mid-infrared. The shape and depth of these features are primarily determined by the effective temperature or spectral type of a brown dwarf (Allers & Liu 2013). Previously published space-based spectroscopic infrared observations of VHS 1256 b and other brown dwarfs are sensitive to water but often limited to resolutions of a few hundred (Zhou et al. 2020). The sensitivity and added resolution of the JWST data provide crucial details regarding the presence of water vapor in the atmospheres of brown dwarfs. Water absorption bands are present in the VHS 1256 b spectrum at 1.3–1.6 and 1.7–2.1 μm, and shape the spectrum beyond 10 μm. Water overlaps with carbon monoxide (CO) between 2.2 and 2.6 μm and between 4.3 and 5 μm. All water features are labeled in Figures 4 and 5.

Brown dwarfs below an effective temperature of ∼1400 K begin to display methane absorption features, as methane is favored in the chemical reaction between methane and carbon monoxide at these temperatures (Lodders & Fegley 2002). Methane absorption at 1.67 μm is also a typical T-dwarf signature (Cushing et al. 2005).

We have an absorption feature in the spectrum at 1.66 μm in VHS 1256 b; however, this feature is slightly blueward of the typical 1.67 μm location, but also too broad to be FeH absorption that is seen in warmer L dwarfs (Cushing et al. 2005). This feature also coincides with excess amplitudes from oscillations in the extracted dithers at wavelengths between 1.65 and 1.75 μm. Only dithers 2 and 3 of the total four show no oscillations in the extracted spectra, but the slight depression remains when only using dithers 2 and 3 of the observations. The crest-to-crest width of the oscillation waves in the calibrator star between 1.65 and 1.75 is ∼0.01–0.02 μm, and the width of the potential methane feature is 0.02 μm. Previously published near-infrared spectra show no methane absorption at 1.67 μm (Gauza et al. 2015; Petrus et al. 2023). If this feature remains after further pipeline updates, the opacity source causing this feature will need to be validated with further atmospheric modeling work.

We also find a relatively shallow methane feature between 2.8 and 3.8 μm that is consistent with the previously published L-band spectrum in Miles et al. (2018). Methane absorption at ∼3.3 μm is more prominent in brown dwarfs than methane absorption at 1.65 μm and is seen even in mid-L dwarfs (Noll et al. 2000). In this effective temperature range, methane has a significant opacity contribution between 7 and 9 μm, but its overall strength compared to water and other molecules is smaller. This region does not visually mirror the predicted opacity profile of methane for this reason.

All of the near-infrared and mid-infrared methane features appear depleted relative to similar-temperature brown dwarfs, indicating that disequilibrium chemistry is influencing the apparent abundance of methane in the upper atmosphere. We will discuss the degree of atmospheric mixing required to recreate these methane features with models in Section 5.

Carbon monoxide (CO) is a common near-infrared spectral feature in very low-mass stars and brown dwarfs with effective temperatures above ∼1400 K (Lodders & Fegley 2002; Cushing et al. 2005). Carbon monoxide produces a strong feature centered at 4.6 μm in warm brown dwarfs. Cooler brown dwarfs that have disequilibrium chemistry driven by atmospheric mixing (Sorahana & Yamamura 2012) also display this feature. VHS 1256 b's effective temperature is cool enough that a small degree of carbon monoxide will be detected at 2.3 μm. The depth and shape of the central feature are similar to previously published K-band spectra in Hoch et al. (2022) and Petrus et al. (2023). We also detect a densely packed set of carbon monoxide features around 4.6 μm in VHS 1256 b's spectrum. It is the best-resolved, highest-S/N set of features from the fundamental carbon monoxide bandhead compared to previously published space-based brown dwarf spectra and photometry from AKARI (Sorahana & Yamamura 2012) and Spitzer (Patten et al. 2006).

Carbon dioxide (CO₂) is another carbon- and oxygen-bearing gas that can provide insight into metallicity or atmospheric effective temperature (Lodders & Fegley 2002; Tsuji et al. 2011; Sorahana & Yamamura 2012). The CO₂ feature at 4.2 μm is inaccessible from ground-based observatories and has only been detected in a handful of mid-infrared observations of T-type brown dwarfs (Tsuji et al. 2011; Sorahana & Yamamura 2012). We show evidence for this CO₂ feature in the JWST spectrum of VHS 1256 b by comparing two versions of an atmospheric model, one with CO₂ opacities added and one without, as shown in Figure 7. The model comparisons reveal a slight absorption feature where CO₂ influences the spectrum from 4.2 to 4.4 μm. In M dwarfs and relatively warm L dwarfs, CO₂ is theorized to primarily trend with effective temperature, while at lower effective temperatures where methane should be the dominant carbon-bearing gas, CO₂ will vary with both atmospheric pressure and temperature (Lodders & Fegley 2002). The presence of CO₂ in VHS 1256 b's spectrum would not be not surprising, but it needs to be verified with more detailed atmospheric analysis.

Zoom In Zoom Out Reset image size

The transition between red to blue near-infrared colors at the L-to-T transition shown in Figure 1 is likely driven by the condensation and eventual descent below the photosphere of silicate grains composed of enstatite, forsterite, or quartz as these objects cool. Low-surface-gravity brown dwarfs and directly imaged exoplanets can retain their silicate clouds at lower effective temperatures, producing their redder colors compared to field counterparts. VHS 1256 b displays a significant silicate cloud feature in the JWST/MIRI spectrum from 8 to 11 μm when compared to a standard brown dwarf and the relatively red brown dwarf 2MASSW J2224438–015852 (2M2224–0158) discovered in Kirkpatrick et al. (2000), which is an outlier along the L-to-T transition (Figure 11). Compared to 2M2224–0158, VHS 1256 b shares the same spectral shape across 8–11 μm, indicating an absorption feature due to silicate clouds of similar composition. The best-fit cloud model for 2MASS 2224–0158 from Burningham et al. (2021) was a combination of enstatite (MgSiO₃), quartz (SiO₂), and a higher-pressure iron (Fe) cloud.

VHS 1256 b's spectrum shows evidence of disequilibrium chemistry based on methane and carbon monoxide; therefore, we explored the potential presence of other disequilibrium molecules.

From the equilibrium Sonora Bobcat models (Marley et al. 2021), we scaled different molecular abundances to search for species that could appear in disequilibrium conditions. In VHS 1256 b's spectrum, there are no obvious signs of acetylene (C₂H₂), ethylene (C₂H₄), ethane (C₂H₆), hydrogen sulfide (H₂S), phosphine (PH₃), or hydrogen cyanide (HCN). Outside of the features listed prior to this subsection, more detailed analysis and approaches such as retrievals and cross-correlation will need to be applied to the JWST VHS 1256 b spectra to fully understand the object's atmospheric chemistry.

Integrating over VHS 1256 b's JWST spectrum and using the best-fit model described in the previous section to estimate flux from wavelength ranges not covered by our JWST spectrum, we find a bolometric luminosity of $\mathrm{log}\left(\tfrac{{L}_{\mathrm{bol}}}{{L}_{\odot }}\right)=-4.55\pm 0.009$, adopting the Gaia Early Data Release 3 distance measurement to the system of 21.15 ± 0.22 pc (Gaia Collaboration et al. 2021). Our JWST spectrum covers most of the full luminous range of VHS 1256 b; 98% of the measured luminosity is derived directly from the spectrum, with only 2% extrapolated from model fits to wavelengths outside of our JWST wavelength coverage. So much of the SED is covered by measurements and so many independent bands are included that the statistical/random error on our bolometric luminosity value is very small. With a precise distance measurement from Gaia Collaboration et al. (2021) for this system, the most significant uncertainty will stem from the absolute flux calibration of the spectrum using the A3V star TYC 4433-1800-1 (see Section 3). Over the course of JWST's operation, a dedicated absolute flux calibration program is planned to enable better than 2% flux calibration over a wide range of objects (Gordon et al. 2022). Thus, to set the error on our bolometric luminosity measurement at the current early stage of the JWST mission, we estimate a conservative error on the absolute flux calibration of 3% and add this in quadrature with the error on the integrated spectrum.

Adopting the Gaia DR3 distance for the system, a spectral type of L7 ± 1.5 and the VISTA Hemisphere survey K_S photometry from Gauza et al. (2015), and a bolometric correction BC_Ks based on the polynomial relationship for young M- to T-type objects from Filippazzo et al. (2015), we find a photometric bolometric luminosity of $\mathrm{log}\left(\tfrac{{L}_{\mathrm{bol}}}{{L}_{\odot }}\right)=-4.60\pm 0.05$, in good agreement with the value derived from our JWST spectroscopy. However, recent bolometric luminosity estimates from ground-based, near-IR spectroscopy for VHS 1256 b are slightly fainter than the value we find with JWST: Hoch et al. (2022) find a best value of $\mathrm{log}\left(\tfrac{{L}_{\mathrm{bol}}}{{L}_{\odot }}\right)=-4.67$, within the range from −4.6 to −4.7, while Petrus et al. (2023) find $\mathrm{log}\left(\tfrac{{L}_{\mathrm{bol}}}{{L}_{\odot }}\right)=-4.67\pm 0.07$. The bolometric luminosity estimate from Petrus et al. (2023) adopts a distance of ${22.2}_{-1.2}^{+1.1}$ pc from Dupuy et al. (2020), rather than the more accurate Gaia eDR3 measurement. Adjusting to the Gaia distance measurement leads to a slightly smaller bolometric luminosity, further increasing the tension between the JWST measurement and ground-based spectroscopic measurements.

Dupuy et al. (2022) determine an updated age for the system of 140 ± 20 Myr from dynamical masses derived from orbital fits to the inner AB binary. VHS 1256 b sits close to the deuterium-burning limit—if VHS 1256 b is slightly above the deuterium-burning limit, it will be actively burning deuterium at this age. For the short duration of its deuterium-burning phase, a lower-mass object may be more luminous than a higher-mass object at the same age; see Figure 10. As a given luminosity can correspond to a range of masses in this case, direct interpolation from this model grid can produce erroneous mass estimates.

Zoom In Zoom Out Reset image size

To robustly estimate the mass of VHS 1256 b, we implemented a rejection-sampling method similar to that described in Dupuy et al. (2022). First, we draw 1 × 10⁶ samples of age and mass from a Gaussian distribution in age around 140 Myr, with σ = 20 Myr, and a uniform distribution in mass from 1 to 50 M_Jup. For each age and mass sample, we then interpolate a model luminosity from an evolutionary model grid and calculate χ² as ${\chi }^{2}=\tfrac{{({L}_{\mathrm{bol},\mathrm{model}}-{L}_{\mathrm{bol},\mathrm{measured}})}^{2}}{{\sigma }_{\mathrm{Lbol},\mathrm{measured}}^{2}}$, then convert to a probability (P) by normalizing by the minimum χ² ($P={e}^{-\tfrac{{\chi }^{2}-{\chi }_{\min }^{2}}{2}}$) value among our 1 × 10⁶ samples. For each sample, we also draw a uniformly distributed number from 0 to 1. We retain the samples where the sample probability is greater than the uniformly distributed variate drawn for that sample.

As VHS 1256 b has strong evidence for the presence of clouds in its spectrum from the detection of the silicate feature at 10 μm, we implemented this rejection-sampling procedure using the hybrid cloud grid of Saumon & Marley (2008), which includes clouds for objects with L-type spectra and clear atmospheres for objects with T-type spectra.

A histogram of the final set of accepted masses is shown in Figure 10. We find a bimodal distribution of masses, with accepted samples both above and below the deuterium mass-burning limit, as also seen in Dupuy et al. (2022). The percentage of samples that fall into each of the two peaks depends strongly on both the bolometric luminosity value and the uncertainty on that value. While we cannot conclusively determine whether VHS 1256 b falls above or below the deuterium mass-burning limit, all accepted samples for this model have masses <20 M_Jup.

Young, planetary-mass objects with mid-to-late-L spectral types are highly variable in the near-IR (Biller et al. 2015; Lew et al. 2016; Biller et al. 2018; Zhou et al. 2020; Bowler et al. 2020), with variability amplitudes >5% over observations a few hours in length. VHS 1256 b is the most variable of this cohort. Over a contiguous six-orbit observation with HST/WFC3 obtained in 2018, Bowler et al. (2020) found that VHS 1256 b varied by >20% between 1.1 and 1.7 μm over 8 hr. In the near-IR, Bowler et al. (2020) fit the observed HST trend with a single sinusoidal model; however, in an additional 15 orbit/42 hr HST/WFC3 observation in 2020 presented by Zhou et al. (submitted), the variability observed is more complex, requiring a three-sinusoid + slope model for a full fit. In the mid-IR, Zhou et al. (2020) obtained a contiguous 36 hr Spitzer 4.5 μm lightcurve of VHS 1256 b and found a best-fit model for this lightcurve of a single sinusoid with a period of 22.04 ± 0.05 hr (interpreted as the rotation period of the object) and a peak-to-peak amplitude of 5.76% ± 0.04%, a significantly lower variability amplitude when compared to the near-IR variability. Both the near-IR and mid-IR variability are likely driven by patchy thin and thick silicate clouds (Apai et al. 2013). Variability is an important probe of the inhomogeneity of the top-of-atmosphere structure of these objects and it is critical to understand the intrinsic variability of this target in order to interpret our JWST observations. To do so, we have conducted a multitelescope, multiepoch photometric variability monitoring campaign starting in 2022 February and continuing through July, directly covering the JWST observation epoch, which will be presented in B. A. Biller et al. (in preparation).

However, over the ∼4 hr timescale of our ERS observations, we expect a relatively low level of variability, based on estimates from the earlier HST and Spitzer observations. To estimate the range of potential variability we expect during the NIRSpec observation, we drew 10,000 sample 2 hr observations from the three-sinusoid model from Zhou et al. (2022) and estimated the intrinsic variability occurring during each simulated observation as the maximum minus the minimum value from the model during that time span. We used a very fine time sampling and did not realistically simulate noise or the actual cadence of our JWST observations. Thus, this method constrains the intrinsic variability, and the actual measurement of variability in any given observation would yield a smaller value than these predictions. From our simulated observations, 50% of samples varied by less than 1.5%, 75% of samples varied by less than 2.5%, and 95% of samples varied by less than 3.7%. For wavelengths >3 μm, drawing 10,000 sample 2 hr observations from the single-sinusoid model used to fit the Spitzer 4.5 μm lightcurve from Zhou et al. (2020), 50% of samples varied by less than 1.1%, 75% of samples varied by less than 1.5%, and 100% of samples varied by less than 1.6%. More conservatively, we estimate a maximum potential variability measurement of 5% over 2 hr for the JWST/NIRSpec 1–3 μm observations, based on the highest amplitude variability epoch, where VHS 1256 b displayed 20% variability over 8 hr (Bowler et al. 2020). Assuming the same scaling between near-IR and mid-IR lightcurves as found between the HST and Spitzer lightcurves of Bowler et al. (2020) and Zhou et al. (2020), we estimate a maximum potential variability measurement of 1.5% over 2 hr at wavelengths >3 μm. Variability has not been measured beyond 5 μm for any planetary-mass object, but assuming a continuing trend of decreasing variability amplitude with increasing wavelength, we expect variability at wavelengths >5 μm to be negligible.

The JWST spectrum has molecular features that are consistent with previous findings of VHS 1256 b and other young, red late-L dwarfs having atmospheres that are out of chemical equilibrium (Chauvin et al. 2004; Gauza et al. 2015; Liu et al. 2016; Miles et al. 2018). The presence of CO absorption and depleted CH₄ absorption compared to equilibrium atmospheric models supports atmospheric mixing forcing the atmosphere into chemical disequilibrium. The CH₄ absorption feature at 3.3 μm is the most prominent absorption feature stemming from disequilibrium chemistry in VHS 1256 b's spectrum and potentially other similar-temperature, directly imaged exoplanets. The fine-tuned forward model shown in Figure 8 has an associated CH₄ abundance profile (Figure 9), but the model does not match the 3.3 μm CH₄ feature or many other molecular features along the spectrum, despite matching the overall SED shape. For sedimentation parameter values of f_sed = 4, CH₄ features at 1.6 and 7 μm appear in model spectra, but for our best-fit model with f_sed < 1, these features are quite muted. The estimated K_zz range of VHS 1256 b spans 10⁸ cm² s⁻¹–10⁹ cm² s⁻¹, which is consistent with the estimated K_zz of 10⁸ cm² s⁻¹ from Miles et al. (2018). However, not matching crucial disequilibrium absorption features is significant and means future modeling work on clouds and mixing will need to be done to estimate a meaningful K_zz for VHS 1256 b.

We show evidence of CO₂ being useful for reproducing the shape of the JWST spectrum at 4.2 μm as shown in Figures 7 and 9. The portions of the JWST spectrum surrounding the CO₂ feature are discrepant from the best-fit atmospheric model. CO₂ is also influenced by disequilibrium chemistry, but the timescale of CO₂'s conversion from CH₄ is much faster than the conversion of CO to CH₄ (Zahnle & Marley 2014); therefore, CO₂ quenches at higher atmospheric pressures. Historically, disequilibrium chemistry driven by atmospheric mixing has only been detectable from a single molecule in exoplanets and brown dwarfs: either CH₄ for warmer L dwarfs or CO for cooler T dwarfs. JWST has the capability to confidently detect two different molecular gases with different quench pressures between 1 and 5 μm. JWST is capable of detecting CO₂ and CH₄ or CO over the entire L-, T-, and potentially Y-dwarf sequence, revolutionizing our understanding of disequilibrium chemistry as traced by carbon species in the atmospheres of brown dwarfs and extrasolar planets.

Brown dwarfs are expected to have clouds in their photospheres as their temperatures drop below the condensation curves of various species (Lunine et al. 1986, 1989; Tsuji et al. 1996; Chabrier et al. 2000; Ackerman & Marley 2001). There are several lines of evidence for clouds in brown dwarfs, including:

• 1.  
  The red colors of L-type brown dwarfs, which quickly transition to blue colors as the brown dwarfs cool through the L-to-T transition and clouds are theorized to sink below the visible atmosphere (Kirkpatrick et al. 1999; Allard et al. 2001; Saumon & Marley 2008).
• 2.  
  The variability seen in brown dwarfs (Artigau et al. 2009; Metchev et al. 2015), which is most prominent near the L-to-T transition and can be modeled with patchy clouds (Burgasser et al. 2002; Saumon & Marley 2008; Radigan et al. 2012; Apai et al. 2013).
• 3.  
  The match in temperature between condensation curves and kinks in brown dwarf color–magnitude diagrams (Fegley & Lodders 1994; Lodders 1999; Morley et al. 2012; Leggett et al. 2015).

There are also alternative explanations for the previous phenomena, such as temperature–pressure profiles that have been perturbed by disequilibrium chemistry (Tremblin et al.2015).

Solid-state spectroscopic features can be unambiguous signatures of clouds. At VHS 1256 b's temperature, the most prominent feature is expected to be from silicate particles, which have a broad 10 μm feature that is commonly seen in the interstellar medium and in the disks of young stars (Draine & Lee 1984). Using the Spitzer Infrared Spectrograph, Cushing et al. (2006) detected a "plateau"-shaped absorption feature from 9 to 11 μm in the spectra of several mid-L brown dwarfs, which they attributed to small amorphous silicate particles. A compendium of all of the Spitzer spectra of brown dwarfs shows that silicate features are common, but not ubiquitous, for L2–L8 brown dwarfs (Suárez & Metchev 2022).

The JWST/MIRI spectrum of VHS 1256 b shows a prominent silicate feature compared to the Spitzer brown dwarfs, with a shape that is well matched to 2M2224–0158 (see Figure 11). Burningham et al. (2021) modeled the spectrum of 2M2224–0158 and found the best fit with clouds composed of submicron enstatite/silicate, quartz, and iron. The presence of a prominent silicate feature in VHS 1256 b is strong evidence for small particles. Detailed modeling of the cloud composition and grain-size distribution will be the subject of a future paper.

Zoom In Zoom Out Reset image size

Various groups have developed models for clouds in brown dwarfs and exoplanets, but they tend to focus on fitting the near-infrared (1–2 μm) part of the spectrum, where extinction from large particles (10–100 μm) mutes the spectral features (Chabrier et al. 2000; Allard et al. 2001; Saumon & Marley 2008; Madhusudhan et al. 2011). It has long been known (e.g., from the comet literature), small particles contribute less to extinction (Gao et al. 2018), but they result in a much more prominent 8–12 μm silicate feature (Min et al. 2004). JWST's broad wavelength coverage allows us to constrain both populations of cloud particles.