GJ 1252b: A Hot Terrestrial Super-Earth with No Atmosphere

Subsequent to the mid-2019 TESS Sector 13 photometry used to first discover GJ 1252b (Shporer et al. 2020), the system was reobserved during the TESS Sector 27 Campaign using Camera 2 from 2020 July 5 to 2020 July 30. In this section, we describe our combined analysis of both the original Sector 13 and the new Sector 27 data. By performing a global fit on data sets separated by nearly a year, we further refine the orbital and planetary properties of GJ 1252b.

We downloaded both Sectors 13 and 27 Presearch Data Conditioning (PDC) time series measurements from MAST. PDC-level data products are corrected for instrumental systematics and contamination from nearby stars. Our analysis used the LightKurve package (Lightkurve Collaboration et al. 2018) to perform five iterations of 3σ outlier rejection on data points above the median flux level. This removed 0.2% of data from the lightcurve. To remove any remaining flux variations, we flattened the lightcurve using a Savitzky–Golay filter (Savitzky & Golay 1964) after first masking out the transits (with one transit duration on either side) before applying the filter to ensure the transit features are not affected.

We fit the flattened lightcurve using the exoplanet package (Foreman-Mackey et al. 2021), which uses a Hamiltonian Monte Carlo (HMC) routine to explore the posterior probability distribution. Assuming a circular orbit with the period (P), time of inferior conjunction (T₀), scaled planet radius (R_P /R_*), impact parameter (b), transit duration (T₁₄), and mean flux offset (μ) as free parameters, we minimized a negative log-likelihood function. Our analysis held quadratic limb-darkening coefficients constant at u₀ = 0.2800 and u₁ = 0.3683 (values taken from Claret 2017). We used the values obtained from the minimizer as initial positions for 32 parallel chains and ran the HMC for 2000 tuning steps and 4000 sampling steps per chain. Loose Gaussian priors (much wider than the final posteriors) were placed on P (mean 0.51 days, width σ 0.05 days) and T_conj (mean 1668.0, width 0.1) to prevent the sampler from wandering too far astray. The final Gelman–Rubin statistics of the HMC runs were <1.01 for all parameters.

The median values and their 1σ uncertainties are listed in Table 1, and the model fit to binned TESS data is shown in Figure 1. Our results agree well with those of the original discovery paper (Shporer et al. 2020). As an independent check, we also analyzed the full, two-sector TESS data set using the transit lightcurve code described in numerous similar K2 studies (e.g., Crossfield et al. 2015, 2016), and found consistent parameters in all cases.

Zoom In Zoom Out Reset image size

Table 1. Planet Parameters

Parameter	Units	Value	Source
Stellar Parameters:
R*	R⊙	0.391 ± 0.020	Shporer et al. (2020)
M*	M⊙	0.381 ± 0.019	Shporer et al. (2020)
Teff	K	3458 ± 137	Shporer et al. (2020)
age	Gyr	3.9 ± 0.4	This work, derived
TESS Transit Parameters:
T0	BJDTDB	2458668.09748 ± 0.00032	This work, fit
P	d	0.51824160 ± 0.00000069	This work, fit
i	deg	84.8 ± 3.2	This work, derived
RP /R*	⋯	0.0277 ± 0.0011	This work, fit
a/R*	⋯	5.03 ± 0.27	This work, derived
T14	hr	0.724 ± 0.013	This work, fit
b	⋯	0.42 ± 0.24	This work, fit
a	au	${0.00915}_{-0.00015}^{+0.00015}$	This work, derived
Sinc	S⊕	${233}_{-41}^{+48}$	This work, derived
RP	R⊕	1.180 ± 0.078	This work, derived
MP	M⊕	1.32 ± 0.28	Luque et al., in review
Spitzer Eclipse Parameters:
TE	BJDTDB	2458668.3575${}_{-0.0007}^{+0.0019}$	This work, fit
δIRAC2	ppm	${149}_{-32}^{+25}$	This work, fit
dt	min	+1.4${}_{-1.0}^{+2.8}$	This work, derived
$e\cos \omega$	⋯	+0.0025${}_{-0.0018}^{+0.0049}$	This work, derived
Tday	K	${1410}_{-125}^{+91}$	This work, derived
f	⋯	>0.40	This work, derived
AB	⋯	<0.41	This work, derived
$f\left(1-{A}_{B}\right)$	⋯	>0.37	This work, derived

Note. Two-sided intervals indicate 68.3% (1σ) confidence; one-sided intervals indicate 95.4% (2σ) confidence.

Download table as:  ASCIITypeset image

We analyze only the transits because the current TESS data cannot usefully constrain the planet's eclipse depth. In the TESS bandpass, the contribution from either thermal emission (Figure 6) or reflected light (Section 3.2.1) is ≲15 ppm, significantly smaller than our TESS transit depth precision of 61 ppm (Table 1).

Finally, we also performed a search for transit-timing variations (TTVs) in the TESS data, again using the exoplanet package. If present, TTV signals could indicate the presence of undiscovered companions due to mutual gravitational interactions or orbital decay due to tidal effects. Although the S/Ns of individual transit events are quite low, we find no evidence for TTVs with amplitudes ≳1 minute, consistent with a lack of strongly perturbing companions. The deviation of GJ 1252b's individually measured transit times is consistent with a linear ephemeris across both sectors of TESS data. With nearly a year separating these two sectors of TESS data, our analysis reduces the uncertainty on the period by an order of magnitude (see Table 1).

2.2.1. Eclipse Observations

2.2.2. Eclipse Analysis

We analyzed the Spitzer photometry using Pixel-Level Decorrelation (PLD; Deming et al. 2015), which models the systematics-dominated Spitzer lightcurve as a linear combination of basis vectors derived from each pixel's time series. Specifically, we use the formulation

$$\begin{eqnarray}&&{f}_{{ik}}=\left(\sum _{j=1}^{N+M}{c}_{{jk}}{v}_{{ijk}}\right){m}_{{ik}},\end{eqnarray} \tag{ 1 }$$

where f_ik is the modeled flux at the ith time step of the kth eclipse visit, c_jk is the scaling coefficient for the corresponding basis vector v_ijk , and m_ik is the purely astrophysical model of a secondary eclipse. The basis vectors v_ijk always include the N individual pixel time series from the kth visit (the "pixel-level" data essential to PLD) and may also include M low-order temporal trends (for which ${v}_{{ijk}}={t}_{{ijk}}^{p}$ for p = 0, 1,...(M − 1)) or other systematic vectors against which the data will be decorrelated. In our analysis, we included a linear trend with time in order to remove a slow, long-term drift. We parameterized the eclipse model (m_ik ) using the Mandel & Agol (2002) formulae for the occultation of an object with uniform surface brightness, with its only free parameters being the time of mid-eclipse T_E and the fractional eclipse depth δ.

PLD is often applied by simultaneously sampling the posterior distribution of the nuisance parameters c_jk as well as the astrophysical parameters of interest. In our case, this turned out to be intractable. With 10 eclipse visits, the use of N = 9 pixels and M = 1 (constant scaling, linear polynomial trend) would require marginalizing over 100 nuisance parameters to obtain measurements of just two astrophysical parameters, T_E and δ.

Instead, we determine T_E and δ, and their uncertainties, as follows. For each combination of these parameters, we calculate a model eclipse lightcurve and divide the observed flux F_ik by it. The result is

$$\begin{eqnarray}&&\displaystyle \frac{{F}_{{ik}}}{{m}_{{ik}}}\approx \sum _{j=1}^{N+M}{c}_{{jk}}{v}_{{ijk}}+{\epsilon }_{{ik}},\end{eqnarray} \tag{ 2 }$$

where _ik represent the measurement errors. We then directly solve Equation (2) at each point for the c_jk using weighted linear least squares. This approach allows us to directly sample the two-dimensional (T_E , δ) plane while also accounting for the interrelationships of these astrophysical parameters on the c_jk nuisance parameters. This approach is similar in some ways to the PLD analysis of Spitzer/IRAC microlensing observations (Dang et al. 2020); the main difference is that that work had more than two astrophysical parameters of interest and so used MCMC sampling instead of directly calculating a grid of likelihood values.

We set the weight of each observation equal to 1/σ², where σ is the 68.3% central confidence interval on the dispersion of the residuals to an initial fit. Flux measurements F_ik are set to zero weight if they deviate by more than 5σ from the nominal model.

The resulting stacked, detrended eclipse lightcurve shown in Figure 2 shows a clear flux decrement of ${149}_{-32}^{+25}$ ppm at the expected time of eclipse. Our measured eclipse parameters are listed in Table 1, while the lightcurves for each individual visit, as well as Allan deviation plots of the residuals to each visit, are shown in the Appendix. As two checks on our measured eclipse depth, we also calculated the weighted mean of the eclipse depths from each individual visit, and we conducted a joint analysis in which the time of eclipse was held fixed to orbital phase 0.5. The weighted mean is 150 ± 28 ppm, while the fixed-time analysis yields a depth of ${140}_{-20}^{+35}$ ppm; both these values are consistent with the value obtained from our joint fit.

Zoom In Zoom Out Reset image size

Finally, to interpret our measurements of GJ 1252b's thermal emission and estimate its atmospheric evolution (described below), we need to estimate a stellar age. Inferring ages for mature, field M dwarfs is notoriously challenging; one promising avenue for all but the latest M spectral subtypes is the use of gyrochronology. Using the spindown analysis of Engle & Guinan (2018) together with GJ 1252's stellar rotation period of 64 ± 4 days (Shporer et al. 2020), we estimate GJ 1252's age to be 3.9 ± 0.4 Gyr. This indicates that GJ 1252 is somewhat younger than LHS 3844 (7.8 ± 1.6 Gyr; Kane et al. 2020), an estimate broadly consistent with the nondetection of stellar flares during the first sector of TESS observations of GJ 1252 (Howard 2022). This result also indicates that the GJ 1252 system is somewhat younger than the 7.8 ± 1.6 Gyr LHS 3844b (Kane et al. 2020).

The offset of the eclipse time from orbital phase 0.5 constrains the combination of orbital parameters $e\cos \omega$ (Winn 2010); for GJ 1252b, we find $e\cos \omega$ = +0.0025 ${}_{-0.0018}^{+0.0049}$. This result is consistent with zero at the 1.4σ level, so we do not take this measurement as evidence of an eccentric orbit. Regardless, the $e\cos \omega$ measurement further justifies the assumption of low eccentricity in the radial velocity analysis of Shporer et al. (2020). We conducted a reanalysis of their radial velocity data while incorporating this new constraint on $e\cos \omega$, finding results consistent with those of the discovery paper.

Although GJ 1252b's orbit would quickly circularize in the absence of other perturbers, small planets orbiting M dwarfs are often found in multiplanet systems, and additional bodies in the system could cause GJ 1252b to stay on an eccentric orbit. Although no such bodies were indicated by our TTV analysis, it is still possible that tidal heating could act as an additional heat source in GJ 1252b. We estimate this heating level following the prescription of Henning et al. (2009) and assuming a Love number k₂ = 0.3 and tidal quality factor Q = 10⁴, approximately appropriate for super-Earths (Miguel et al. 2011; Millholland & Laughlin 2019). Further assuming that e = 0.0025, we find that tidal heating should contribute only 4 × 10¹⁶ W to GJ 1252b's total energy budget; this amount is negligible (unless tidal heating is enhanced via a significantly nonzero axial tilt; Millholland & Laughlin 2019) compared to the roughly 6 × 10¹⁹ W of starlight absorbed by the planet.

Here, we consider the implications of our eclipse measurement on the surface and atmospheric properties of GJ 1252b. We first consider our results in the context of global energy balance in Section 3.2.2, and then in the context of a suite of one-dimensional radiative transfer models in Section 3.2.3.

3.2.1. From Eclipse Depth to Brightness Temperature

3.2.2. Energy Balance and Atmospheric Circulation

One phenomenological framework for interpreting single-band secondary eclipse measurements is to assume the planet radiates as a blackbody at the measured brightness temperature, then to use the measurement to constrain some combination of day-to-night heat redistribution parameter f and Bond Albedo A_B (e.g., Seager 2010). In particular, the combination f(1 − A_B ) directly determines the planet's dayside equilibrium temperature via

$$\begin{eqnarray}&&{T}_{\mathrm{eq}}={T}_{\mathrm{eff}}\sqrt{\displaystyle \frac{{R}_{* }}{a}}{\left(f\left[1-{A}_{B}\right]\right)}^{1/4}.\end{eqnarray} \tag{ 3 }$$

In this formulation, the limiting values of f are $\tfrac{2}{3}$, indicating no heat redistribution (e.g., consistent with no atmosphere) and $\tfrac{1}{4}$, indicating uniform heat redistribution around the planet. Figure 3 shows our joint constraints on f and A_B , assuming flat priors on both quantities and the system parameters listed in Table 1. In all cases, the most likely values are the ones that give the highest dayside T_eq, i.e., $f=\tfrac{2}{3}$ and A_B = 0. We set upper limits (at 95.4%, or 2σ confidence) of a low albedo of A_B < 0.41; a high redistribution parameter of f > 0.40; and a high combination of the two, f(1 − A_B ) > 0.37.

Zoom In Zoom Out Reset image size

3.2.3. One-dimensional Atmosphere Models

Finally, we also present a large suite of atmospheric models and spectra of GJ 1252b. These models are all available as machine-readable supplements to this paper. Our models and spectra are generated using the open-source, 1D radiative transfer code HELIOS (Malik et al. 2017, 2019a, 2019b), which simulates the planet in radiative-convective equilibrium and also provides the functionality to include the radiative effects of a non-gray surface on both the atmosphere and the planetary spectrum (see E. A. Whittaker et al. 2022, in preparation).

In HELIOS, the temperature profile and surface temperature are obtained using the k-distribution method, with 420 wavelength bins (0.245–10⁵ μm). Then, starting from the equilibrium temperature profile, the planetary spectrum is calculated using opacity sampling with a resolution of R = 4000. Convectively unstable atmospheric layers are corrected using convective adjustment. We model dayside-averaged conditions and use the scaling theory of Koll (2022) to estimate the amount of heat transported from the dayside to the nightside of the planet. In the bare-rock case, the heat redistribution parameter (f in Equation (3)) is set to 2/3, equivalent to no horizontal heat transport (Burrows et al. 2008; Hansen 2008).

Gaseous opacities are calculated with HELIOS-K (Grimm & Heng 2015; Grimm et al. 2021), including O₂ (Gordon et al. 2017), N₂ (Gordon et al. 2022), H₂O (Polyansky et al. 2018), CO (Li et al. 2015), CO₂ (Rothman et al. 2010), CH₄ (Yurchenko et al. 2017), and SO₂ (Underwood et al. 2016). All opacities are calculated on a fixed grid with a resolution of 0.01 cm⁻¹, assuming a Voigt profile truncated at 100 cm⁻¹ from line center. For H₂O, CH₄, and SO₂ the default pressure broadening coefficients provided by the Exomol database ¹² are included. For O₂, N₂, CO, and CO₂, the HITRAN broadening formalism for self-broadening is used. Further included are collision-induced absorption (CIA) by O₂–O₂, O₂–CO₂, CO₂–CO₂, N₂–N₂, and N₂–CH₄ pairs (Richard et al. 2012) and Rayleigh scattering of H₂O, O₂, N₂, CO₂, and CO (Cox 2000; Sneep & Ubachs 2005; Wagner & Kretzschmar 2008; Thalman et al. 2014).

To model the radiative effects of the surface, we use the geometric albedo spectra from Hu et al. (2012). For the bare-rock scenario, HELIOS again iterates the surface temperature until the surface is in radiative equilibrium, i.e., the downward stellar radiation equals the reflected plus emitted radiation at the surface boundary. This takes the non-gray surface albedo into account across the range of 0.3–25 μm, thus correctly treating both the stellar flux absorption and reflection as well as the planetary emission.

When modeling the planetary envelope, we include the main infrared absorbers that may be plausibly found in secondary atmospheres—H₂O, CO₂, CO, CH₄, and SO₂ (see, e.g., Gillmann et al. 2022)—and vary their mixing ratios between 1 ppm and 1% (Gaillard & Scaillet 2014). As background gas, we use O₂ or N₂ (Wordsworth & Pierrehumbert 2014; Luger & Barnes 2015; Schaefer et al. 2016; Lammer et al. 2019). As limiting cases, we also approximate post-water-runaway, Venus-like, and carbon-rich (elemental C/O ∼1) atmospheres by adding pure H₂O, CO₂, and CO scenarios (Madhusudhan 2012; Kane et al. 2014; Goldblatt 2015). Specifically, the full range of our models included either CO or CH₄ in an N₂-dominated atmosphere; CO₂, SO₂, or H₂O in an O₂-dominated atmosphere; and atmospheres of pure CO₂, SO₂, or H₂O. These models are not intended to be exhaustive (nor may they all be chemically stable on geological timescales); our goal is to explore a representative range of atmospheres without overinterpreting our single-channel measurement.

For each model emission spectrum, we calculated the eclipse depth that would be measured in the 4.5 μm IRAC2 bandpass. Figures 4 and 5 compare our measured 4.5 μm eclipse depths to the predictions of our atmospheric models dominated by N₂ and O₂, respectively. For the N₂-dominated models, Figure 4 shows that the only models consistent with our eclipse measurement at 2σ or better have P_surf ≲ 1 bar (for CO as the active IR absorber) and ≲10 bar (for CH₄). Similarly, Figure 5 shows that the O₂-dominated models consistent with our measurement at 2σ have P_surf ≲ 1 bar (for CO₂) and ≲10 bar (for SO₂ or H₂O).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We also find that models lacking any atmosphere are also consistent with our eclipse measurement. Figure 6 shows that, while only the gray and metal-rich bare-rock models are consistent with our measurement at <1σ, models with an oxidized Fe, basaltic, and ultramafic surface composition are all consistent with our eclipse at <2σ. Models with a feldspathic or granitoid surface composition are inconsistent with the data at >2σ. However, we note that the substellar region of GJ 1252b is likely hot enough for such materials to melt. Although our atmosphere-free models may not be strictly accurate for this planet's surface, we leave more detailed models involving both solid and melted regions for future study.

Zoom In Zoom Out Reset image size

We first estimate the atmospheric-loss rate from GJ 1252b using the formalism of energy-limited atmospheric escape (Salz et al. 2016), leaving more involved estimates of the planet's mass-loss rate for future work. Using the MUSCLES treasury survey's spectra of nearby M dwarfs (France et al. 2016; Loyd et al. 2016; Youngblood et al. 2016), we estimate an XUV flux incident on GJ 1252b of (6–8) × 10³ erg s⁻¹ cm⁻². Assuming a heating efficiency of 0.3 (Salz et al. 2015) and that the planet's optical transit radius is the same as its effective radius of XUV absorption, this XUV flux translates into an atmospheric mass-loss rate of roughly 0.1M_⊕ Gyr⁻¹. The mass of a planet's atmosphere is just $4\pi {R}_{p}^{2}{P}_{\mathrm{surf}}/{g}_{\mathrm{surf}}$, which for GJ 1252b is $\left(5.3\times {10}^{18}\,\mathrm{kg}\right)\left({P}_{\mathrm{surf}}/1\,\mathrm{bar}\right)$. Thus, even a 100 bar atmosphere would be ablated in <1 Myr. Note that, although GJ 1252 exhibited no detectable stellar flares during its first sector of TESS observations (Howard 2022), the star was presumably more active, and thus mass-loss rates from GJ 1252b would have been even higher, earlier in the system's lifetime.

To further evaluate the prospects for volatile loss from GJ 1252b, we modify the model of Foley & Smye (2018), Foley (2019) to apply to GJ 1252b and explore varying initial mantle CO₂ inventories that will still allow for a completely desiccated planet at the estimated planetary age of ∼3.9 Gyr. Mantle gravity and core-radius-fraction model inputs were calculated using the mass–radius-composition solver, ExoPlex (Unterborn & Panero 2019). As the planet lacks any significant volatile atmosphere, we assumed the planet was made entirely of a FeO-free silicate mantle and pure-liquid-Fe core. Using a planet radius of 1.213 R_⊕ and a mass of 1.32 M_⊕, and assuming an Earth-like core radius fraction of 0.55%, ExoPlex calculates an average mantle density of 5026 kg m⁻³ and gravity 14.4 m s⁻². Due to the high surface temperature of the planet and the low likelihood of liquid water on the surface, we assume that the planet is in the stagnant lid regime of tectonics (as is, e.g., Venus; Gillmann et al. 2022). Our model then assumes all CO₂ outgassed from the mantle will accumulate in the atmosphere, as there is no known method of weathering or recycling carbon without water (Walker et al. 1981; Kasting & Catling 2003; Foley & Driscoll 2016). Our model also assumes an Earth-like heat-producing element budget, initial mantle temperature of 2000 K, and reference viscosity μ_ref = 1.3 × 10²⁰ Pa s (Foley & Smye 2018).

Assuming an atmosphere mass-loss rate of 5 × 10⁶ kg s⁻¹, we vary the initial mantle CO₂ inventory of our model between an Earth-like initial inventory of 0.011 wt% (mass percentage), based on an estimate of 10²² mol of CO₂ in the mantle and surface reservoirs of Earth (Sleep & Zahnle 2001), to 7.3 wt% (over two orders of magnitude more CO₂ than on present-day Earth, by mass fraction).

Figure 7 shows the time evolution of total atmospheric pressure with varying initial mantle CO₂ inventories. For the larger carbon inventories considered, the planet's initial rapid outgassing is greater than the atmospheric loss rate, allowing the atmospheric pressure to build before gradually being eroded once the mantle's carbon store is depleted. However, for the Earth-like initial CO₂ (1.1 × 10⁻² wt%), the atmospheric loss rate is always greater than the outgassing rate and so the planet never builds a significant secondary atmosphere. All models tested result in a planet with a completely eroded atmosphere by 3.9 Gyr. Thus, GJ 1252b may have had orders of magnitude greater carbon inventory than Earth yet still have no remaining atmosphere today.

Zoom In Zoom Out Reset image size

Taken together, our model spectra and escape calculations strongly indicate that GJ 1252b has no significant atmosphere.

GJ 1252b joins the handful of small planets (<2R_⊕) with infrared flux detections. The prior examples are 55 Cnc e (whose size and mass imply a sizable volatile mass fraction), LHS 3844b, and K2-141b; ¹³ some relevant parameters for all these systems (including GJ 1252b) are listed in Table 3. GJ 1252b is smaller than all of these other planets, but intermediate in irradiation. We used the system parameters in Table 3, along with BT-Settl model stellar spectra (Allard 2014) interpolated to these stars' parameters, to homogeneously calculate the irradiation and 4.5 μm dayside brightness temperatures of all these planets. The model spectra used for all four planets are included as machine-readable files with this paper.

Table 3. Planets < 2R_⊕ and with Measured Infrared Emission

Parameter	GJ 1252b a	LHS 3844b b	K2-141 b c	55 Cnc e d
Rp /R⊕	1.180	1.32	1.53	1.88
a/R*	5.03	7.08	2.36	3.51
Teff/K	3458	3036	4599	5172
${T}_{\mathrm{eq},\max }$ /K	1390	1030	2705	2550
δ4.5 μm/ppm	${149}_{-32}^{+25}$	380 ± 40	143 ± 39	83 ± 23 e
TB,4.5 μm/K	${1410}_{-125}^{+91}$	${1011}_{-33}^{+32}$	${2200}_{-330}^{+310}$	${2040}_{-290}^{+270}$
ESM7.5 μm	16.7	30.0	15.4	70.1 f

Notes.

^a From this work and Shporer et al. (2020). ^b From Vanderspek et al. (2019) and Kreidberg et al. (2019). ^c From Malavolta et al. (2018) and Zieba et al. (2022). ^d From Demory et al. (2016a) and Bourrier et al. (2018). ^e We adopt the standard deviation on the mean of the eclipse depths of Demory et al. (2016a) as a conservative estimate of this uncertainty. The weighted mean and its uncertainty from their first and second seasons of observations are 51 ± 17 ppm and 171 ± 27 ppm, respectively, corresponding to 1620 ± 230 K and 3070 ± 300. ^f For stars as bright as 55 Cnc, ESM-like metrics typically overestimate the achievable S/N (Kempton et al. 2018).

Download table as:  ASCIITypeset image

As shown in Figure 8, a curious dichotomy emerges. The two smallest and coolest planets (GJ 1252b and LHS 3844b) both have dayside brightness temperatures $\left({T}_{B,4.5\,\mu {\rm{m}}}\right)$ consistent with the maximum possible dayside equilibrium temperature (i.e., $f=\tfrac{2}{3}$ and A_B = 0 in Equation (3)):

$$\begin{eqnarray}&&{T}_{\mathrm{eq},\max }={\left(\displaystyle \frac{2}{3}\right)}^{1/4}{T}_{\mathrm{eff}}\sqrt{\displaystyle \frac{{R}_{* }}{a}}.\end{eqnarray} \tag{ 4 }$$

On the other hand, the two largest and hottest planets (55 Cnc e and K2-141b) both have notably lower normalized dayside temperatures that are consistent with uniform heat redistribution ($f=\tfrac{1}{4}$).

Zoom In Zoom Out Reset image size

This dichotomy may be coincidental: 55 Cnc e's emission measurements are best interpreted as indicating a massive atmosphere, while the interpretation for K2-141b is a nonzero albedo but negligible atmosphere (Zieba et al. 2022). Some calculations predict that rocky planets with sufficiently intense irradiation could exhibit a high substellar albedo induced by photovolatilization of dayside rocky materials (Kite et al. 2016; Mansfield et al. 2019). One possibility is therefore that there is an irradiation threshold for this albedo enhancement, with GJ 1252b and LHS 3844b lying below it and K2-141b lying above it.

Alternatively, both 55 Cnc e and K2-141b may have atmospheres thick enough that they can transport sufficient heat to measurably cool their daysides. Evidence of 55 Cnc e's thick atmosphere is seen in its asymmetric 4.5 μm phase curve (Demory et al. 2016b), but K2-141b's 4.5 μm phase curve showed no such evidence for a thick atmosphere (Zieba et al. 2022)—this despite K2-141b's dayside being heated high enough above the silicate solidus that an optically thick, ∼0.1 bar mineral atmosphere is predicted. Future modeling and observations will both be needed in order to determine whether Figure 8 represents a coherent trend between some combination of irradiation, planet size, heat redistribution, and albedo.

Fortunately, GJ 1252 is bright enough to offer high S/N while faint enough to be observable with all of JWST's instruments. With an Emission Spectroscopy Metric (ESM; Kempton et al. 2018) greater than K2-141b and within a factor of two of LHS 3844b (see Table 3), GJ 1252b is likely to join the select group of eminently observable and highly irradiated terrestrial exoplanets and to be subjected to many productive future investigations.

We have presented our measurement of 4.5 μm thermal emission from the highly irradiated terrestrial planet GJ 1252b. With a radius of just 1.180 R_⊕, our target is the smallest planet for which such a measurement has been reported. After presenting our Spitzer data analysis, along with an updated transit analysis using new TESS data in Sections 2 and 3, we compared these measurements to a large suite of atmospheric models and simulated spectra. Our modeling demonstrated that, for a broad range of possible atmospheric compositions, surface pressures P_surf ≲ 10 bar are required in order to be consistent with the measured eclipse at 2σ confidence. Furthermore, Section 4 then showed that the energy-limited atmospheric mass loss from GJ 1252b could quickly erode atmospheres with P_surf > 100 bar on timescales far shorter than the system lifetime.

We therefore conclude that GJ 1252b possesses no significant atmosphere. In this case, it presumably retains only a tenuous mineral exosphere; such an atmosphere would be expected to have P_surf ≲ 10⁻⁶ bar and to be dominated by species such as Na, molecular O₂ and atomic O, and K (Miguel et al. 2011; Ito et al. 2015). Such an atmosphere is likely thin enough that atmospheric circulation would contribute negligibly to global heat and mass transfer (e.g., Nguyen et al. 2020), although detailed simulations will be needed to confirm this. Since GJ 1252b's atmosphere would be quite optically thin, observations of GJ 1252b therefore offer the opportunity to directly probe surface minerology via emission spectroscopy during eclipse and throughout the planet's orbit.

Infrared emission has been measured from only three exoplanets with sizes placing them firmly in the terrestrial planet regime. A larger sample is urgently needed in order to better identify the surface and atmospheric properties of this class of planets; fortunately, this number is already set to increase somewhat in the dawning JWST era (see discussion by Zieba et al. 2022). A further pressing need is to obtain more precise mass measurements for these planets: no mass has been reported for LHS 3844b, and GJ 1252b has only a roughly 3σ mass. The combination of more precise masses and bulk densities, more precise eclipse spectra, and stellar abundance measurements will ultimately enable more accurate models to link these planets' surface minerologies and atmospheres to their observed thermal emission.