The Hubble PanCET Program: Emission Spectrum of Hot Jupiter HAT-P-41b

Emission spectroscopy of exoplanets allows us to probe into the deeper layers (1–10 bar) of the dayside atmosphere compared to transmission spectroscopy which measures the upper layers (∼1 mbar) of the planetary limbs. The difference in the radiative transfer path geometry means that the emission spectrum is much more sensitive to the vertical thermal structure of the planet. Depending on the atmospheric chemical composition and the temperature–pressure (TP) profile, molecular absorption, or emission features can be imprinted onto the emission spectrum. Notably, water features at 1.4 μm (Fu et al. 2017) have been the most robustly detected in both absorption (Kreidberg et al. 2014) and emission (Evans et al. 2017; Fu et al. 2022). The presence of H- continuum opacity was inferred in multiple ultra-hot Jupiters (Arcangeli et al. 2019; Fu et al. 2021a). In the longer wavelength, the CO/CO₂ features were indicated by the deviations of Spitzer photometric points in the 3.6 and 4.5 μm channels from the blackbody approximation (Garhart et al. 2020; Fu et al. 2021b).

HAT-P-41b is an inflated hot Jupiter ($R={1.685}_{-0.051}^{+0.076}$R_Jup M = ${0.8}_{-0.1}^{+0.1}$ M_Jup T_eq = ${1940}_{-38}^{+38}$K; Hartman et al. 2012) with detailed atmospheric characterization via transmission spectroscopy (Lewis et al. 2020; Wakeford et al. 2020; Sheppard et al. 2021). The planet shows a high-metallicity (Wakeford et al. 2020; Sheppard et al. 2021) atmosphere with an increased H- opacity abundance (Lewis et al. 2020). With a dayside temperature of ∼2300 K, HAT-P-41b sits between hot and ultra-hot Jupiters where physical processes such as molecular dissociation and H- opacity are starting to become important (Parmentier et al. 2018). HAT-P-41b allows us to pinpoint the transitional parameter space of various atmospheric processes which makes it a valuable data point in our understanding of hot-Jupiter atmospheres on a population level (Baxter et al. 2021; Mansfield et al. 2021). Here we present the 1.1 to 4.5 μm emission spectrum of HAT-P-41b with combined data from Hubble Space Telescope (HST)/Wide Field Camera 3 (WFC3) G141 and Spitzer channel 1 and 2.

The HST WFC3/G141 data set was observed as a part of the Panchromatic Exoplanet Treasury Survey (PanCET GO 14767; PIs: Sing & Lopez-Morales) on 2016 October 9 and the Spitzer data set was published in Garhart et al. (2020). The WFC3/G141 eclipse observation was taken in spatial scan mode for five consecutive HST orbits. Each frame was taken with the 256 × 256 pixel subarray in the SPARS10 and NSAMP = 12 settings. The forward scanning rate is 0065 s⁻¹ and the exposure time is 81 s.

All the orbital parameters for the WFC3/G141 data reduction have been fixed to the same values used in Sheppard et al. (2021) which are identical to what were used in the (Garhart et al. 2020) Spitzer analysis.

The data reduction process starts with applying the standard flat-field, background subtraction, and bad pixels, as well as the cosmic ray removal on the ima frames (Fu et al. 2021a). Then we extract the nondestructive reads from each frame (Deming et al. 2013). There is a spatially resolved companion star (Sheppard et al. 2021) located 36 away. Due to the large spatial separation, the nondestructive reads of the spatial scan from the two stars do not overlap which allows for clean removal of the companion star spectra. The companion star removed reads are then combined to the complete spatial scan frames. Next we summed each frame in the vertical direction for the 1D spectrum and normalized it by its own median flux to calculate the wavelength shifts. We then used the scipy. interpolate. interp1d function to interpolate the 1D spectrum in the wavelength direction and calculated the relative subpixel level shifts of each frame. The largest shift between any two frames is under 0.1 pixel which does not induce any excessive systematics (Stevenson & Fowler 2019). Wavelength shifts corrected non-normalized frames were then summed in all wavelength channels to form the white light eclipse lightcurve. The lightcurve is then fitted with a combination of BATMAN (Kreidberg 2015) with the RECTE charge trapping systematics model (Zhou et al. 2017), second order polynomial of the HST orbital phase, and the wavelength shifts. Each wavelength channel is then fitted using the same method but with the midtransit time fixed to the best-fit white light value. The best-fit lightcurves of each wavelength channel and corresponding residuals are shown in Figure 1. We also compared our spectrum with that in Mansfield et al. (2021) and they are in good agreement (Figure 2).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We performed the retrieval analysis on the reduced HAT-P-41b emission spectrum with ATMO (Amundsen et al. 2014; Tremblin et al. 2015; Drummond et al. 2016; Tremblin et al. 2016; Wakeford et al. 2017; Goyal et al. 2018). The retrieval setup is equilibrium chemistry with a fixed solar C/O ratio. There is a total of four free fitting parameters (Table 1) including metallicity (Z) and three parameters κ_IR, γ_O/IR, and β for the TP profile as defined in Line et al. (2013). We fixed the C/O ratio to the solar value due to the lack of constraint this data set has on the carbon-bearing species. Also, the transit spectra retrieval (Lewis et al. 2020; Sheppard et al. 2021) results were all consistent with the solar C/O values.

Table 1. ATMO Eclipse Retrieval Priors and Posteriors

Parameter	Priors	Posteriors
log(Z/Z⊙)	${ \mathcal U }$(−2.8, 2.8)	$-{0.342}_{-1.277}^{+1.677}$
log(KIR)	${ \mathcal U }$(−5, −0.5)	$-{2.887}_{-0.907}^{+0.982}$
log(γ/IR)	${ \mathcal U }$(−4, 1.5)	$-{0.907}_{-1.716}^{+0.672}$
beta	${ \mathcal U }$(0, 2)	${1.326}_{-0.168}^{+0.068}$

Download table as:  ASCIITypeset image

The best-fit model to the emission spectrum (Table 2) from the ATMO retrieval (Figure 3) gives a ${\chi }_{\nu }^{2}$ of 1.24 (4 degrees of freedom) with a near-isothermal TP profile. The best-fit blackbody temperature using the PHOENIX stellar model (Husser et al. 2013) grid (logg = 4.5 and logZ = 0) interpolated to T_eff = 6390 K is 2283 ± 64 K with a ${\chi }_{\nu }^{2}$ of 1.23 (1 degree of freedom). Based on the formalism described in Cowan & Agol (2011), this dayside temperature would suggest a circulation efficiency ε of 0.44 assuming zero Bond albedo and the predicted nightside temperature would be 1572 K. ε = 0 represents the noncirculation limit where the nightside temperature is 0 K and ε = 1 represents the full-circulation scenario where the nightside temperature is the same as the dayside. The measured dayside temperature is consistent with the dayside-only heat redistribution scenario. The retrieved metallicity is consistent with the solar value (Figure. 4). While Sheppard et al. (2021) retrieved metallicity ∼1σ higher than solar, Lewis et al. (2020) reported values consistent with solar metallicity. Considering the large uncertainties on metallicity from all three studies, the current data sets cannot well constrain the models.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Lewis et al. (2020) found H- abundance several orders of magnitude larger than equilibrium chemistry is needed to best fit the transmission spectrum. We did not find evidence for significant H⁻ opacity from the emission spectrum and we believe this could be due to the following: (1) The blackbody-like emission spectrum and near-isothermal TP profile make the abundances largely unconstrained. (2) The H- constrain in Lewis et al. (2020) comes from a combination of WFC3 G280 and G141 data. Wakeford et al. (2020) did not find evidence for H- solely based on the G280 optical data set. Therefore any uncorrected offsets between G280 and G141 spectra can potentially lead to high H- abundance. (3) Sheppard et al. (2021) did not retrieve significant H- abundance with both G280 and G141 data sets using a different retrieval code. So it could also be due to different forward model assumptions.

Heavy metals such as Fe i, Fe ii, and molecules like TiO are considered major optical absorbers that can induce temperature inversion in hot-Jupiter atmospheres (Fortney et al. 2008; Lothringer et al. 2020), and they have been detected in the optical transit spectrum with HST/STIS G430L and G750L (Fu et al. 2021a). There are no significant near-UV heavy metals nor optical TiO absorption detected in both the Space Telescope Imaging Spectrograph (STIS; Sheppard et al. 2021) and Ultraviolet Imaging Spectrograph (UVIS; Lewis et al. 2020; Wakeford et al. 2020) transmission spectra for HAT-P-41b. The noninverted TP profile retrieved in this HAT-P-41b emission spectrum is consistent with the nondetection of near-UV/optical absorbers in the transit spectra.

To quantify the WFC3/G141 emission spectra from various hot Jupiters and compare HAT-P-41b to them, we took the 1.4 μm water feature strength index (${{\rm{S}}}_{{{\rm{H}}}_{2}{\rm{O}}}$) reported in Mansfield et al. (2021) and plotted them in the dayside temperature versus gravity parameter space (Figure 5) first introduced in Parmentier et al. (2018). The important physical transitions including TiO, H⁻, and molecular dissociation are also the same as shown in Parmentier et al. (2018). The larger negative ${{\rm{S}}}_{{{\rm{H}}}_{2}{\rm{O}}}$ values (redder) represent stronger water emission features while the larger positive values (bluer) indicate more prominent water absorption features. The index is centered at zero for a blackbody-like featureless spectrum. In addition, we also scale the size of each point based on the host star temperature (T_eff) to reflect the effect of increased far-UV/UV radiations from earlier type stars (Lothringer & Barman 2019; Fu et al. 2022).

Zoom In Zoom Out Reset image size

We have only seen strong water absorption features among cooler hot-Jupiter (T_day < 2200 K) atmospheres (Kreidberg et al. 2014; Line et al. 2016) driven by their decreasing TP profiles. This is due to heavy metal absorbers condensing out of the atmosphere at low temperatures. On the other hand, when the atmosphere becomes too hot (T_day > 2700 K), water molecules dissociate and H⁻ continuum opacity starts to dominate. As a result, we have only observed featureless emission spectra among the hottest planets. In between the two regions (∼2200 to 2700 K) we see a transitional parameter space where atmospheres can be inverted (Evans et al. 2017; Fu et al. 2022) or isothermal (Fu et al. 2021b; Mansfield et al. 2021). HAT-P-41b sits in the middle of this transitional space. Its mostly featureless WFC3/G141 emission spectrum is similar to that of WASP-76b, WASP-74b, WASP-19b, and KELT-7b. However, it is different from WASP-121b and KELT-20b where we saw evidence for water emissions. Although previous models have indicated multiple important physical transitions (Parmentier et al. 2018) happening at this temperature and gravity range, it is not yet well understood what exact physical processes drive the different thermal structures and emission spectra of these planets. In addition, host star type could be another important determining factor since strong far-UV/UV radiations from the host star can strengthen atmospheric thermal inversion (Lothringer & Barman 2019; Fu et al. 2022) as they deposit significant energy into the upper layers of the planetary atmospheres. This is supported by the large emission features seen in KELT-20b which orbits an A-type star. The lack of far-UV/UV flux from F-type host star of HAT-P-41b could be another cause for its isothermal atmosphere.

We present the most complete emission spectrum for inflated hot Jupiter HAT-P-41b. The spectrum is close to blackbody-like with no significant molecular absorption or emission features. The best-fit ATMO model shows an isothermal TP profile agreeing with the dayside heat redistribution scenario and a metallicity value consistent with the solar value. The noninverted TP profile is consistent with the nondetection of near-UV/optical absorbers in the transit spectra. Significant H⁻ opacity suggested in Lewis et al. (2020) is not required in the model to adequately fit the emission spectrum. We also do not retrieve a metal-rich atmosphere as indicated in Sheppard et al. (2021). However, the emission spectrum does not well constrain the atmosphere metallicity due to the limited wavelength coverage. The featureless emission spectrum of HAT-P-41b indicates planets with dayside temperatures around 2300 K may have relatively isothermal TP profiles in the absence of heavy metal absorbers and strong host star far-UV/UV radiations to drive thermal inversion. The comparison of HAT-P-41b to other similar hot Jupiters paints a murky picture of how atmospheric physical properties transition from cooler to hotter planets. The combined effect of surface gravity, thermal dissociation, H⁻ opacity, heavy metal absorbers, and host star type is yet to be disentangled. Future observations of more similar planets and follow-up JWST infrared measurements of HAT-P-41b will be the key to solving the mystery of hot-Jupiter atmospheres.